Report No. 39809-KH Cambodia Sharing Growth: Equity and Development in Cambodia Equity Report 2007 June 4, 2007 Poverty Reduction and Economic Management Sector Unit East Asia and the Pacific Region Document of the World Bank ABBREVIATIONS AND ACRONYMS ADB Asian DevelopmentBank MOEYS Ministry of Education, Youth and Sports ANC Antenatal care MOH Ministry of Health ARI Acute respiratory infection MOPS Movingout of Povetry Study BCG Bacillusof Calmette and Guerin: vaccination against TB NCD Non-communicablediseases CISC CommuneI SangkatCouncil NGO Non-GovernmentalOrganization CARERE Cambodia Area Rehabilitationand RegenerationProgram NIPH NationalInstituteof Public Health CAS Center for Advanced Study NIS NationalInstituteof Statistics CDC (i)CommuneDevelopment Committee; (ii) Councilfor the National Strategic Development Plan Developmentof Cambodia NSDP 2006-2010 CDHS Cambodia Demographicand HealthSurvey OCA OverlappingClaims Area CDRI Cambodia DevelopmentResource Institute ODA Official Development Assistance CIPS Cambodia Inter-CensalPopulation Organisation for Economic Cooperation Survey OECD and Development CNPA CambodiaNationalPetroleum Agency OOPS Out-of-pocketspending CSES Cambodia Socio-Economic Survey ( 1 997,2004) PETS Public Expenditure Tracking Survey D&D Decentralizationand deconcentration PFM Public financial management DIW District IntegrationWorkshop PPA ParticipatoryPovetry Assessment DK Democratic Kampuchea PRK People's Republicof Kampuchea EF Equity Fund RGC Royal Government o f Cambodia FDI Foreign Direct Investment RH ReferralHospital FGD Focus group discussion Socio-Economic Survey of Cambodia SESC (1993/4) FTC Free-standingTechnical Cooperation SLC Social LandConcession GDP Gross Domestic Product TB Tuberculosis GMAC Garment Manufacturers' Association of Cambodia TOFE Table of FinancialExpenditures GTZ Gesellschafifur Technische Zussamenarbeit (German Agency for Technical Cooperation) USMR Under-fivemortality rate HC Health Center UNDP UnitedNationsDevelopment Program HFA Height-for-age VAT Value-added tax ILO InternationalLabour Organisation WDR World Development Report IMR Infant mortalityrate WFA Weight-for-age IRL Indochina Research Limited WFH Weight-for-height Law on the Managementand Administration of Communes LMAC (CHECK) WTO World Trade Organisation MEF Ministry of Economy and Finance MMR Maternalmortality rate ACKNOWLEDGMENTS This report has benefited from valuable support and debate at various stages from H.E. Keat Chhon (Senior Minister, Ministry of Economy and Finance), Dr.Aun PornMoniroth (Chairman of SNEC and Secretary of State of the Ministry of Economy and Finance), Dr. Hang Chuon Naron (Secretary General of the Ministry of Economy and Finance), H.E. Vongsey Vissoth (Deputy Secretary General of the Ministry of Economy and Finance) and other members of the Supreme National Economic Council (SNEC). An early note on the theme of equity and development in Cambodia, presentedat a one-day workshop in June 2006, benefited greatly from comments from Helen Appleton (DFID), Hong Sokheang (UNDP), Gijs Koop (NFO Forum), Mao Moni Ratana (ADB), Jenny Pearson (VBNK), Horn Vuthy (CDRI) and Kim Sedera (CDRI). The authors very much appreciate the support and ideas received from members of the World Bank team that led the preparation of the 2006 World Development Report on equity and development (Jose Edgardo L. Campos, Francisco Ferreira, Alan Gelb, Giovanna Prennushi and Caroline Sage); and duringthe visit of the Chief Economistand Senior Vice- President, Franqois Bourguignon, in which he participated in a discussion on equity hosted by SNEC. The authors of the report wish to thank Homi Kharas and Mun S Ho for generously sharing their time and knowledge to provide invaluable advice and suggestions. The authors also appreciate the interaction with Sarthi Acharya and others working on the UNDP Cambodia national human development report. The Concept Note was reviewed within the World Bank by Daniel Adler, Nisha Agrawal, Beng Simeth, Luis Benveniste, Bun Veasna, Chea Huot, Heang Path, Mia Hyun, Zhi Liu, Kazi Mahbub-Al- Matin, NilVanna, Ian Porter, Denis Robitaille, Rob Taliercio, Charis Wuerffel and Albert Zeufack (all World Bank). Others who have providedhelpful feedback include Brett Ballard(CDRI), Chan Sophal (CDRI), Brian Lund (Oxfam America), Carol Mortensen (Save the Children Australia), Sok Hach (EconomicInstituteof Cambodia) and Eiichiro Hayashiand colleagues at JICA. The report was preparedunder the overall coordinationand guidance of IndermitGill (Sector Manager, EASPR) and Nisha Agrawal (Country Manager) with support from Ian Porter (Country Director) and Kazi Mahbub-Al-Matin (Lead Economist, Cambodia). The core team was comprised of Chorching .... Goh, Tim Conway, Luis Benveniste, Xubei Luo, Mia Hyun, and Neak Samsen. Lead responsibilities were as follows: The Executive Summary was written by ChorchingGoh and Tim Conway; Chapter 1 (introduction): ChorchingGoh and Tim Conway with inputs from Huot Chea; Chapter 2 (rural inequality): ChorchingGoh and Tim Conway; Chapter 3 (urban inequality): ChorchingGoh; Chapter 4 (land): Chorching Goh and Tim Conway, drawing heavily on inputs from Daniel Adler (World Bank Social Development), DougPorter(University of Queensland) andMichael Woolcock, . (World Bank DEC); Thomas Markussen (consultant); Steven Schonberger (World Bank ARD); and Dominiquevan de Walle (WorldBank DEC). Chapter 5 (infrastructure):ChorchingGoh with inputsfrom Yusuf Ahmad (consultant). . Chapter 6 (servicedelivery and human development): Tim Conway with inputs from Toomas Palu and Luis Benveniste (World Bank Human Development), Chris Sakellariou (Consultant), Betsy .. Vanleit (consultant), Jan Willem Rosenboom(World Bank Water and SanitationProgram), Jeremy Ockelford(OPM consultants) and Andy Robinson (Consultant). Chapter 7 (gender): Mia Hyun with inputs from Cheryl Urashima (consultant), Ingrid FitzGerald (consultant)and Dr. Sok Sokun and other staff of UNFPA. Chapter 8 (institutions): Tim Conway and Daniel Adler (World Bank Social Development)with inputs from Doug Porter (University of Queensland) and Michael Woolcock (World Bank DEC); and RobTaliercio(World Bank PREM). Chea Huot (PREM) prepared the data annexes; James Knowles and Nong Zhu (consultants) ably provided statistical analysis. Bou Saroeun (Communication Specialist) has coordinated public communication of this program of work. Assistance on the preparation of the report has been providedby Chea Bunseang, Iv Ek Nimnuon, Lynn Gross, MichaelFigueroa, and Doris Chung. The key sources of data for this report were the Cambodia Socio-Economic Survey (CSES) 2004, conducted by the National Institute of Statistics(NIS); the Cambodia Demographic and Health Survey (CDHS) of 2005, conducted by the National Instituteof Public Health (NIPH) and National Institute of Statistics (NIS); the Moving Out of Poverty Study (MOPS) and Tonle Sap Participatory Poverty Assessment (PPA), both conducted by the Cambodian DevelopmentResource Institute(CDRI); and a national opinion poll designed and conducted by Indochina Research Limited (IRL). Valuable background papers were commissioned from Yusuf Ahmad (consultant), David Chandler (Monash University, Australia), Thomas Markussen (University of Copenhagen), Meas Nee (Krom Akphiwat Phum), James Robinson (Harvard University, USA), Steven Schonberger (EASRE, the World Bank), Kheang Un (Northern Illinois University, USA), and Dominique van de Walle (DECRG, the World Bank). Lastly, the authors and team want to thank the report's peer reviewers, GiovannaPrennushi(PRMPR) and Ed Campos (SASPR), for their advice and suggestions. TABLE OF CONTENTS EXECUTIVE SUMMARY Overview .............................................................................................................. i Background: transition and growth inCambodia ................................................ ...11.... Creating equality of opportunity ........................................................................ Sharing the benefits of growth: trends inper capita consumption...................... 111 Summary: findings and policy implications ..................................................... v11 xiv CHAPTER 1: Recent economicand social developments Summary .............................................................................................................. 1 Why examine inequality?..................................................................................... How has the Cambodian economy evolved? ....................................................... 21 Trends ininequality of monetary dimensions of welfare..................................... Sources andmeasures .......................................................................................... 6 9 13 Perspective and focus ......................................................................................... Trends ininequality of non-monetary dimensions of welfare ........................... 17 CHAPTER 2: Ruralinequality What took place?................................................................................................ Summary............................................................................................................ 18 19 Which groups drove inequality? ........................................................................ 22 Drivers of inequality inthe rural population...................................................... 28 CHAPTER 3: Urbaninequality What happenedto inequality?............................................................................ Summary............................................................................................................ 40 41 The components of urbaninequality .................................................................. How did the transformation of the urban economy influence inequality?.........46 Why did urban inequality remainhigher than rural inequality? ........................ 44 49 CHAPTER 4: Addressingasset inequality:a focus on land Summary ............................................................................................................ 53 Why is land so important?.................................................................................. Inrespect to land distribution, equity andefficiency are fully complementary 56 Landlessnessandpoverty................................................................................... .54 58 Land titles and agricultural development........................................................... 63 CHAPTER 5: Addressinginequalityin access to infrastructure: a focus on rural roads 75 Distributions o f infrastructure ............................................................................ Summary ............................................................................................................ 76 Disparity o f outcomes from the disparity o f access to roads ............................. Substantial progress inclosing the rich-poor gap .............................................. 80 83 CHAPTER 6: Promotingequity through service delivery Human development. equity and growth ........................................................... Summary ............................................................................................................ 87 88 Health and wealth............................................................................................... 88 Educational equity and inter-generational mobility......................................... Policy recommendations .................................................................................. 120 136 CHAPTER 7: Progressand challenges in gender equity Summary .......................................................................................................... Economic growth has created factory jobs for women................................... 139 .......................................... 140 Menand women face differentform ofinsecurity........................................... Expanded service delivery has closed gender gaps 142 144 Participation in decision-making...................................................................... . . * . . . 145 CHAPTER 8: Institutionsand equity inthe development process 149 Institutions and development ............................................................................ Summary .......................................................................................................... Promoting equity though public spending ....................................................... 150 154 169 Equitable institutions require open-ended process. ................................ Legal andjudicial institutions and access to justice for the poor..................... 171 The roles of civil society and external development partners.......................... 179 REFERENCES........................................................................................................... 181 ANNEXES 2. Empirical framework and regression results for analysis o f urban inequality... 1. Inequality and growth: Cambodia incontext o f ASEAN and rest o f world 188 3. How can titling be translated into poverty reduction? .............................. ........................................................................................................... 189 196 4.Technical tables......................................................................................... 197 List of Boxes Box 1: Poor householdslack the resourcesto cope with healthshocks .......................... x Box 2: Roads reduce spatial inequalities and expand opportunities for the poor ............................................................................................ xi Box 3: RuralCambodiansperceive andworry about growing inequality ..................................................................................................... .. xi1 Box 1.1: Poverty trends inCambodia, 1993/4to 2004....................................................... 3 Box 1.2: World Development Report 2006: Equity and development............................... 4 Box 1.3: Defining terms and concepts: equality and equity ............................................... 5 Box 1.4: Qualitative researchshows widespread concern with growing 7 Box 1.5: Commonly usedinequality measures................................................................... inequalities ........................................................................................................... 8 Box 1-6: Stratification or churning? Potential insights from longitudinal 10 Box 2.1: Initial conditions and the re-emergenceof difference........................................ analysis............................................................................................................... 28 Box 2.2: Peace has opened up opportunities for investment, trade and Box 2.3: Poverty and debt traps........................................................................................ accumulation ...................................................................................................... 30 34 Box 2.4: Individuals with a "strong back" (connections or corrupt influence) enjoy greater-often illegal-economic 39 Box 4.1: Patterns of landholding: findings from a landsurvey ........................................ opportunities ...................................... 55 Box 4.2: Competition and confusion inland managementreflects Box 4.3: Lessons learned and applied inCambodia's land distribution program ............61 Cambodia's turbulent history ............................................................................. 56 Box 4.4: Inmuch of the Cambodian countryside, land markets appear to work reasonably-for Box 4.5: Interpreting household-level irrigationdata....................................................... now-despite weak or non-existent titles ..................64 66 Box 4.6: Systematic titling has positive effects, which can be expectedto increase with time and with the extension of the program into more Box 4.7: Among recipients, titles are perceived as improvingtenure security ................69 remote and less densely-populated areas ........................................................... 71 Box 6.1: Rural healthcarehas improved inmany respects............................................... 91 Box 6.2: Technical note on selected healthindicators...................................................... 99 Box 6.4: Poverty and illness inthe CSES....................................................................... Box 6.3: Sanitation and hygiene may be as important as drinking water quality .............91 100 Box 6.5: Rural Cambodiansperceive, and have explanations for, worse Box 6.6: Private providers have helped to expand access to modern healthcare............105 healthoutcomes amongst poor households...................................................... 100 Box 6.7: The poor often delay using public services until problems are serious ...........109 Box 6.8: Despite formalization of user fees and (limited) introduction of fee exemptions, cost remains a barrier to care ............................................. 110 Box 6.9: Funding public healthand protecting the poor: user fees and 112 Box 6.10 Healthcare-induced debt and land10s remain serious problems ..................... equity funds ...................................................................................................... Box 6.11: Poor households lack the resources to cope with health shocks .................... 113 113 Box 6.12: Around a third of all disabilities originate inpoor health ..............,.............. 118 . Box 6.13: Preference or constrainedchoice: do the poor value education?................... 120 Box 6.14: Since 1993, the Government has beencommitted to equity in education ,.,.,,, ., ........................,, .,,.,..,......... ...,,.,...,....,.... ...,, ,,, .,,, , .,. .,,,....,......... 124 Box 6.15: For poor households, demands on children's time is akey reason Box 6.16: Inter-ethnic differences ineducation.............................................................. for over-age enrolment or dropout ,............................................I.............IIIIII..129 131 Box 6.17: Government-NGO partnershipshave reducedout-of-pocket costs for the poor and delivered improvements inhealth equity ,......,.......,,,............ 137 Box 7.1: Qualitative research suggests a decline in domestic violence in some 145 Box 7.2: Women tend to be responsiblefor managing borrowing................................. villages ............................................................................................................. Box 8.1: What do we meanby "institutions"? ...............,... ... ....................,.................... 146 , 150 Box 8.2: The quality of local governance i s a major, generally negative influence upon opportunities, which will probably require national solutions .....,......,,,,......,........,,.,.,,...,,... ......................,...,............ .................... 152 ,, Box 8.3: The quality of local governance is a major, generally negative influence upon opportunities, which will probably require national ...................... ,.... .,.. ..............I.............................,..153 Box 8.4: "Rights inpractice": power confers greater opportunities for the rich............154 solutions ....................... I Box 8.5: Core concepts inpublic expenditure policy..................................................... 155 Box 8.6: Non-revenue financing of public spending......... Box 8.7: Difficulties inidentifying optimal allocation for goals ...... ....,............ ......,............. 157 , .......,,....,, .....,................ 162 Box 8.8: Civil service salaries and development outcomes............................................ Box 8.9: Infrastructure and equality: cross-country and locality findings.......,,,............ 163 Box 8,lO:Current estimates o f Cambodian petroleum reserves ..................... ................ 165 Box 8.11: Justice systemsplay a crucial role inlong-term development ......,........... , Box 8.12 "Interim institutions" for governance reform Box 8.13 Monitoring and dispute resolution inthe garment industry ,..........,,.............. 173 ...,...,,, ..,............ ,.,................ 171 ..... 166 169 ,,,,,, , Box 8.14 Commune Development Committees and District Integration Workshops ...,,, ........ .............., ..............,......,... ...,.................,.......................175 , ,,, , Lists of Figures Figure 1: Since 1994, growth has beenvariable but high....,..... ......,.............................ii Figure 2: Living standards improved at very differentrates, widening inequalities ....... .,.,, .................,.,..,,.., ............ ..,,.,................... ..................111 ... ,, ,, , , Figure 3: The rise innational inequality has beendriven by growing inequality within the countryside ..,....................,..........................................111 ... Figure 4: Inruralareas, the richinitially maderapidgains while others lagged; later, improvements were spread evenly across the population......... .,, .,. , .,.,,.. ... , .,,. ....,......., ... ..... .... .... ..... .....,., .,.,.iv Figure 5: Investments inproductive asset matter ......................................................... ,, , , ,, , ,,,,,, , , , ,, , , , ,,, vi Figure 6: Inequalities inliteracy are closing over time ......................,...........,,..........VIH... Figure 7: Child care and survival by wealth ................................................................. ix Figure 8: Landownership is highly unequal ................................................................. x Figure 1.1: From 1994to 2004. per capita consumption amongst the poorest fifth has risen by only 8 percent-compared to 45 percent amongst the richest fifthofthe population.................................................................. 3 Figure 1.2: Richandpoor Cambodianscare about the same issues, albeit with different emphasis.................................................................................. 6 Figure 1.3: Changes in consumption, percentile-to-percentile, for Phnom Penh, 11 Figure 1.4: Over the last decade, rural inequality has risen sharply............................... Other Urban, and Rural Areas, between 1994-2004.................................... 12 Figure 1.5: Most people believe inequality has risen; and will continue to do Figure 1.6: Distributions of consumption, by sector, 1993, 1997and 2004...................14 so .................................................................................................................. 13 Figure 1.7: Increasing equality inprimary enrolment is closing gaps inliteracy rates amongst the younger generation.......................................................... 16 Figure 1.8: Faster gains from a betterbase inurban areas are widening existing inequalities...................................................................................... 17 Figure 2.1: Initially, the very richmade rapid gains while others lagged; later, improvements were spread evenly between wealth groups. ...............21 Figure 2.2: A decomposition of rural Theil Indices ....................................................... 24 Figure 2.3: Poverty rates in upland areas are significantly higher than on the coast.............................................................................................................. 25 Figure 2.4: Inthe MOPS dataset, most villages recorded an increase inthe income Gini averaging 0.05; two registered muchhigher increase............26 Figure 2.5: The relationships of provincial Gini Coefficients andthe population shares inthe richest quartile group (rural areas), 1993-2004 .................................................................................................... 27 Figure 2.6: Levels and Growth of Provincial (rural) Mean Real Consumption, 31 Figure 2.7: Inequality rose with greater education stocks.............................................. 1993-2004 .................................................................................................... 36 Figure 2.8: Inequality and investments (actual and potential), 2004.............................. 38 Figure 3.1: Changes inreal consumption, percentile-to-percentile, for 43 Figure 3.2: A decomposition of UrbanTheil Indices..................................................... periods (i)1993-2004, (ii)1993-1997, and (iii)1997-2004......................... 46 Figure 3.3: The relationship of inequality and education stock.,.................................... 48 Figure 3.4: The relationship of inequality and population shares inthe secondary sector (i.e,, manufacturing, construction) ................................... 49 Figure 4.1: Land distribution inCambodia ranks amongst the most unequal inAsia .......................................................................................................... 54 Figure 4.2: Small farms are a better economic proposition than large farms .................57 Figure 4.3: For most crops, there is no significant economy of scale effect ..................62 67 Figure 5.1: access to road and employment inthe secondary sector.............................. Figure 4.5: The benefits of title are clearly seen to outweigh any problems..................70 Figure 4.4: The relationship between investments and secure land titles....................... 86 Figure 6.1: Availability and use of healthcarehave both expanded 90 Figure 6.2: While some health outcomes have improved, others have not .................... significantly.................................................................................................. 91 Figure 6.3: Children now born are more likely to live to adulthood and middleage .................................................................................................... 93 Figure 6.4: Faster gains from a better base inurban areas are widening existing inequalities...................................................................................... 95 Figure 6.5: Child malnutritionamongst the poor i s roughly twice that of the 96 Figure 6.6: The poor are at far greater risk of water-related disease.............................. rich................................................................................................................ 97 Figure 6.7: Children from richfamilies are significantly more likely to be 98 Figure 6.8: The richpractice better householdhygiene than the poor ........................... covered by preventative healthservices....................................................... 99 Figure 6.9: From birth onwards, poor childrensuffer significantly worse Figure 6.10: Frequency and size of out-of-pocket-spending on health. 2004 ................101 healthoutcomes-including survival rates................................................ 103 Figure 6.11: The richest quintile spends far more on healthcare than do other groups: but this reflects a similar gap inoverall living standards..............104 Figure 6.12: There are substantial wealth-baseddifferences inuse of Figure 6.13: For the poor, cost and distance are the main barriers to healthcare...........108 hospital treatment....................................................................................... 105 Figure 6.14: There are pronounced spatial variations inbarriers to healthcare..............109 Figure 6.15: Access to care during pregnancy and delivery vary greatly 114 Figure 6.16: I11health reduces labor force participation................................................. betweenrural and urban populations ......................................................... 116 Figure 6.17: There is a strong associationbetweeneducational achievement andthe quality of employment and income ............................................... 121 Figure 6.18: For the poor, cost and distance are the main barriers to healthcare educational attainment ............................................................................... Average earning amongst the self-employed appear to reflect 123 Figure 6.19: Although shrinking over time, educational inequalities remain large............................................................................................................ 125 Figure 6.20: Increasing equality inprimary enrolment is closing gaps in literacy rates among the younger generation ............................................. 126 Figure 6.21: The richest parents outspendthe poorest parentsby a factor of 25 ........................................................................................................... 128 Figure 6.22: Richer parents can afford muchhigher education spending ...................... 129 Figure 6.23: Distance is not now a major constraint to primary education -but is likely to act as abarrier to secondary education for the poor ............................................................................................................ 130 Figure 6.24: Parents' education appears to be a string influence upon a child's educational achievements............................................................... 133 Figure 6.25: A mother's education influences a child's chances of obtaining healthcare, from conception onwards ....................................................... 134 Figure 6.26: A mother's education strongly shapes her children's health outcomes .................................................................................................... 135 Figure 7.1: Women are more likely than men to be engaged in services in the informal sector, and manufacturing inthe formal sector ........................... 140 Figure 7.2: Inwage labor, women are concentrated ingarments; mendominate construction, public sector employment, and other services...................... 141 Figure7.3: Men, particularly men from the richest quintile, dominate public sector employment ..................................................................................... 142 Figure7.4: Educational outcomes amongst recent school leavers show much Figure 7.5: Most household decisions involve both husband and wife ........................ less inequality............................................................................................. 143 146 Figure8.1: Cambodia ranks 49 out o f 51 low income countries for revenue collection .................................................................................................... 158 Figure 8.2: The economic classification ofpublic spending, 1996-2005 ..................... 164 List of Tables Table 1: Urban consumption inequality appears not to have risen-but i s still much higher than inthe countryside ................................................................. v Table 2: There is greater poverty inremote villages with less connectivity..................vi Table 3: There is limited trust inpublic institutions, particularly police and the courts .................................................................................................. .. vi1 Table 1,l: According to household surveys, inequality (Gini coefficients o f real per capita consumption) rose considerably inthe countryside, but remainedlargely unchanged inthe towns................................................. 11 Table 1.2: There has been a dramatic expansion inbasic education inrecent years ................................................................................................................ 15 Table 2.1: Welfare differences inthe countryside appear to have widened rapidly between 1993 and 1997, but more slowly after that ........................... 20 Table 2.2: There has been increasing divergence inwealth withinthe richest 25 percent o f rural households ............................................................ 23 Table 2.3: Theil Indices: a decomposition o f within-quartile and between- quartile inequality o f per capita consumption, 1993, 1997 and 2004 .............24 Table 2.4: Differences o f mean consumption betweenconnected and remote households in rural areas (compatible samples), 1997 and 2004................................................................................................................. 32 Table 2.5: Percentage o f population below the total poverty line, by village size and distance to the nearest all-weather road, 2004 .................................. 33 Table 2.6: Ability, family background and education were seen as the most important reasons why some people can get ahead ........................................ 34 Table 2.7: Mean levels o f consumption are significantly higher for households with skilled headsthan those with unskilledheads......................................... 35 Table 2.8: Household consumption and crop yields, by shares of land with land papers, 2004 .................................................................................................... 37 Table 3.1: Gini coefficients of consumption for comparable samples of urban areas (Le,, Phnom Penh and other urban), 1993, 1997 and 2004 .................... 41 Table 3.2: Gini coefficients o f real household consumptionby quartile groups. 1993. 1997 and 2004 ....................................................................................... 44 Table 3.3: Theil Indices: a decomposition of within-quartile and between- quartile inequality o f household consumption. 1993. 1997 and 2004 ............46 Table 3.4: Mean consumptions o f richest and poorest quartiles. 1993. 1997 and 2004................................................................................................................. 47 Table 3.5: Summary statistics o f paid employees inurban areas. 1997 and 2004................................................................................................................. 50 Table 3.6: Effects o f education and industry on wage differences between 1997 51 Table 3.7: Rates o f returnto various levels o f schooling, 1997 and 2004 ....................... and 2004 for urban employees ........................................................................ 52 Table 4.1: Poverty and land-holding status inCambodia by zone and Table 4.2: Landless poor are not more disadvantaged than the landedpoor ...................59 urbadrural sector in2004 ............................................................................... 59 Table 4.3: The richest are 50 percent more likely than the poor to possess Table 4.4: Secure land tenure i s positively correlated with land investments .................64 papers supporting land ownership................................................................... 66 Table 4.5: Simulated effect o f titling programs on poverty ............................................ 68 Table 4.6: Public perceptions o f land titling are unambiguously positive ....................... 71 Table 5.1: Population distribution by national per capita consumption quintile Table 5.2: Percentage o f population indwelling with water-piped or public tap ............77 and by region and region*sector ..................................................................... 77 Table 5.3: The distribution o f population indwellings with lighting, powered 78 Table 5.4: The distribution o f population with access to electricity (%) ......................... by city power, generator or battery ................................................................. Table 5.5: Average distance to nearest all-weather road (inkm)..................................... 79 79 Table 5.6: Average distance to permanent market (inkm) .............................................. Table 5.7: Average distance (inkm) to motor roads by quartile, 1997 and 2004 ............80 81 Table 5.9: Average distances (inkm) to schools, 1997 and 2004.................................... Table 5.8: Distributionso f utilities and amenities by quartile, 1997 and 2004 ...............82 83 Table 5.10: Average distances to primary schools, 2004.................................................. 83 Table 5.11: Provincial averages of total household revenue, crop profits, non-agriculture revenue, and wage income, by existence o f all- weather road invillage, (annual, inRiel)........................................................ 85 Table 6.1: Health outcomes inrich provinces and municipalities are often twice as good as those inpoor areas ............................................................... 94 Table 6.2: There is a large gap betweenthe poor and the rich inrates o fhealth service utilization .......................................................................................... Table 6.3: The poor have to travel considerably further to obtain healthcare Table 6.4: There are notable inequalities inthe quality o f public healthcare ................106 ................105 107 Table 6.5: Increasingly. Cambodians can obtain healthcare without needingto poverty........................................................................................................... resort to measures which cause them to fall into, or remain in, 111 Table 6.6: Rural-urban inequalities inmaternal and reproductive healthcare 115 Table 6.7: Improved water and sanitation has multiple positive effects ........................ are declining; wealth-based inequalities remain considerable ...................... 119 Table 6.8: Wage rate differentials inpaid employment ................................................. 122 Table 6.9: There has been a dramatic expansion in basic education in recent years .............................................................................................................. 124 Table 6.10:Dimensions o f inequality ineducation ......................................................... 127 Table 7.1: Estimates o f domestic violence. 1996-2005.................................................. Table 7.2: Women are under-represented inall aspects of government ........................ 144 147 Table 8.1: As core Government costs have been contained. developmental spendinghas increased.................................................................................. 160 Table 8.2: Spendinghas risen three-fold since 1996...................................................... 166 Executive Summary Overview Over the last one-and-a-halfdecades, Cambodia has achieved high rates of economic growth and significant poverty reduction. However, the benefits o f growth have been spread unevenly, resulting in a rise in consumption inequality. Drawing on survey and other data, this report concludes that the rise in inequality occurred in the early part of the decade (circa 1994-1997) and only in rural areas; there was no significant change in the distribution of consumption between 1997 and 2004. The findings suggest that the pattern of growth is not structurally destabilizing. The report then examines how expansion in service delivery and the provision of infrastructure have begun to reduce inequalities in human development outcomes and increase equality of opportunity; but also how increased concentrationof land ownership, together with insecure land tenure, risk creatingthe conditions for widening inequality in the future. Looking at the roles of institutions in shaping economic and social development processes, the report describes ongoing reforms in public financial management and discusses the potential for using public spending to promote equity and poverty reduction. It notes that practices in the legal and judicial sector fail to meet the ideal of providing equality before the law; and analyzes how gains have been achieved on a number o f fronts by adopting a flexible, iterative approach to institutional reform. Cambodia's changing distribution of income-related outcomes is consistent with the process o f transition from a plannedto an open market economy, and the accompanying growth of incomes. This transformation has promoted better resource reallocation, expanded the spectrum o f gainful activities, and widened the distribution of earnings. Aided by robust economic growth and improved capacity for implementing public policies, Cambodia has seen most other welfare outcomes (notably health and education indicators) become more equal over time. While a radical shift of development strategy is not necessary, improvements in public spending and selected interventions can further promote equality of opportunity, and public policies can promote an expansionof private investment opportunities. The findings of the report point to the following sets of government actions: (i)continuingeconomicreformto promoteinvestmentsandsustaingrowth; (ii)protecting land tenure through systemic titling, and addressing landlessness- related poverty through various measures (e.g., social land concessions; non-farm employment); (iii) building upon progress to date with further improvementsin the quantity, quality and affordability of schooling and healthcare; and underpinningservice delivery reforms with stronger and more responsive state institutions through continued reforms of public financial management and merit-basedcivil service pay. ll Epi&,growth andpoveq redtlctionin Cambodia Background: transition and growth in Cambodia Cambodia has made considerable progress in the one-and-a-half decades since the Paris Peace Accords of 1991, That settlement marked the beginning of a transition from conflict to peace, bringing most of the parties to the low-intensity civil war of the 1980s into an agreement to compete for power through elections rather than military struggle (although full peace was only achieved in 1999 with the final collapse of the Khmer Rouge insurgency). This agreement also clearedthe way for international recognition, an inflow of foreign investment and development assistance, and a transition from an isolated, subsistence-oriented economy to one based on international integration and markets. The results have been high rates of economic growth (Figure 1) and a rise in national living standards. Between the first household survey in 1993/4 and the most - recent in 2004, poverty fell by around Figure 1 Since 1994, growth has been variable but high a quarter (from around 47 percent in 1 13.5% h 1993/4 to 35 percent 11.9% in 2004). Quality of housing, ownership of consumer goods and access to electricity v) 6 8.8% 5.6% have all improved, v L: including amongst the 6.5% poorest fifth of the a population. Non- n c3 monetary aspects of - welfare have also I . 94 1995 96 97 98 99 2000 01 02 03 improved. Health 04 2005 06 service coverage and Source: NIS 2007. child survival outcomes have improved dramatically between the health survey of 2000 and that of 2005. The rise in HIV/AIDS has been contained and reversed; and primary school enrolment has expanded rapidly, and been reflected in rising literacy rates among new school leavers. This report attempts to understand the causes and consequences of this increase in inequality; examine whether the pattern of growth has structural characteristics that are generating social or economic instability; and assess whether Cambodia needs to change policies and public spending interventions. Executive Summa9 lu ... Sharing the benefits of growth: trends in per capita consumption The benefits of economic growth have been unevenly distributed In those parts of Cambodiain which trends can be compared directly (Le., those areas that were covered by the first survey in 1993/4), average living standards, measured as consumption of goods and services per capita per day, Figure 2 Living standards improved at very rose by 32 percent in real different rates, widening inequalities terms between 1994 and 2004. However, this rise was 1994 t! 2004 associated with widening differences between rich and -1 26`4, poor. In 2004, the living standards of the poorest fifth of the population were only 8 percent higher than they were a decade earlier; over this same period, the living standards of the richest fifth rose five times as fast (45 percent). Similarly, rural living standards rose more vE Poorest Wchest Rural Other Rnom slowly than those in Phnom quintile quintile urban Penh Penh and other urban centers (Figure 2). Source: SESC 1993/4, CSES 2004 The result has been a rise I in the Gini coefficient (a commonly-usedsummary measure of inequality, ranging from a value of zero-signifying perfect equality-to one-signifying perfect inequality). The Gini for the comparable sampling frame rose from 0.35 in 1993/4 to Figure 3 The rise in national inequality 0.40 in 2004. In Cambodia as a has been driven by growing whole (that is, looking at the full inequality within the countryside national sample, not just that part of it that correspondsto the 1993/4 ,b - 0.4 sampling frame), the Gini in 2004 was 0.42. 0 3 rural-urban inequalityw [thin Rising inequality has been *8#T 2 0 2 urban areas driven by increasing intra- 2; E 1 rural inequality -E` 3 0 1 O.'" inequality within It is striking that the rise in rural areas total inequality in Cambodiacan be 0 attributed primarily (Figure 3) to 1993/4 2004 widening differences in standards Source,SESC 1993/4,CSES 2004 1v Equity, growth andpovenj reduction in Cambodia of living in the countryside. Although the gap in urban areas remainedlarge, its share in aggregate inequality has fallen. Inequality appears to have widened in the mid-I990s, then stabilized Further insights can be obtainedby breakingthe analysis down into two time periods, (i) 1993/4-1997, and (ii) 1997-2004.Plotting change in per capita consumption for every percentile in the respective samples reveals how growth was distributed for the two sub- periods. In rural areas the level of inequality widened considerably between 1993/4 and 1997 as consumption grew rapidly at the upper tip o f the distribution, but contractedfor the rural masses; however, inequality remained unchangedfrom 1997 to 2004, as growth in consumption was similar for the rich and the poor (see growth incidence curves in Figure 4). Figure 4 In rural areas, the rich initially made rapid gains while others lagged; later, improvements were spread evenly across the population - Changes in real consumption, percentile-to-percentile Growth incidence curve Rural, 1993-1997 (in 1993 sampling frame) 0 Percentile o f thezsI population (ranked by per capita household consumption) I I I so 7s loa Growth incidence curve Rural, 1997-2004 (in 1993 sampling frame) 0 Percentile of the21population (ranked by per capita household consumption) 10 7, la0 Source: SESC 199314,CSES 1997, CSES 2004. Note: Only comparable samples from the identical (1993) sampling frame are used for each survey. Executive Summay V In urban areas, inequality remained stable throughout. Table 1 Urban consumption inequality Although urban inequality changed appears not to have risen-but little over the last decade, the is still much higher than in the absolute levels of inequality in countryside Phnom Penh and other towns-were 1993 2004 more pronounced than in the Rural 0.265 0.354 countryside (Table 1). It is worth (i0.005 (* 0.007) bearing in mind two caveats. The Urban 0.434 0.431 first is that there are reasons to (* 0.010) (i0,009) suspect that consumption measures PhnomPenh 0.393 0.367 under-estimatethe full magnitude o f (* 0.023) (i0.014) inequalities. Especially in urban Other Urban 0.439 0.434 settings, income or wealth which (* 0.037) (i0.018) may be more appropriate than Cambodia 0.347 0.403 consumption in obtaining a rounded (* 0.006) (i0.005) picture of the extent of inequality, Source: SESC 1993/4,CSES 2004 tends to show more dispersion than consumption. The second is that the small number of extremely rich households in any population are very unlikely to be interviewed in a household survey. Such households are more numerous, and their omission thus more significant, inurbansettings. As such, household surveys that under-sample the extreme rich or poor probably underestimate total urban inequality. With these caveats in mind, these inequality trends suggest that the nature and patternof Cambodia's growth have not been structurally destabilizing. What explains changing trends in consumption inequality? In urban areas... Economic growth in Cambodia has been heavily reliant upon garment manufacturing, tourism and construction,all of which are concentrated in urbancenters, and in particular Phnom Penh, Siem Reap town and Sihanoukville. The expansion of trade, investment and employment created two countervailing forces on the distribution of urban incomes. On one hand, global integrationand an increasinglydiverseurban economytendto widen the inequality of earnings, as wages for skilled labor rise much faster than those for non- skilled labor. On the other hand, there were also equalizing forces arising from the structural transformation of the urban economy. Trade and foreign direct investment resulted in a higher labor force participation in urban labor markets and a shift of employment from the low productivity sector of agriculture to higher value-added manufacturing (especially garments) and services (especially construction and tourism). The rapidly growing sectors in the economy have also been those that are intensive in unskilled labor, Cambodia's relatively abundant resource. In rural areas... The initial widening and subsequent stabilization o f inequality amongst the rural populationappears to reflect the interactionbetweenthree different sets of factors. v1 Eqaity,growth andpovenj redaction in Cambodia Firstly, geography plays a key role. As growth and public services improved first around urban centers and in accessible densely-populatedlowland areas, households in these areas enjoyed better access to factor and product markets and pulled further ahead of those in remote areas, resulting in widening differences. Over time, as conflict subsided, and in 1999 ended with the integration of the Khmer Rouge, infrastructure investments and improvements in schooling and healthcare reached these peripheral areas. The report's analyses find that the incidence of poverty is lower and mean consumption levels are higher in connected villages with easier access to road than in remote villages (Table 2). Even with the end of conflict, however, remote areas are more likely to have poorer soils and more limited potential for irrigation. Table 2 There is greater poverty in remote villages with less connectivity Village size Distance to the nearest all-weather road Less than 5km More than 5 km More than 1,000persons 33% population are poor 38% population are poor Less than 1,000 persons 43% population are poor 53% population are poor Source: CSES 2004. Secondly, rural households differ in their abilities to take advantage of new opportunities or manage risk. Innate individual abilities, differences in human capital (education and health status), and different endowments of land and other productive assets enabled some rural families to pull ahead oftheir neighbors. Analyses in the report suggest that differential stock of human and physical assets have promoted disparity in incomes. Increasing returns to higher levels of schooling attainments as well as higher yields and revenuein householdswith greater land investments havewidened disparity in rural incomes(Figure 5). Figure 5 Investments in productive asset matter - Provinces with greater human capital (e.g., higher levels of education) and more physical capital (e.g., land investments) have higher inequality 500h provincial Gini coefficients ver8us provincial Oh 1 -/. population with 10+years of schooling 80% - + * * 270% - E 960% - ++ `50% + +s-`5-+ n - $40%- + + + f + cb30% - :c20% - ?O% - I 0% : lo% 1 Pooled Rovinces for years 1997 and 2004 l o 20 40 60 8C 5 10 % households with over half of their % provinoialpopulation with lOyrr+rchoolln0 plots irrigated in at least 1season Source: CSES 2004 ExecutiveSummarv Vii Finally, local governance and the quality of local institutions play a role. Through social connections or simply greater capacity to pay bribes, wealthier rural households find it easier to navigate governance structures which are more often than not characterized by limited resources, professional capacity, transparency and accountability. These institutional weaknesses are reflected in low trust scores (Table 3). Faith in local authorities is particularly low in communities which depend heavily upon access to and management of natural resources such as forests and fishing waters, where rent seekingand corruption work to the benefit of richer, betterconnectedhouseholdsand against the poor (CDRI2007a, 2007b). Table 3 There is limited trust in public institutions, particularly police and the courts Institution I actor honest effective treat poor and rich treat men and equally women equally NGOs 73 78 71 86 banks 69 80 58 81 media (TV, radio) 50 ._- 64 ~- _. -I___._I_. 6 3 - _ !?."_ _ national government _I____-_-_ I 47 60 40 80 CommuneI sangkat officials 44 55 46 79 teachers 59 68 60 82 healthcareprofessionals 48 66 40 79 money lenders 41 57 36 75 POllCe 29 50 31 75 judges 18 46 20 70 middlemen 10 44 29 67 Source IRL 2007 Note An example of the statement was "NGOs are honest", with the respondent asked to agree or disagree on a scale from 1 ("not at all") through 2 ("somewhat"),3 ("neutral I don't know"), 4 ("much") and 5 ("very much") The percentageof respondentswho picked 4 and 5 to the statement are recorded in the Table above. Creating equality of opportunity Equity in service delivery and human development Equity in effective access to basic services and in human development outcomes are important both as ends in their own right and as means to the ends of equality of economic opportunity. Intrinsic value is particularly obvious with regard to health: no- one chooses to suffer illness or early death when they have the option to avoid them. However, health and education status also have a very important instrumental value, as determinants of economic opportunities and predictors of income, consumption and wealth. Finally, there are very strong cross-sectoral links. The level of a mother's education exercises a strong influence upon not only her own health but also that o f her children; conversely, illhealthis a major barrier to obtaining an education. Viii Equio,growth andpoveq reduction in Cambodia In contrast to living standards as measured in terms of per capita Figure 6 Inequalities in literacy are closing consumption, non-monetary over time have shown a broadly positive g 84 84 a3 83 80 trend towards greater equality over 5 * * * 77 75 time. Gains in education in recent h: 6: - a 3 0 6o 55 years are notable for their pro- 70 O . B 74 0 . 0 0 * poor, pro-girl bias. Primary E 66 60 60 62 57 enrolment has increased fastest f 45 . *36 amongst these traditionally under- 6 0 - El Urban 39 represented groups, resulting in pP 0 Rural 2a shrinking urban-rural, male-female T I , , , , , , , , , , , - Nonetheless, significant 2 86 a2 79 80 7a 76 7a a3 a3 differences remain between urban S * * 77 70 and rural populations and rich and p - 8 2 59 poor households in access to c 71 0 E - 0 m 62 65 60 services and outcomes. While 0 0 . 53 54 52 access to primary schools and to a 02 42 lesser degree health centers have 0 Female 0 0 become more equal over the last 2> 25 - 19 0 7 I I I I I , I I I , ! G > ExecutiveSumma? ix Figure 7 Child care and survival by wealth Consequently, inequalities in Children born in richer households access to services and resulting receive better preventative care health- and education-based from before they are born... opportunities do still vary I Ln considerably. The life chances o f a w Poorest child born into the poorest quintile are significantly reduced even Next before birth (see Figure 7): poorest compared to a child born into the Middle richest quintile, his or her mother is four times more likely to have gone without any contact with 0Next richest antenatal care. A child born into Mother received no the poorest quintile is then over antenatal care during twice as likely to suffer from pregnancy severe malnutrition; and three times as likely to die before ...are less likely to be reachingtheir fifth birthday. malnourished ... There is also a pressingneed to improve the ability of the health system to respond to key health issues faced by women. Although data issues make definitive interpretation hard, it appears that the maternal mortality rate has remainedunchanged at a very high level (472 per 100,000 live births). Poor households are more Stunted Underweight likely to need to withdraw their ...and children from school to help make much less likely to die before age five. ends meet; and more likely to have to adopt extreme measures to manage the costs of a health crisis (see Box 1). Low education, low income and consumption, and vulnerability to illness and illness- induced economic shocks act in mutually reinforcing ways to limit opportunities and trap households in poverty. infant rortality rate Under-fivemrtaliy (deaths c 1year old) rate (total deaths c 5 years) Suorce: CDHS 2005 x Equig, growth andpoveq reduction in Cambodia ______I_____I______I_ll_lI_ I ___I____I________ Box 1 Poor households lack the resources to cope with health shocks Results from surveys in two villages south of Phnom Penh suggest that while crop failures and illnesses entail similar magnitudes o f economic damage (a few hundred thousand riels), households find it harder to cope with illness. The negative consequences from health shocks have greater impact on both immediate and long-run livelihood because they require an 'immediate lump-sum of money: as most households do not have sufficient savings and rural credit markets do not operate well, households are often forced into distress sale of productive assets (including land) and/or enter long-term debt, reducing their future income streams and increasing their non-consumption expenditures, respectively. This broadly confirms earlier studies which found that half of all distress sales, or around 40 percent of cases o f once- landowningfamilies losing land, involvedhealthcrises. Sources: Yagura 2005; Biddulph2004; Ballardand So 2004. Land ownership is highly concentrated In 1989, land was privatized and distributed largely on the basis of need (household size), resulting in a highly equal distribution and effectively zero landlessness. Since then, landlessness and inequality in landholding have risen, reflecting the interaction between population growth and land management policies which designate much of Cambodia as state land and prioritize large-scale Figure 8 Land ownership is highly unequal concessions when allocating this land to use in cultivation. As a result. bv 2004. 5.7 Thailand 0 41 percent of the rural population 'naoneS'a -' . . 0 49 was both landless and poor, Malaysia and the distribution of land in hdia Cambodia is amongst the most unequal in the region (Figure O o d i a 8). Gini coefficient (0 = perfect equality, 1= perfect inequality) Inequality in ownership of Source: WDR, IFAD a key productive asset such as land is a major concern because it implies the likelihood of a trend to growing inequality of production, income and consumption in the future. In the absence of dramatic expansion of off-farm employment or livelihood opportunities to absorb landless labor, the concentration of land in fewer hands implies a growing gap in the opportunities available to different types of households. There is also a strong growth case to be made against polarization in land holdings. Evidence from Cambodia, as in most other rural- majority developing countries, shows agricultural productivity in an inverse relationship with size of farm, so that small farms are significantly more productive than are large farms. All things being equal, equality and efficiency (growth) are fully complementary. There is thus much to be said for the proposed program of redistributing idle economic land concessions (which it was hoped private investors would develop as large-scale commercial plantations) to local landlessand land-poor householdsthrough a program of social land concessions. There are numerous challenges-redistribution requires logistically complex complementary policies and programs to be effective-but the case Executive Summary xi remains strong for a Government agricultural policy that prioritizes assistance to small- scale family farms. The second major policy issue in land administration concerns the security of land tenure. Analyses in the World Bank's Poverty Assessment show that secure land tenure, as manifested in various forms of paper proofs of land ownership, has a significant impact on land productivity, profitability and agriculture income. The importance of tenure security is corroborated, in this report, by additional analyses of the CSES 2004, semi-structured interviews during field research, and results from a national poll. This report also finds that secure land tenure, in forms ranging from paper receipts to titles, is highly correlated with land investments. As the price of land rises and local markets become more integrated, full land titles will be increasingly usehl for avoiding disputes, encouraging productivity-enhancing investment, and protecting against outright seizure by more powerful actors. Systematic land titling began in 2004 and has resulted in the allocation of almost one million titles. Early indicators from small-scaleresearch suggests that the receipt ofthese titles is perceivedto improvetenure security and make disputes less likely; make it easier to sell to buyers from outside the community; and improvethe terms on which credit can be obtained. Infrastructure and inequality Infrastructure is a powerful influence upon livelihood opportunities(particularly off- farm opportunities) and welfare outcomes: household incomes in villages with an all- weather road connectionare typically twice as high as incomes in villages without a road. As such, the uneven distribution of economic infrastructure (in particular roads and irrigation structures) helps perpetuate differencesbetween localities; and the extension of such infrastructure can be a powerful tool to start reducing these spatial ,Box 2 Roads reduce spatial inequalities. Equality or difference in opportunities for the poor inequalities and expand access to roads interacts with the , distribution of social infrastructure Before we had a good road, the poor, ' (schools, health centers, referral 1especially female heads of households, rarely hospitals)to create spatial inequalities in idared to go to sell labor farporn the village ... access to basic servicedelivery. It took a long time to reach where they wanted to go [Now] it costs only 1,000 riels and ~ Inequalities in access to ... infrastructure are pronounced, first and jtakes an hour to travel to Prey Veng, compared foremost between urban and rural to 1,500 riels and four hours before the road populations, but also within both was built ... Thepoor landless and others have migrated farther from the village in search of segments. Both national surveys and ~ work; before 2002, when travel was expensive locality studies confirm the importance and it took longer to reach the work of road construction-and, critically, destination, they had to think twice. (Babaong road maintenance-in improving village, Prey Vend opportunities and livelihoods (Box 2). Source:CDRI 2007b (forthcoming) I I xii Eqztity,growth andpoveq redztction in Cambodia Perceptions of inequality In Cambodia, different tools suggest slightly different pictures with regard to the perceptionof inequality and its severity. In a national public opinion poll, when asked to specify the most important problems facing Cambodia, most (33 percent) focused on poverty, with only around 2 percent mentioning inequality and unfairness, a proportion similar to those who mentioned roads, education and health care. However, qualitative research also suggests that people are very conscious of and concerned about growing inequalities in opportunities and outcomes (Box 3). Box 3 Rural Cambodians perceive and worry about growing inequality As of now, the gap between the rich and the poor is getting wider. The rich get richer and the poor getpoorer and it is not easy to move upward, (Krasaing Village, Battambang) Over time, stratification has been getting worse. ... It will be very hardfor the poor to move out of poverty in the nextfive years because of landlessness, lack of capital and growing inequality. Thegap between the rich andpoor has grown tremendously. (Andong Trach, Battambang). The poor remain poor because the poor feed the rich. The rich are getting richer because they have capital to invest to make more money. ...The rich drop somethingfor thepoor to continue, but thepoor earn tofeed the rich. (Prek Khmeng, Kandal) Source: CDRI 2007b (forthcoming) . _ I Gender-based inequalities remain significant Gender norms and opportunities are shifting significantly in response to economic and social change. Positive trends towards greater equality include increasing girls' enrolment in primary education (and resulting rise in female literacy) and expanded employment opportunities: the vast majority of garment factoryjobs are female. However, significant traditional inequalities persist, and new ones are emerging. As women are assumed to have primary responsibility for childcare and domestic tasks, they have limited opportunities for more remunerative employment. These difficulties are reinforced by lower average levels of education, prevailing attitudes regarding "appropriate" occupations for women, and the fear of trafficking or sexual violence when traveling away from home, especially to urban centers or across the border. With migration and more disposable income, new, risk-taking patterns o f male behavior are emerging and resulting in new forms o f vulnerability for women (vulnerability to sexual harassment or to contractingHIV from their husbands). Finally, women lack significant voice in social or political affairs, with very limited representation in Government or policy-making or planning processes, from the local level to the national. While 20 of the 123 parliamentarians elected to the National Assembly in 2004 (16 percent) and 14.6 percent of council seats filled in the 2007 Commune /Sangkat council elections are women, women constitute only two out of 37 Ministers (5.4 percent) and 8 out of 127 Secretaries of State (6.3 percent). In the total labor force, one-third of professional and technical workers are women but only 14 percent of legislators, senior officials and managers are women. ExectltiveStlmmaty xiii Inequality and public action: the role of institutions Public spending and equity The state can influence national economic and social development through several channels. The first is through raising revenue and directing public spending. Cambodia's public financial management (PFM) system has a number of serious problems, but a PFM reform program over the last two years has begun to make significant headway in reforming this system. These reforms open up possibilities for more effectively and efficiently directing public expenditure towards poverty reduction. Priorities would include: .. Increasingrevenue collection (particularly through improvedtax administration) Putting in place strong, transparent measures for the effective use of the very . considerable revenue that is expected to flow from offshore oil and gas within a few years Reallocating spending towards priority goals (shifting spending from general .. administration to priority sectors; changing the balance of ODA from technical assistance to capital spending) Increasing capital spending with emphasis on economic sectors (roads, irrigation, energy) that shape the distribution of opportunities andthe patternof growth. Increase recurrent spending, both on maintenance (to sustain the value of capital investments) and, critically, on public sector salaries. Better pay for public sector workers, combined with better management, is essential to create strong state institutionsthat can formulate and implement nationaldevelopmentstrategy. Current legal practice falls far short of the ideal The allocation and management o f spending is not the only tool by which the Government can guide the course of national development. The Government can also influence economic and social change through its regulatory functions: that is, by setting and implementing laws and policies that shape incentives and interaction between different actors. However, Cambodia faces some quite pronounced difficulties formulating and implementing laws and policies. State capacity remains low, reflecting limited financial and material resources; the small numbers of well-educated public servants; limited data with which to design and monitor evidence-based, results-oriented policies; and institutional arrangements which are, as in many low-income countries, not well aligned to responsiveand accountable policy-making. In particular, the legal andjudicial sector in Cambodia notably fails to deliver services which conform to the ideal of equality before the law. Lack of resources and low levels of transparency and accountability are reflected in widespread distrust of the police and courts, which are seen as either corrupt or subject to political direction. Faced with a dispute, most Cambodians will seek to resolve it without recourseto the court system. XiV Equity,growth andpoveq reduction in Cambodia Fair institutions typically evolve in stages Both theoretical and applied thinking now emphasizes the role of institutions in shaping economic and social change. However, limited success has been achieved with attempts to transplant ideal models of "international best practice" from one country to another: effective institutional arrangements for given policy outcomes are likely to vary considerably from one country to another. This is in part because in Cambodia as in many countries, formal state law is in practice only one among several normative reference points: values and behavior are also shaped by "traditional" understandings of rights (e.g. of a household's right to clear land for subsistence farming or extract resources from forests), or by perceptionsamongst powerful actors that position confers discretionary rights to manage or dispose of public resources in ways not bound by formal rules. This legal pluralism is by no means unique to Cambodia, but does create a major challenge for Government. Rather than attempt to create organizational structures perfectly designed to achieve defined public policy functions, more success may be achieved with a more iterative- and messy-process that acknowledges the useful role to be played by "interim" institutions. In a number of cases, progress in Cambodia has been achieved in an enclave, in the hope of developing robust institutions and later building outwards, rather than through an attempt to implement a comprehensive across-the-board reform of institutional arrangements. Thus, ties to trade quota agreements were used to create incentives for achievingcore labor standards in the garment export industry. Principles of participatory and accountable local planning were introduced through donor-supported "project law" (new institutional rules and principles developed under the Seila program): over time, these have taken root and come to influence Government thinking on the direction of decentralization and deconcentration policy. In each case, outcomes fall short of the ideal that is envisaged under the liberal concept of the rule of law, but do mark a significant advance on what existedbefore, and do open up possibilities for future stepwiseprogress. Summary: findings and policy implications Using the national household surveys, qualitative studies, and opinion polls, the report finds the following: Rural inequality of consumption rose during the last decade, driving up aggregate inequality in Cambodia, while urban inequality remained unchanged. Rural inequality of consumption rose during 1993-1997, but then stabilized between 1997 and 2004. In contrast to consumption inequality, inequality in non-consumption measures of living standards such as school enrollment and mortality rates has fallen over time. The disparity in access to infrastructure (roads), amenities (clean water, electricity), and basic services (schooling, healthcare) has also narrowed between 1997 and 2004, as have gender gaps in literacy and schooling. Inequality in rural and urban areas comes from two sources: the gap between the richest 25 percent and the rest, and inequality within the richest quarter of the Executive Summa? xv population. Rising rural inequality was mainly the result of rising inequality among the richest quarter. Based on the findings and analysis, the report concludes that policies that directly tackle poverty will also address inequality, because income growth of the poor will narrow the rich-poor disparity. This suggests a number of policy priorities for Cambodia's responseto inequality concerns: First, there is a need for continued effort in economic reform and improvement of investment climate. Maintaining the economy's openness and global outlook will be the best strategy for continuedgrowth and sustained poverty reduction. Second, priority in land policy should be to accelerate and intensify systemic titling, extending it throughout the country, including to remote, more conflict-prone areas. Land policies within the agricultural strategy should give priority consideration to household farming, including by redistributing idle plantation land to landless poor households. Third, more needs to be invested in building and then maintaining basic infrastructure to connect up remoter parts o f the country to economic opportunities and improved social services. Rural roads clearly have a significant impact in equalizing geographical inequalities. There is also a pressing need for increased investment in improvedwater supply and sanitationin rural areas: the rural-urbangap in access to these facilities remains extremely wide. Fourth, there is a need to build upon progress to date with further improvements in the quantity, quality and affordability of basic education, with an emphasis upon tackling the causes of late enrolment and drop-out. Equally important is a comprehensivenation- wide adult literacy program, as a first step to replenish (and equalize) the human capital of the adult generation. Fifth, healthcare provision needs to be focused on preventative and primary care, with emphasis on expanding physical access and affordability (targeted fee exemptions backed by equity funds) and concerted effortsto improve maternalhealthand bring down the persistently high maternal mortality rate. Expansion of investments in improved water supply and sanitation would play a major role in reducing disease burden and the economic costs associatedwith it. Lastly, significant improvements in service delivery will depend on creating stronger and more responsive state institutions. Continued reforms of public financial management, with emphasis upon improving civil service pay and management and putting in place arrangements for the transparent and effective use of future oil revenues, will be critical. 1. Recent economic and social developments - . . - Summary Over the last year, Cambodia has continued to show strong growth, with garments and tourism both expanding by 20 percent. The balance of payments has improved and FDI reached a new high. Newly-released health indicators also show considerable improvementover the last five years (with maternal mortality a notable exception). This report analyzes the evolution of living standards (per capita consumption) over time and the changes in distributions of accesses to services and assets. The purpose of the report is to analyze the forces underlying the observed rise in inequality of consumption between 1994 and 2004, when the Gini coefficient rose from 0.35 to 0.40. The report aims to find out whether the pattern or nature of Cambodia's growth has been destabilizing as well as whether Cambodia's policies have perpetuatedinequality of opportunity or poverty traps. Analysis begins with an examinationof headlinetrends inconsumptioninequality in rural and urban areas, based on data form the 1993/4 and 2004 household surveys (recognizing the limitations of this type of data). Findings suggest that within the general population, growth and rising living standards in urban areas has been pro- poor. In rural areas, by contrast, inequality has increased. Inequality of non-monetary dimensions of well-being, specifically health and education, has in most respects narrowed between 1997 and 2004. Access and outcomes have improved for all, and in particular more rapidly among disadvantaged groups (e.g., the poor, girls, and the rural population). Thus, disparities of healthand schoolingoutcomes betweensub-groupshave decreased. - .. - . - - l_l How has the Cambodian economy evolved? Cambodia's economy has undergonedramatic changes over the past two decades, the most significant being from central planning to a market economy. Beginning in 1989, private property rights were reintroduced, price control abolished, state-owned enterprises privatized and investments liberalized. The Paris Peace Agreement in 1991 led to a UN-sponsoredelection in 1993 and the establishment of the Royal Government of Cambodia(RGC) in September 1993. Annual GDP growth averaged 8.4 percent per annum over 1994-2006. During 2004- 2006, Cambodia grew at double digit rates, averaging 11.4 percent per annum for three consecutive years. Consumer price inflation dropped from an average of 56 percent over 1990-1998 to less than 3.0 percent in 2006. Growth continues to be driven by garment exports, tourism, construction, and agricultural expansion. The garment sector, employing 10 percent of the total labor force and accounting for 14 percent o f GDP, expanded with exports rising by 20 percent in 2006. During 2006, 20 percent growth in 2 Sharinggrowth: egg@ and developmentin Cambodia tourist arrivals resulted in a 26 percentjump intourism receipts.Agriculture continuedto expand at 5.5 percentwhile construction slowed slightly. External developments were positive in 2006. The balance of payments improved as the current account deficit (excluding transfers) declined to -7.7 percent in 2006 (from - 9.6 percent in 2005) and gross internationalreserves expanded by 20 percent to US$ 1.1 billion. FDI continued its upward trend, reaching a record high of US$ 475 million. Consumer price inflation declined to 2.8 percent (from 6.7 percent in 2005) and is expected to remain low (below 5 percent) for 2007 and 2008. IncreasingFDI is expected to be sustained and the discovery of offshore oil and gas reserves will likely bring about even higher growth, with production expected to start in 2009 or 2010. However, the new oil sector will also pose additional challenges to macroeconomic and fiscal management. The garments sector is increasingly facing stiffer competition from Vietnam's accession to the WTO and greater competition from China in 2008. A recently released demographic and health survey shows considerable progress in health outcomes. Infant mortality fell from 95 per 1,000 live births in 2000 to 66 in 2005; child (under 5) mortality decreased from 124 deaths per 1,000 live births in 2000 to 83 in 2005. HIV prevalencealso appears to have fallen remarkably.The notable lagging indicator is the maternalmortality ratio, which appears to have remainedunchanged. Why examine inequality? The principal motivation for this report is the finding in the 2006 Cambodia Poverty Assessment that inequality rose in Cambodia (World Bank 2006). The second motivation is the synthesis of international experience with regard to growth-poverty-equity relationships contained in the 2006 volume of the World Bank's annual global flagship report, the World Development Report (WDR) 2006: equity and development (World Bank 2005). With peace, Cambodia has enjoyed growth and poverty reduction Cambodia's transition from civil war to peace, from one-party to multiparty politics, and from an isolated, subsistence-oriented economy to an open market economy, has made possible impressive rates of economic growth (averaging 7.7 percent per annum between 1994 and 2004) which has in turn resulted in rising average levels of per capita income and consumption. With financial and technical assistance from external partners, basic services have also begun to improve. The Poverty Assessment found that poverty fell from 47 percent to 35 percentbetween 1994 and 2004 (Box 1.1). However, inequality has risen Even as all wealth groups have, on average, benefited, the benefits of growth have clearly accrued disproportionately to the wealthier groups in society, widening the gap betweenthe rich and the poor. Within the sample of comparable areas, the average per capita consumption of the poorest quintile in 2004 was 8 percent higher than in 1994; for the richest quintile, by contrast, consumptionwas 45 percent higher in 2004 than in 1994. Recent economic andsocial developments 3 Box 1.1 Poverty trends in Cambodia, 1993/4 to 2004 It has been hardto establish trends in poverty in Cambodiabecause questionnaire design and sampling have changed from one survey to the next, making it hard to compare findings. Most importantly, the geographical sampling frame has expanded each time. In the first survey, the 199314 Socio-EconomicSurvey of Cambodia (SESC), peripheral areas that were hard to reach and which were still under Khmer Rouge control were not covered: so, this cannot be directly compared to the 2004 Cambodia Socio-Economic Survey (CSES), which sampled from the whole country. However, looking only within the same (SESC 1993/4) sampling frame in 2004, poverty fell from 39 to 28 percent over the decade. Onthe basis of this trend in directly comparable estimates and what is known about the distribution of population and poverty inside and outside the sampling frame in 2004, it is estimated that poverty in the whole of Cambodia fell from between 45-50 percent in 199314 (the point estimateis 47 percent)to the nationalestimateof 35 percentseen inthe 2004 survey. Figure 1. 1 From 1994 to 2004, per capita consumption amongst the poorest fifth has risen by only 8 percent-compared to 45 percent amongst the richest fifth of the population - Real average per capita household consumption, 1994 and 2004, within the comparable sample frame (riels, 1993/4 constant prices 7,500 i 1 5,000 2,500 0 Poorest Next Middle Next Richest Cambodia poorest richest Source: SESC 1994, CSES 2004. Growth, inequality and poverty Drawing on the WDR 2006 (Box 1.2) and other sources, we summarize the main arguments for fostering equity in public policy. The WDR 2006 argues that very high levels of inequality and poverty traps are fundamentally inimical to the efficient operation of a market economy. There are several different ways in which high levels of inequality and absolute deprivation may contribute to market failures and reduce investment and growth. In outline, there are three main effects that are generally distinguished: unequal access to economic opportunities, due to the inability of the very poor to participate in markets for credit or invest in human capital development, results in an inefficient allocation of investment. 4 Sharinggrowth: egui5 and developmentin Cambodia 1I Box 1.2 World DevelopmentReport 2006: Equity and development The 2006 WDR analyzes the relationship between equity, growth and poverty reduction, both within and between nations. The report documents the persistence of inequality traps, highlightingthe interactionbetweendifferent forms of inequality; and presents evidencethat the inequality of opportunity that arises is wasteful and often weakens prospects for overall prosperity and long-term growth. It derives policy implications that center on the broad concept of leveling the playing field, both politically and economically and in both the domestic and global arenas. The report recognizes the intrinsic value of equity but focuses on the instrumental case that attentionto equity supports long-rundevelopment. WDR 2006 is structured in three parts. Part Ireviews the evidence on inequality of opportunity, within and across countries. Part I1 asks why equity matters, discussing the two channels of impact (the effects of unequalopportunities when marketsare imperfect, andthe consequences of inequity for the quality of institutions a society develops) that give equity an instrumental importance, as well as intrinsicmotives.Part 111asks how publicaction can levelthe politicaland economic playing fields. Inthe domestic arena, it makes the case for investing in people through equitable access to services; expandingaccess to justice, land, and infrastructure; and promoting fairness in markets. In the international arena, it advocates leveling the playing field in the functioning of global markets and the rules that govern them-and the complementaryprovision o f aid to helppoor countries andpoor peoplebuildgreater endowments. Source: World Bank 2005. . - - - - _I pronounced inequalities in wealth give rise to institutions (for example, the institutions regulating the protection of property rights or the institutions that . determine the voice of individuals in influencing Government policy) that systematicallyfavor the rich over the poor. Finally, highly unequal societies (especially highly unequal societies in which many are very poor) are typically proneto social and political instability. This suggests that, within sensible boundaries, improving equity promotes growth. Extreme deprivation interacts with market imperfections (in the form of asymmetries in access to information and credit, and in ability to take on risk) which divert investments in capital (including human capital) away from optimal allocations: without any asset base, the poor cannot obtain access to finance and cannot afford to invest in their children's education. In highly unequal societies, these growth-constrainingeffects are often embedded over time in institutions which consistently favor the rich over the poor, regardless oftheir innateabilities. This basic ethical aversion to extreme inequity would seem to be present in Cambodia, too. The concepts of equality and equity seem to translate well between Khmer and English (see Box 1.3): while the place o f these concepts in the Cambodian worldview is complex and difficult to summarize without grossly over-simplifying (as in any society), it would be hard to claim that they have no indigenous roots. Examining these social values needs to recognize some distinctive cultural legacies (e.g. the concept of merit, which is sometimes taken to imply that luck or misfortune in this life reflect an individual's good or bad behavior in a previous life), without oversimplifying Cambodian belief systems, or implying that values do not change over time-which they clearly do, and have. Interestingly, monks interviewed during the 2001 participatory poverty Recent economic and social developments 5 assessment explicitly rejected most of the karma-basedexplanations of poverty, locating causality for wealth differencessquarely in secular processes(ADB 2001). Box 1.3 Defining terms and concepts: equality and equity The key concept of this report will be equity. Given that the English words equality and equity can be used by different people to mean different things, it is worth trying to define these terms with some precision. Whereas equality is a positive concept (that is, one that is objective and measurable), equity is a normative concept (that is, one based in values). Followingthe WDR, equity is understoodin this report to encompass issues of both relative and absolutedeprivation, andto incorporatetwo components: . equality of opportunities: a concern with equity implies attention to the fairness o f wealth generation processes, and justifies interventions to ensure that individuals are able to . benefit from their innate abilities (talent and hard work), regardless of the wealth, ethnicity, geographical locationor sex into which they are born. aversion to extreme poverty: a concern with equity implies that public action (by state or non-state actors) is necessary to help those whose standard of living falls below a minimumthreshold, regardlessofthe processesthat have broughtthem there. The relatedbut distinct meanings attachedto "equity" and "equality" inthe WDR map quite well with equivalent Khmer concepts. The Khmer word usually given as the translation for equity (samatoah) incorporates the root sama (common to both Sanskrit and Pali), meaning "even, equal, impartial": it thus relates to the "fairness" of rules or process (or, in a modern phrasing, to equality of opportunity or treatment). By contrast, samakpheap or pheap smoer khnear relate to a state o f beingthe same, equivalent to the concept of similar outcomes inthe English"equality". As outlined below, both logic and evidence suggest that inequality traps (inequality of opportunity and pockets of extreme deprivation) result in growth which is slower and less sustainable than it might otherwise be; and, conversely, that tackling inequity is likely to I sustain or increase rates of growth (andgrowth-ledpoverty reduction). While trade-offs may I be necessary in the short term, equity (or distributionalconcerns) and efficiency (or growth) are at a fundamentallevelcomplementary over the longterm. I In a public opinion poll conducted recently by Indochina Research Limited (IRL) for the World Bank, respondents were asked to list (without prompting) the three most important issues facing Cambodia that the Government should tackle. Most respondents (33 percent of respondents) chose "poverty". Poverty was also the most frequently quoted pressing problem, according to the richest and best educated respondents. Inequality did not feature significantly (2 percent o f respondents) in the answers volunteered. However, if people focus first on absolute outcomes (their own living standards and problems encountered in making a living) rather than relative position and process (how they compare to others, equality of opportunity), they do not appear to be insensitive to equity issues. When asked "Do you think inequality of wealth is an issue?," 88 percent answered yes. Finally, concern about growing inequality of wealth, and the implications of this in terms o f growing inequality of opportunity, emerged strongly in mixed-methodslocality-basedresearch (see Box 1.4). 6 Sharinggrowth: epio and developmentin Cambodia Figure 1.2 Rich and poor Cambodians care about the same issues, albeit with different emphasis 7% c3 > $300 la $107-200 10% 11% % of respondents (NB could list up to % of respondents (NB could list up to 3 issues) 3 issues) Source: IRL 2007 That inequality does register as a concern is important as the popular legitimacy of a particular elected government-and of a political system more broadly-is strongly affected by citizens' perceptions as to whether this system operates in the interests of all citizens, rather than just a privileged few. Low levels of state legitimacy make it harder to govern effectively: if it is widely believed that wealth differences result from unfair rules (i.e. that the rich are rich largely because of their connections), this undermines people's faith in politicians and officials, making it harder for Government to retain people's loyalty and ensure their compliance with law and policy'. High levels of inequality also undermine the trust that is needed for citizens to work together to solve their own problems through collective action; and is clearly related to the incidence and severity of crime within society (Demobynesand Ozler 2005). Sources and measures A core measure of welfare is per capita consumption: that is, the average value, in monetary terms, of food, clothing, housing and so on, consumed by an individual over a ' See Mekong Economics 2006 for this argument as applied to the analysis o f growing inequality in Vietnam. Recent economic and social develobments 7 ' Box 1.4 Qualitativeresearchshows widespreadconcern with growing inequalities As of now, the gap between the rich and the poor is getting wider. The rich get richer and the poor get poorer and it is not easy to move upward. (KrasaingVillage, Battambang:MOPS) In 1993,we were not so different. At that time we [each] had around 5ploan [0.5 ha.] of rice land that was given by the government. ..,There is now a huge difference. The rich are reaching the clouds and the poor are sinking deeper...The rich lend out money and make a profit ... whereas the have-nots are getting worse off and sooner or later sell all their land, (KompongThnoat, Kampot: MOPS) Over time, stratiJication has been getting worse. .., It will be very hardfor the poor to move out of poverty in the next five years because of landlessness, lack of capital and growing inequality. The gap between the rich and poor has grown tremendously. (Andong Trach, Battambang:MOPS). have capital to invest to make more money. ...The rich drop something for the poor to Thepoor remain poor because thepoor feed the rich. The rich are getting richer because they I continue, but thepoor earn tofeed the rich. (Prek Khmeng,Kandal: MOPS) Source: CDRI 2007b (forthcoming). I I specified period. Per capita daily consumption is recorded in national multi-purpose surveys such as the 1993/4 Socio-Economic Survey of Cambodia (SESC) and the 1997 and 2004 Cambodia Socio-Economic Surveys (CSES): the consumption level of each individual is compared to the consumption poverty line level, and the proportion with consumptionvalues falling under the poverty line defines the poverty rate. In this report, we attempt to describe and explain changes in the level of inequality of consumption along various dimensions (rural and urban, richest and poorest quintile, those with a secondary educationand those with none, and so on). Chapters2 and 3 will, respectively, answer the following questions: .. Why does the level of inequality o f consumption appear to have risen in rural areas, but appearedto remain unchanged or fall in urban areas? What contributedto the high levels of inequality in levels of consumption in certain segmentsofthe populations? Per capita consumption as derived from a national sample survey is a good general- purpose measure of living standards (and thus poverty and inequality) for a variety of reasons (Grosh and Glewwe 2000). However, it does also have some limitations. Household surveys typically under-samplethe very poor and the very rich. In poverty analysis, this may not be critical if it does not result in anything more than a small underestimation of the poverty rate. When the interest is in equality or inequality of distribution, however, it matters more if both the tails (the poorest and richest) extremes are underrepresented, resulting in an underestimationofthe true level of inequality. Consumption is however only one dimension o f welfare, and national surveys only one tool. In the case of Cambodia, it's unlikely that the richest households or the slum dwellers o f Phnom Penh will be sampled in the household surveys. Thus, the results from householdsurveys must be interpretedwith this caveat in mind. 8 Sharinggrowth: equig and developmentin Cambodia Box 1.5 Commonly used inequality measures ..... There are many alternative ways in which inequality might be measured or communicated (McKay 2002). These include (amongothers): Average value (or shares oftotal) for the welfare measureheldby differentgroups; ratios ofthe welfare measure betweenthe top andbottomof a distribution; I a simple plot o fthe distributiono fthe welfare measure a Lorentz curve, which plots cumulative percentage o f the populationagainst cumulative percentageofthe wealth measure. The Gini coeflcient, derived from the Lorentz curve, is one of the most widely-used measures of inequality, describing the difference between an observed cumulative distributionand an (ideal) perfectly equal distribution. It ranges from zero to one, with zero signifying total equality and one denoting complete inequality. It provides a measure of disparity or inequality by assessing the relative shares of outcomes vis-a-vis the respective population shares in the distribution. For a Gini coefficient of 0, every . percentage of the populationhas 1 percentage share of the total outcomes (e.g., wealth or , income or wages). Inother words, the total outcomesare equally shared. I The Theil Index also ranges from from zero to one, with zero signifingtotal equality and ' one denoting complete inequality. One advantage o f the Theil Index is that it allows decomposition of total inequality. In other words, total inequality is the summation of inequality within each sub-component. Theil Index also measures inequality by comparing the relative outcome shares vis-a-vis the respective population shares, and weightingthe differentials by their respective(sub-component)incomeshares. I Each of these measures has strengthsand weaknesses. Thus the Giniregistersas a decrease in inequality any transfer between a richer and a less richperson, without makingany distinction as to whether the transfer went to someone in the middle or the bottom o f the wealth distribution, which is obviously of interest whenthe concern is poverty reduction. A different picture of inequality may be obtained if data on distributionis presentedas a ratio of values betweenthe top andbottom 5 percentofthe population,-or the top-andbottom 20.percent. - - - - - I _ - - There are also other aspects and indicators of materialwellbeing which are potentially more important in the longer run, and other tools for collecting and analyzing data. The measures commonly used to describe inequality (Box 1.5) can all be used with any continuous variable. The value of household assets or quality of housing may be good indicators of long-term welfare. Measures of human development (or human capital) such as literacy, level of educational attainment or health outcomes (e.g. life expectancy) have both intrinsic value in defining a good standard of living, and instrumentalrelevance in expanding an individual's economic opportunities. As the future living standards of Cambodian children depend on the education and healthcarethat their families currently enjoy, differences in access to basic services are a critical influence upon equality of opportunity for the next generation. Similarly, it is rewardingto examine inequalities in access to private productive assets (land) and publicly-provided infrastructure (such as roads or irrigation), as these expand opportunities for productionand employment. Given the importance o f these other outcomes, the report also examines the distribution of these non-monetary aspects of wellbeing and opportunity. In particular, Chapters 4, 5 and 6 will, respectively,document and analyze the extent of inequality in: Recent economicand social developments 9 ... access to productiveassets (with a focus on land) access to infrastructure (with a focus on rural roads) access to services (with a focus on schoolingand healthcare) Apart from the SESC and CSES, there are a number of other important sources for this analysis. The primary sources of information on intra-household differences in consumption and access to productive assets, services, and infrastructure are the 1993/4 SESC and the 1997 and 2004 CSES. The key source of information on health outcomes is the 2005 Cambodia Demographic and Health Survey (CDHS), the findings of which can be compared to those in the previous(2000) CDHS. Information on educationcan be derived from the consumption surveys, the CDHS, and a number of other national sources. Finally, information on popular values and perceptions with regard to living standards, nationalchallenges, and performanceof various political and social institutions were gauged with a nationally-representativepublic opinion poll of 2,565 adults. These various nationally-representative statistical sources are complemented with a variety of thematic, contextual studies focusing on particular localities. Such research, using either quantitative or qualitative methods (or both), may help us to understand relationships between different aspects of household livelihood strategies and welfare (e.g. between access to loans and investment decisions), and between households and wider institutional structures (communities, local authoritiesand market chains), in ways which are hard for a national sample survey. However, it is hardto then generalize from these local findings to make claims that they represent broader national phenomena (at least in the way that a well-designed sample survey can). Useful outcomes can be achieved by using both types of information in tandem to validate findings, explore behavioral explanations for observed statistical relationships, and so on. A key source is the Moving out of Poverty Study (Box 1,6), conducted by the Cambodia Development Resource Institute (CDRI 2007b). Other sources include the Participatory Poverty Assessment (PPA) conducted in the Provinces surrounding the Tonle Sap lake (CDRI 2007a); a number of sector-specific studies; and a short piece of fieldwork-based research into the current and potential future effects of systematic landtitling (Markussen2007). Trends in inequality of monetary dimensions of welfare As mentioned above, in those parts of the country for which direct comparison is possible, living standards rose for every consumption quintile. While every segment of the population has gained, the magnitude and pattern of growth differ between urban and rural domains. In Phnom Penh, the growth in consumption was spread fairly evenly across the whole range, averaging 30 percent over the decade between surveys. The poorest at the bottom end of the distribution experiencedthe highest rate of growth while the upper tail grew at the slowest rate. In "other urban" areas, consumption gains from the 25`h percentile through the 90thpercentile were phenomenal, averagin 40 percent during 1993-2004. Even for the poorest urban group below the 25 B percentile, consumption growth was 10-25 percent. Growth in other urban areas could thus be characterizedas rapid and widely shared (Figure 1.3). 10 Sharinggrowth: eguizj and development in Cambodia Box 1.6 Stratification or churning? Potential insights from longitudinal anaIysis The Movingout of Poverty Study (MOPS) has usedmixedmethodsresearch(that is, research which purposively combines quantitative and qualitative analysis) to describe and explain village- and household-level wealth and poverty dynamics. This has involved revisiting890 householdsinnine villages which were previously surveyedfor an earlier study in2001. This panel dataset (Le., a dataset that tracks the same households between repeated surveys) provides a valuable complement to national sample surveys such as CSES. In particular, it yields insights into how many households have moved out of poverty over a three year period-and, conversely, how many households which were not poor in 2001 had fallen into poverty by 2004, and how many had not seen their status (poor or non-poor) change. With this information,it becomes possible to estimate what proportionof the poor at any one time are trapped in chronic poverty, relative to the transitory poor who may cycle in and out of poverty in responseto life cycle events or householdshocks. Inprinciple,the level of stasis or dynamismgives clues as to the level o f equality of opportunity in rural society: if the rich stay rich and the poor stay poor over several rounds, it strongly suggests the existence of poverty traps and inequality of opportunity. If by contrast there is found to be considerable "churning" around the poverty line, that might suggest a healthier situation in which there is more equality of opportunity, with poverty status at any given date as likely to representluck, life cycle factors or choices (e.g. to defer current consumption or savings). The qualitative component of MOPS, meanwhile, used semi-structured interviews with groups and individuals in the nine villages to obtain their views on how their livelihoodshave changed, and what factors, at the household, village or national levels, help to explain these changes. Preliminary analysis and findings from the draft MOPS report are reproduced in this document:the full researchprojectwill be presentedas a CDFU WorkingPapers later in2007. , By contrast, growth in rural areas accrued primarily to the richest end of the distribution. The growth incidence curve here increased almost monotonically: consumption gains of the bottom half of the rural population were less than 10 percent, and reached only 20 percent at the 90thpercentile. One can characterize growth in rural areas as highly concentratedat the extreme top end. As a result, the Gini coefficient for per capita consumption in the geographically comparable sampling frame rose from 0.35 in 1994 to 0.40 in 2004. This rise in inequality is principally a result of rising rural inequality. Inequality in Phnom Penh and in other urban centers remained largely unchanged while inequality amongst the rural population increased considerably. Even though inequality in rural areas rose over time, the absolute level o f inequality continuedto be higher in urban areas today,just as decade ago. The urbanGini coefficient remainedat 0.43 between 1993 and 2004, while the rural Gini coefficient was only 0.354 in 2004, despite a rapid rise (Table 1.1). Because analysisof trends in consumption inequality is based upon a geographically- incomplete subsample of the 2004 data which is a subsample weighted towards the more accessible parts of the country, the 2004 figure derived from this trend analysis somewhat underplaysthe current level of inequality in Cambodia. Inthe whole of the country-that is, including now the poorer, peripheral areas that were not covered in the first 1993/4 survey-the Gini in 2004 stoodat 0.42. Recent economic and social develobments 13 Figure 1.3 Changes in consumption, percentile-to-percentile, for Phnom Penh, Other Urban, and Rural Areas, between 1994-2004 6o 1 Growth incidence curves, by sector or domain, 1993-2004 -- I J-*\ /'\Phnom Penh /.p/\/''-.. LJ,; i I Otherurban ,*I \$ \% / I \ fir"* I i (, \.*JWl..\ e'1 l .-.. )v --."- v q ; I/- '.*.* I/ / e .._.-- ..- C . I V __...-- . * 'Bk 3 ~ ,* * Rural o ] :** , 0 10 20 30 40 50 60 70 80 90 100 Percentile of the population (ranked by per capita household consumption) Source. SESC 199314; CSES 2004. Note. only comparable sub-samples from an identical (1993/4) sampling frame are used for each survey year. Table 1.1 According to household surveys, inequality (Gini coefficients of real per capita consumption) rose considerably in the countryside, but remained largely unchanged in the towns 1993 2004 Rural 0.265 0.354 (+ 0.005 (*0.007) Urban 0.434 0.431 (+ 0.010) (* 0,009) Phnom Penh 0.393 0.367 ('t0.023) (5 0.014) Other Urban 0.439 0.434 (* 0.037) (* 0.018) Cambodia 0,347 0.403 (* 0.006) ('t0.005) Source: Socio-economic Survey Cambodia, 1993; and Cambodia Socio-economic Survey 2003-04. Note: estimated standard errors in parentheses; real consumption in 1993 Phnom Penh prices. The comparable sub-sample of 2004 is taken from the full 15rnonfh sample. The results are based on similar sampling frames. While Gini coefficients allow us to easily compare inequality across groups, they do not allow us to readily compare the relative contribution of each group to the total inequality. The Theil index, another commonly used measure, provides ease of 12 Sharinggrowth: egtri9 and developmentin Cambodia decomposition and can reveal the relative importance of each component-in this case the contributions of (i) rural inequality, (ii) urban inequality, and (iii) between-urban- rural inequality-to Cambodia's total inequality. Figure 1.4 shows the decompositiono f Cambodia's inequality into these three components. Rural inequality has risen and its share in total inequality has also increased during this decade. Urban inequality, on the other hand, has fallen in point estimate and in its contribution to total inequality. Between-sectorinequality is insignificant and unchanging. Figure 1.4 Over the last decade, rural inequality has risen sharply o Inequality between rura -.-$0.3 - and urban areas Y5, -P0.2 - Inequality within urban areas 0.1 Inequality within rural areas O t 1994 2004 Source: SESC 199314;CSES 2004. Note: For 2004, we only select the comparablesub-sample,based on the full 15-monthsample, taken from an identical(1993) samplingframe. The between-urban-ruralgap is small because the rural and urban distributions have substantial overlapping areas. Figure 1.6 presents distributions of consumption for urban, rural and Phnom Penh populations by year. The overlap between these three domains is large: hence, even if the means are significantly different across distributions, disparity between distributions is trivial. Recent economicand social developments 13 Very few people feel that inequality has not changed: the vast majority thinks that the gap between the richest and poorest in their community has changed over the last five years, with 59 percent feeling that it has increased. Opinion is more divided on whether local wealth differences will diverge further in the next five years: one in five think not, while those forecastingthe gap widening outnumberthose who predict it would shrink. Figure 1.5 Most people believe inequality has risen; and will continue to do so a. Compared to five years ago, in your b. Infive years' time, in your communitythe community the gap between richest and gap between the richest and the poorestwill poorest is: be: 48% 43% Wch Smaller The Greater Wch Smaller The same Greater smaller same greater Source IRL 2007 p. 31 Note Answers total to 100 percent: missing values answered "don't know I didn't answer' Trends in inequality of non-monetary dimensions of welfare There has been progress with regard to other aspects of inequality. In education, there have been very positive trends towards increasing equality in educational uptake and outcomes, with closing gaps in the enrolment rates of girls and boys and upper and lower quintile groups. In health, there are some positive trends in outcomes. However, the pattern o f public provision can be made more progressive. Currently, much is spent in the capital, whereas some 85 percent of the population and 90 percent of the poor live in rural areas, and with significant amounts allocatedto tertiary rather than primary care. 14 Sharinggrowth: equip and developmentin Cambodia Figure 1.6 Distributions of consumption, by sector, 1993, 1997 and 2004 Kernel density ofper capita household consumption (1ogarithmic scale) Cambodia, 1993 10 Lop o fpir capita hous8eholdconsmgption (weighted bypopul&:on) Phnorn Penh, 1993 Rural.1993 Kernel density of per capita household consumption (logarithmic scale) Cambodia, 2004 i og ci 0 ' l o Log ofpzr capita hous8ehold consu2ption (weighted by popu1:t:tfon) _ _ _ Phnom Penh,2004 Other urban.2004 Rura1.2004 Kernel density of per capita household consumption (logarithmic scale) Cambodge, 1993/2004 -1 I . 6 Logo f per capita household consumption (weighted by population) 7 8 9 10 1! _---_ _ _ _ _ Phnom Penh,1993 ---Otherurban,l993 Rural, 1993 Phnom Penh,2004 Other urban,2004 --------- Rura1,2004 Source Socio-economic Survey Cambodia, 1993, and Cambodia Socio-economic Survey 2003-04. Note only Comparable sub-samples from the same (1993) sampling frame are used Recent economic and social developments 15 Equality in access to primary education is improving As a result of expanding coverage, equality in access to education is improving as increasing numbers of girls and children of rural and/or poor families are drawn into the school system. Although there are Table 1.2 There has been a dramatic expansion in still large inequalities basic education in recent years in outcomes between men and women, rich Performance indicator 2000 2004 change and poor, and rural Pre-schools 952 1,275 34% and urban Lower secondary schools 511 688 35% populations, these Classes 1,628 2,249 38% group averages mask Primary enrolment (millions) 2.4 2.7 14% considerable progress Primary enrolment in remote areas 54,000 81,000 50% in recent years. Primary net enrolment rate (NER) 84% 90% 7yo Despite problems Lower secondary enrolments 284,000 460,000 62% with dropout, Source.Sakellariou 2007 citing Quinio 2005. repetition and teaching quality, greater equality in access to primary education is starting to result in closing gaps in education outcomes. Amongst the younger age groups which have more recently passed through the (improving) educationsystem, literacy rates are higher across the board, and the inequalities are shrinking rapidly (Figure 1.7 and Table 1.2). Although pronounced, health inequalities are starting to close In most of the key measures, health outcomes are becoming less pronounced over time. Thus for example, the gap between the best and worst performing provinces in terms o f child survival has narrowedbetween 2000 and 2005: Mondolkiriand Rattanakiri posted 40 percent and 39 percent declines in infant and child mortality respectively,while indicators held steady in Phnom Penh. Pursat province experienced the second highest rates of early childhood mortality across all age groups in 2000: the CDHS 2005 has shown although still relatively high, there has been a marked reduction by more than a third for rates of infant and under-fivemortalit?. This picture of narrowing inequalities is not, however, universal. Inequalities in child survival between the rural and urban populations are widening as improvements occur in both, but occur more rapidly-and starting from a much better initial level-in the urban centers (see Figure 1.8 ). * Changes in methodology between CDHS 2000 and CDHS 2005 saw Siem Reap included with Oddar Meanchey, Preah ViheadStung Treng was included with Kratie and Kompong Som was includedwith Kampot/ Kep for the CDHS 2000 which may give an under-estimate o f 2005 ratios. 16 Sharinggrowth: equity and developmentin Cambodia Figure 1.7 Increasing equality in primary enrolment is closing gaps in literacy rates amongst the younger generation 100 - 91 a9 e a4 a4 + a3 a3 ao .c 7 5 - a3 e 74 77 75 u e 70 a, e 67 e 66 0) ;503 e 60 e 60 62 + 60 L e 57 55 v 0 0 45 E 39 -.?e i 36 h 2 5 - 2a 0 4 15-19 20-24 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-64 65-69 70+ age (years) *-a7 -*a7 X *,76 0 0 71 + 8 6 8 x a, 71 ' + 62 62 0 @ B 61 "'\,+ 59 c e + + 5 5 0 - . e + & 52 57 e 53 53 4 49 147 ' ca, E 46 49 Richest 0 -r i 41 42 h 0 2 5 - 28 6 25 Next poorest 24 22 o Poorest 0 7 Recent economic andsocial develobments 17 ~~ Figure 1.8 Faster gains from a better base in urban areas are widening existing inequalities A: Infant mortality rate B: Under-five (under one year) mortality rate 150 150 146 100 100 50 50 0 0 2000 2005 2000 2005 0ratio:ruralchildrendyingforevery 100urban children dying -o- Ruraldeath rate (deaths per 1,000 live births) &Urban death rate (deaths per 1,000 Source:CDHS 2000 p. 125;CDHS 2005 p 126 Perspective and focus In the context of Cambodian development policy, the focus is for now on the bottom end of the distribution. The burning question on inequality should be why this poorest segment o f the population-chiefly in rural areas-has been unable to benefit more from productive opportunities. This report thus focuses on addressing the barriers to participation in the economy by all members of society, and how the government could reduce these barriersto generate greater productivity and real income for all. . Based on the analyses, the rest of the report examines sectoral as well as cross-cutting issues in-depth. Recommendations for public policiesrelate mainly to: .. creatingan environmentthat promotes individuals' investmentsin capabilities; expandingopportunitiesto access services and productive assets; and removing systemic barriers so the poor and disadvantaged can participate in the country's developmentand experience higher growth. More specifically, we suggest policy options to facilitate or ensure greater inclusion o f the poor in governmentprograms(e.g., land titling, social land concessions); accessing public services (e.g., healthcare, schooling); and sharing public resources (e.g., infrastructure,fiscal resources, common property resources). Each o f these measures is a strongcandidate for inclusion in a set of equity-enhancingpro-poor interventions. 2. Rural inequality I _ Summary Inequality in living standards (per capita consumption) in rural areas rose between 1993 and 1997 but stabilized between 1997 and 2004. This trend suggests that has been nothing structurallydestabilizingin the patternof Cambodia's rural growthto date. Rural inequality has been driven by inequality among the richest quarter of the population, and inequality between the richest andthe poorest. Inequalityamong the rural majority (the bottom three-quarters of the population) is negligible. Over time, the rich-poor gap (the disparity between the bottom three-quarters and the top quarter) has narrowed as poverty has fallen. Much of the increase in rural inequality is, therefore, due to increaseddisparities amongst the richest quarter of rural Cambodians. Geographic isolationcontributes to rural inequality, as less connected localitieshave lower average levels of income and consumption and have experienced lower rates of growth. Remote villages tend to have higher poverty incidencethan better connectedvillages. Thus, the geographic expansion of public investments in infrastructuresuch as rural roads will be instrumental in improving connectivity and expanding opportunities for remote communities andreducingpoverty and inequality. Household and individual factors underlie growing inequality in rural Cambodia. Differentials in household-level human and physical capital result in disparities in returns and income. Differences in intrinsic capacities and skills naturally result in varying ability to participate in and benefit from expanding economic opportunities; the ability to which innate abilities can be realized is in turn significantly influenced the availability of productivity-enhancinginfrastructure (as well as by access to education and healthcare). Thus, poverty reduction and equity goals will be served by public investments in productivity-enhancinginfrastructuresuch as irrigation; public spendingon basic services to enhance human capabilities; and policies to improve land tenure, and so encourage investments in land. In particular,infrastructuresuch as irrigation is crucial for the improving and stabilizingthe livelihoods of rural households. While a small number of farmers can afford to invest in small-scale irrigation structures, the majority depends on the Government to provide these resources. Investments in irrigation are needed to raise and stabilize agricultural productivityand diversify crop production. This chapter examines why the rural majority did not experience more rapid gains between 1994 and 2004, and how policiescouldaddress this imbalance of gains between the rich and the rest. The first section examines what took place during the past decade by describing trends in inequality of consumption in the sub-periods of 1993-1997 and 1997-2004. The second section examines within- and between-group changes in inequality to identify which groups contributed to rising inequality in rural areas. The third section answers why inequalityrose inthe ways it did. 19 Sharinggrowth: egu$ and development in Cambodia What took place? A long decade, divided into two sub-periods A decade is a long time during which structural economic changes are likely to take place. This is especiallytrue for a country like Cambodiathat went through simultaneous transitions from war to peace, from one-party to multiple-party politics, and from a closed and planned regime to an open market economy. Running through this period there has also been an important fourth transition in demographics,as the children born in the post-1979 and post-1989 baby booms have worked their way through the schooling system and into the labor force at the same time that the fertility rate has declined markedly. Given these numerousstrands of rapid structural change, it helpsto divide this past decade into two sub-periods for a closer look at the changing pattern of economic activity and income distribution. Making the best of deficient data The Cambodian Government implemented four rounds of household surveys during the 1990s. Unfortunately, each has a different set of deficiencies. The CambodiaPoverty Assessment (World Bank, 2006, Box 2.1) documents the shortcomingsof these household surveys. Despitethe many flaws ofthe 1997 CambodiaSocio-Economic Survey (CSES), it is clearly superior to CSES 1996 and CSES 1999. We proceed on the working assumption that the downward biases in consumption strongly suspected to exist in the CSES 1997 data (which was collected during a period of political tension) would have affected the entire distribution equally, without being more or less exaggerated in different parts of the distribution: on these grounds, analysis of shifts in the distribution over time will still be robust. It can also be safely assumed that non-consumption measures (e.g. access to infrastructure, services, and public amenities) will be less sensitiveto the tensionsthat existed in June 1997, and comparingthese measures in 1997 with those in 2004 providesa reliable and robust measure of trends. Thus, imperfect data from CSES 1997 is usedjudiciously to complementthe 1993 and 2004 data. Analysis also suffers from some severe limitations of the Socio-economic Survey of Cambodia (SESC) of 199314 (henceforth referred to simply as SESC 1993 for simplicity). The entire SESC 1993 dataset contains only limited pieces of useful information (variables): many of the household-levelvariables recorded in later surveys are not recorded here, others were collected but have since been lost, and there was no village questionnaire module, and thus no information on local amenities of the villages in which the householdsare located (e.g. whether there was a school or a healthfacility in the village). Thus, while the SESC 1993 will be utilized to its maximal potential, analysis is constrainedin terms of depth, breadth,and rigor for the year of 1993. Interestingly, when it comes to dissecting overall change between 1993 and 2004, contextual mixed-methods research suggests a slightly different storyline from that suggested by analysis of the national sample survey data. According to respondents in the nine MOPS case study villages, intra-village differences in livelihoods and living standards began to grow from 1993, but accelerated rapidly after the 1998 elections-the reverse of the periodization outlined above (CDRI 2007b, forthcoming, chapter 2). Possiblereasons for these different perspectiveswill be outlined below. Rural inegtlalip 20 Rural inequalityjumped between 1993 and 1997, then leveled off between 1997 and 2004 Table 2.1 presents the Gini coefficients for real consumption in years 1993, 1997 and 2004. Both per capita and total household consumption measures are shown as household structure changed dramatically during this decade, as average household size fell by about 10 percent, especially for households at the bottom end of the distribution. The inequality indices in Table 2.1 indicate a substantial increase in inequality in rural consumption during the four years between 1993 and 1997. The Gini coefficient for real per capita consumption leaped significantly from 0.27 in 1993 to 0.33 in 1997, and then edged upward modestly to 0.35 over the next seven years. The trend for inequality of total household consumptionwas similar, with a somewhat smallerjump in the first four years. Table 2.1 Welfare differences in the countryside appear to have widened rapidly between 1993 and 1997, but more slowly after that Gini coefficients of real consumptionfor comparable samples of rural-areas, 1993, 1997 and 2004 1993 1997 2004 per capita real consumption 0.265 0.330 0.354 (*,005) (1.007) (i.007) total household real 0.281 0.338 0.367 consumption (*.006) (1.008) (1,007) Source SESC 1993/4, CSES 1997, CSES 2004. Note Only comparable samples from the identical (1993) sampling frame are used for each survey. More insights can be obtained from growth incidence curves which show the percent change in real per capita consumption between two points in time for each percentile of the rural consumption distribution. Again, the decade is split into 1993-1997and 1997- 2004. Figure 2.1 presents three growth incidencecurves for (i)the decade of 1993-2004; (ii) the sub-periodof 1993-1997;and (iii)the sub-periodof 1997-2004,respectively. The top panel reveals two features between 1993 and 2004 that contributed to rising inequality: Consumption growth amongst rural households increased steadily by distribution percentile, i.e. the richer the households, the bigger their proportionate consumptiongains. Except for the richest 7 percent, the entire rural population experienced only sluggish growth in consumption. Even households at the 75thpercentile only gained 20 percent while those at the top of the poor group, at the 93rdpercentile, grew by a mere 25 percent. By contrast, at the richest end of the distribution, growth was 40 percent. The CSES data suggests that rising rural inequality between 1993 and 2004 masks two very different sub-periods. The consumption changes for each percentileduring the earlier sub-period of 1993-1997 are plotted in the growth incidence curve in the middle panelof Figure2.1 while those for 1997-2004are representedin the bottom panel. 21 Sharinggrowth: epi9 and developmentin Cambodia Figure 2.1 Initially, the very rich made rapid gains while others lagged; later, improvements were spread evenly between wealth groups - Changes in real consumption, percentile-to-percentile Bk //----- ----- ~ - // Growth incidence curve Rural, 1993-1997 (in 1993 sampling frame) i 0 25 50 75 1w Percentile o f the population (ranked by per capita household consumption) ch-owthincidence curve Rural, 1997-2004 (in 1993 sampling frame) E - * - -..-. - .. Growth in median ~ ..- . c i ; b t - vm ~~ R - - . ,. 7----- ~~ ~ Source SESC 199314, CSES 1997, CSES 2004. Note: Only comparable samples from the identical (1993) sampling frame are used for each survey. Rtlral inegualip 22 1993-1997: the very rich pulled ahead, while others fell behind Between 1993 and 1997, only a handful of the population (the top three percent) enjoyed rapid improvement in living standards (growth in real per capita consumption exceeding 20 percent). The period 1993-1997 can be characterizedas one in which the extremely rich pulled ahead while the poorest, the poor, and the middle of the wealth distribution stagnated. The result was a substantial, statistically significant rise in inequality between 1993 and 1997. 1997-2004: widely- shared growth The changes recorded in the second sub-period are in sharp contrast with the first. Between 1997 and 2004, almost every percentile in the distribution experienced similar high rates of growth in per capita consumption(between 20 and 28 percent). The growth incidence curve is practically flat at the mean growth rate across the entire distribution. The exceptions are among the upper middle class segment (70th-90thpercentiles)and the poorest (1S'-15thpercentiles) with slightly smaller growth, around 22-24 percent. One can characterizethis period as equitable, with growth broadly shared. Thus, the combination of different trends in 1993-97(growth concentratedamong the rich) and 1997-2004 (broadly-sharedgrowth) has resulted in rising rural inequality. The trend during 1997-2004 is promising, and suggests that there is not necessarily a structural problem with Cambodia's economic growth. Projecting simply on the basis of trends in this period, growth dynamics do not suggest a structuraltendency for growth to increase inequality. Which groups drove inequality? The analysis in the previous section hinted that a comparison of within-segment inequality for various income (consumption) segments might shed further light. Using real per capita consumption, rural householdsare ranked into four quartile groups. Divergence amongst the rich promoted inequality Table 2.2 presents Gini coefficients for each o f the quartile groups in years 1993, 1997 and 2004. Levels of inequality within the two middle quartiles of the distribution were strikingly low throughout these 11 years, implying that the majority of rural households enjoyed similar levels of per capita consumption clustered around the average. However, inequality in the richest quartile was the highest in all three survey years, with the Gini coefficient for this quartile rising steadily (from 0.18 in 1993, to 0.23 in 1997, to 0.28 in 2004) betweenthe three surveys. At the other end ofthe spectrum, there was some modest inequality within the poorest quartile group, at a level that remainedessentiallyunchangedthroughout the decade. As a result, inequality in the richest quartile group in 1993 was double the inequality in the poorestquartile group; but becametriple the level ofthe latter.by 2004. 23 Sharinggrowth: equig and development in Cambodia Table 2.2 There has been increasing divergence in wealth within the richest 25 percent of rural households Gini coefficients of real per capita consumptionby quartile groups, 1993, 1997 and 2004 Quartile of rural households In comparable sub-samples 1993 1997 2004 Poorest 0.094 0.103 0.114 Next poorest 0.044 0.050 0.056 Next richest 0.050 0.065 0.063 Richest 0.179 0.232 0.277 Total 0.265 0.330 0.354 Source: SESC 199314,CSES 1997, CSES 2004. Note; Only comparable samples from the identical(1993) sampling frame are used for each survey. Rising rural inequality is a result of rich-poor disparity and rising inequality among the richest quarter of the population While Gini coefficients allow comparison of inequality across groups, they do not readily allow comparison of the relative contribution of each group to total inequality. The Theil index, another commonly used measure of inequality, provides ease of decomposition and can easily identify the relative importance of each component (in this case the contribution of each quartile group to total rural inequality). Table 2.3 presents a decomposition of inequality, using the Theil Index. Total rural inequality is decomposed into within-quartile inequality for each of the four quartiles and (between) inequality across quartiles.Three main findings emerge: Within-quartile inequality was insignificant among the bottom three quartile groups . (i-e., between the lst percentile and the 75thpercentile): their joint contributions to total rural inequality is essentiallyzero. Inequality within the richest quartile group became increasinglysignificant over time. . Its contribution to total rural inequality increased from 24 percent in 1993 to 40 percent in 2004. Between-quartile inequality remained the largest component, even as its share has been eroded by inequality in the richestquartile. These results, as summarized in Figure 2.2, suggest that the gap between rich and poor (i-e., between-quartile inequality) is one of the two determinants of rural inequality. The other determinant of rural inequality is the inequality within the richest quartile group. This within-richest-quartile inequality is rising, and contributing much more weight, or importance, to total rural inequality over time. By contrast, the rich-poor gap is closing, albeit very gradually, as a result of economic growth that has benefited the rural majority, in particular between 1997 and 2004. This suggests that pulling the bottom end of the distribution out of poverty will also directly address inequality. That is, a direct way to reduce inequality will be to understandand remove the impedimentsto growth among the poorest. Generating absolute increases in income for the rural majority (i.e., the bottom 50-70 percentiles) will close the income gap, and lower between-quartileinequality. Rural inegualig 24 Table 2.3 Theil Indices: a decomposition of within-quartile and between-quartile inequality of per capita consumption, 1993, 1997 and 2004 Quartlle of rural households in comparable sub-samples 1993 1997 2004 Poorest 0.002 0.002 0.003 Next poorest 0.001 0.001 0.001 Next richest 0.001 0.001 0.001 Richest 0.031 0.054 0.103 Between Quartiles 0.095 0.147 0.151 Total Theil Index 0.129 0.205 0.259 Inequality in the Richest Quartile Group 4 as a % of Total Rural Inequality 24 26 40 Between Quartile Inequality as a % of Total Rural Inequality 76 74 60 Source: SESC 1993/4,CSES 1997, CSES 2004. Note Only comparable samples from the identical (1993) sampling frame are used for each survey. Figure 2.2 A decomposition of rural Theil Indices 0.25 - 3 0.2 - - `El 0 151 0.15 - 0 0% quartile I 0 103 0 054 nchest quartilk I I 1993 1997 2004 Source: SESC 1993/4, CSES 1997, CSES 2004. Note: Only comparable samples from the identical (1993) sampling frame are used for each survey. Differences in growth rates and inequality There are clear geographical components to patterns of inequality of opportunity and outcomes within the countryside. These spatial inequalities can be picked out at a variety of scales. In terms of outcomes there are, for example, very large inter-Provincial variations in the incidence and severity of poverty (Figure 2.3). These Provincial averages may themselves conceal considerable spatial variation at a more fine-grained resolution: with severely underdeveloped rural roads network, communities next to a national road can enjoy dramatically better opportunities, and as a result demonstrate significantly betteroutcomes, than communities locateda mere 5 kilometers away. The MOPS research found that over the three years between surveys, inequality in income had grown both between the nine study villages; and between rich and poor households within most of the villages. In 2001, average incomes in the richest village were 2.4 times those in the poorest village: by 2004/5, this had grown to 2.9. 25 Sharinggrowth: eqzrig and development in Cambodia Figure 2.3 Poverty rates in upland areas are significantly higher than on the coast Source: CSES 2004 analyzed in Knowles 2005a. Within the MOPS villages, meanwhile, groups with different mobility status saw their incomes diverge further. The chronically poor (those who remainedpoor in both rounds of the panel) saw their incomes rise by 33 percent; the comfortably rich (those who remainednon-poor in both rounds) experienceddoublethat rate (61 percent). As a result, the ratio between the incomes of the two groups changed from 1:3.2 to 1:3.9. This widening gap is reflected in the income Gini, which increased by around 0.05 in most villages, with more dramatic jumps in inequality in two sites and small declines in inequality in another two sites (Figure 2.4). Richer provinces have higher inequality Ifwithin-quartile inequality in the richestquartile group is one ofthe main causes of inequality, we would expect to see that Provinces with greater population shares in the richest quartile group experienced higher intra-provincial inequality. Figure 2.5 demonstrates the relationships between provincial Gini coefficients and rural population shares in the richest quartile group'. The top panel (a) of Figure 2.5 presents the level- by-level relationshipbetween inequality and populationshares in the richestquartile. By pooling all observations in the survey years of 1993, 1997 and 2004, it is possibleto plot provincial inequality levels on the y-axis and provincial population shares in the richest quartile group on the x-axis. The plot shows a (statistically) significantly positive relationship. Richer provinceswith higher percentages of their populations belongingto the rural richest quartile group tended to experience higher inequality. ' All analyses in this section are carried out on only the sample o f rural households. Note that only compatible samples with identical sampling frame o f 1993 are selected from each survey to be analyzed. b r a / ineqtlalzp 21; Figure2.4 In the MOPSdataset, most villages recorded an increasein the income Gini averaging 0.05; two registered much higher increase Krasaing (Battanbang) Babaong ( R ~ Veng) Y Andong Trach 0.33 (Battanbang) .31 }-0.02 Kanchor (Kratie) 045 } t o o 9 Trapeang Rey (KompongSpeu) }-0.03 Rek Khmeng } t0.05 (Kandal) Konpong Thnaot }-0.04 Wanpot) Dang Kdar }+0.04 (KompongThorn) Khsach Chiros }t0.05 (Konpong Thorn) inequality of per capita income (Gini coefficient, 0 = perfect equality, 1 = perfect inequality) 0 200415 Source: CDRl 2007b (forthcoming) Provinces that grew much richer have faster growth in inequality The bottom panel (b) o f Figure 2.5 presents the difference-by-differencerelationship of provincial inequality and population in the rural richest quartile. By pooling the two sub-periods of 1993-1997 and 1997-2004, we plot percentage point differences o f provincial Gini coefficients on the y-axis and percentage point differences o f provincial population shares in the rural richest quartile on the x-axis. The results also indicate a (statistically) significantly positive relationship between changes in inequality and changes inthe proportionof rich people. These results suggest that provincesthat grew richer or more rapidly as more of their residents becoming rich (i.e., greater movement into rural richest quartile) tended to experience rising inequality. Equally striking is the finding that provincesthat regressed or observed falling shares of their populations falling in the rural richest quartile tended to experience stable or declining inequality. 27 Sharinggrowth: equity and development in Cambodia Figure 2.5 The relationships of provincial Gini Coefficients and the population shares in the richest quartile group (rural areas), 1993-2004 a. Levels: Gini coefficientsvs. populationshares in richestquartile groups I; - 045 - * ; 'Z 0 4 - * * *e * V .-z03053: -025- * = 0 2 - - 0 1 5 - * .- b. Differences: changes in Gini coefficientsvs. changes in populationshares in richestquartile 0.4 .c: 1 01.* 0 " ja f I percentage point differences-0.3provincial rural population shares in 1 in ~ rural richest quartile groups Source: SESC 1993; CSES 1997; CSES 2004. Note: Only comparable samples from the identical (1993) sampling frame are used for each survey. The analysisand findings inthis sectionsuppliesthree messages: . Rising rural inequality was due to rising inequality within the richest quartile group. Rising provincial inequality coincided with increasing shares of the richest (provincial population). Rising rural inequality coincidedwith faster growth. The first message is a description of basic facts. The second message is a result from time-series analysisthat shows provinceswith increasingproportionso f rich residentsare more likely to experience rising inequality. A note to highlight is that there is no evidence of concentrated wealth in any particular provinces and the fluctuations of provincial inequality appear random. Thus, the speed and magnitude with which the RNral inequality 28 population in any province makes it into the rural richest quartile are very likely to be a result of economic growth. The third message is a direct implication from the data that provinces with higher inequality are those with greater shares of rich people; and provincesthat have more rapidly rising inequality have also grown muchricher. Drivers of inequality in the rural population Current patterns of inequality within the Cambodian countryside reflect a complex interactionamong many processes. These processes operate at the individual, household, community and national level; and came into play-and in some cases, have faded out again-at different stages over the last fifteen years of transition. The remainder of this chapter will provide analyses of some underlying factors of rural inequality. History and geography Equality in initial conditions? The state that replaced the Khmer Rouge, the People's Republic of Kampuchea (PRK), was premised upon socialist ideals of equality. In the early years, as the PRK sought to prevent famine following the collapse of the Khmer Rouge regime, low-level collectivism was mandated, with land and scarce productive assets such as draft animals shared in krom samaki (solidarity groups). Within a few years, however, there was increasing popular pressure for a return to individual family farming, and the krom samaki were gradually abandoned (Frings 1993). Research in the late 1990s (Box 2.1) suggests that at this stage-and perhaps even earlier-small but significant differences began to emerge in householdasset endowmentsand production opportunities. I Box 2. 1 Initial conditions and the re-emergence of difference Overall, the Khmer Rougerevolutionsucceeded in deliberately and comprehensively breaking the pattern of wealth which had existed in the 1960s, to the advantage of some and the disadvantageof others. (In a few cases people were able to reclaim wealth that they had held before the revolutionary cataclysm: some families managedto dig up gold, jewellery or other wealth that they hadburiedbeforethe Khmer Rougetook over.) Nonetheless, devastatedas Cambodia was when the new regime took PhnomPenh in January 1979 and set about re-establishingthe state, the situationof "equality in poverty" was subject to some small but important variations at the local level. Villagers in Kompong Chhnang villagers related how those who were able to gather and transport cattle, buffalo, carts, tools and the like back from the hills when the Khmer Rouge cadre fled, effectively gained ownership of these assets. Many respondentsstressedthat single women and the seriously ill were disadvantaged at this stage, lacking the physical strength or power (komluhng, a term with inherently masculine connotations: Ledgerwood 1990 pp. 23-4) required to claim and move such goods. Although these assets were then collectivizedand came to belong to the born samaki as the PRK establishedcontrol, with defacto decollectivizationin 1982 or 1983 they revertedto those who had"owned" them in 1979. In 1982, such productiveassets were in short supply andthose families who didhavethem hada headstart on productiveuse ofthe landthey received. Source: Conway 1999 pp. 207-8. 29 Sharinggrowth: egtrip and development in Cambodia The lack of private markets for products or imports, reinforced by Cambodia's political and economic isolation, meant that the economy before 1990 was largely subsistence-oriented, with very few opportunities for individuals to exploit their differential sets of endowmentsor to accumulatewealth. The equality that existed before the Paris Peace Agreement was somewhat less complete in practicethan principle, which meant that when market relations were reintroduced form 1989 some households did enjoy a head start, even if only a small one. To some extent the difference that has emerged since the early 1990s reflect these different starting positions. For example, favoritism in the allocation of land provides a second caveat to the picture of decollectivisation as an egalitarian process in which distribution of resources reflected consumption needs at that time. On a number of occasions, relatives of village and communeauthorities receivedmore than their fair share (Frings 1994). "Location and legacy" explain the geography of inequality The complex set of interacting environmental, economic and political processes have thus created and sustained expanded opportunities in some localities relative to others. Households living in settlementsclose to towns, roads and markets had an early lead on households living in remote areas. These advantages took the form of better access to social services (preventative and curative health services, basic education) that might enable the accumulation of human capital; dramatically lower time and cost to move goods to market; and much greater probability of benefiting from a donor, NGO or Government development scheme. Lowland settlements with national roads running through them tend to be located where they are in the first place because of favorable production conditions (better soils or, more typically, better water control conditions); communities away from roads are almost by definition generally upland, poor quality, and difficult or impossible to irrigate. The final aspect to geographical "legacy" is the date at which armed political conflict ended in the locality: this effectively defines the point at which economic activity and accumulation could start. Those parts of the country which came out of conflict late-in MOPS, several of the sites were vulnerable to Government-Khmer Rouge clashes right up until 1998-lag those regions which have had peace since 1993 (Box 2.2). Well-connected provinces enjoyed higher levels and faster growth of consumption Using the CSES data, this section2examines the impact of isolation and remoteness on inequality or the rich-poor gap. Figure 2.6 shows the levels and growth in mean daily real consumption for all provinces (in comparable samples, based on an identical 1993 sampling frame) during 1993-2004. The top Panel (A) reveals that mean consumption levels in Battambang,connectedto fast-growing Thailand, were more than twice those in a poor, upland, central province such as Kompong Spew. The growth rates of consumption for provinces connected to the world markets (e.g. Kandal surrounding Phnom Penh; or Kampot on the coast) were also higher than inland provinces, 2 All analysis in this section is carried out on only the sample o f rural households, and only betweencompatiblesamples within the identical(1993) sampling frame. h r a l ineauality 30 Box 2.2 Peace has opened up opportunities for investment, trade and accumulation In the MOPS research project, participants consistently emphasized how the end to armed politicalconflict liftedthe burdenof fear andopenedup livelihoodopportunities. (KsachChiros, KompongThom) Before Pol Pot troops defected to the Government, most villagers dared not sleep in their houses; most of the time they slept on the ground close to a safety hole. Whenever Khmer Rouge or government troops came to the village, they took our chickens and ducks or burned our houses ifwe were not willing to give our belongings to them. In 1993, we could not movefieely because we had to askfor a permitfrom local authorities. By contrast, in 1998 we got morefreedom, because we could go everywhere and our community was secure after the integration of the Khmer Rouge. (DangKda, KompongThom) Before the Khmer Rouge integration in 1998, this community was unsafe because this area was a stronghold of the Khmer Rouge. At that time, the situation was not stable. People had to be well prepared and ready to escapefrom the bombing. Theyfelt unsafe, so never cared about building or improving their houses... That's why security always cameJirst to our mind and we had less time to think about our economic life. At that time I cared only about my pots and dishes rather than other useless things when each time serious fighting took place in this community. I never thought about earning money and how to improve myfamily's standard of living because I had no time to think about it. The only one thing I had to do was to learn how to escapefiom bombing. Kanchor, Kratie This village wasfree from armed conjlict after the defection of all Khmer Rouge troops in 1998. Having security signijkantly helps improvepeople 's livelihoods in this area. Source: CDRI2007b (forthcoming). particularly Tonle Sap provinces. Siem Reap is a curiosity, and highlights the importance of examining spatial variations at a fine-grainedlevel: although the provincial town is the center o f a multi-billiondollar tourism industry, the surrounding province, much o f which has poor soils, limited irrigation and poor roads, remains extremely poor. However, lagging provinces are catching up overtime The bottom Panel (B) plots growth during 1997-2004on the y-axis and growth during 1993-1997 on the x-axis. The plot suggests evidence of catching-up among provinces. In other words, provinceswhich grew less or grew more slowly during 1993-1997tended to grow more rapidly in the subsequent period o f 1997-2004. These results suggest not only that geographic isolation plays an importantrole, but that roads play a critical role in overcoming this isolation. As remote areas become more connected through additional road investments, an increasing proportion of the rural population can participate in the country's developmentand growth. During this process, some locationsor provinceswill grow sooner than others. Encouragingly, the data suggest that over time lagging provincesare catchingup. 31 Sharinggrowth: equity and development in Cambodia Figure 2.6 Levels and growth of provincial (rural) mean real consumption, 1993- 2004 Panel (A) Better connected provincesexperiencedgreater mean and higher growth in consumption Panel (B) Lagging provinces during 1993-1997 grew faster subsequently during 1997-2004 A A Source: Socio-economic Survey Cambodia, 1993; Cambodia Socio-economic Survey 1997; Cambodia Socio-economic Survey 2003-04. Note: Only comparable samples from the identical (1993) sampling frame are used for each survey. Welfare levels are lower in remote localities Differences in living standards can be compared by classifying householdsinto either remote or connectedvillages on the basis of location-relatedattributes. Two measures of remotenessor connectivity are used. The first defined by the existenceof any all-weather road in the village: villages with at least one all-weather road are classifiedas connected, and all other villages classifiedas remote. The second measure is defined by the distance to the district headquarters: villages further than 10 km are classifiedas remote. Table 2.4 shows the mean levels of consumption for the two types of households for 1997 and 2004. Partition (1) uses the all-weather roads measure, and partition (2) uses the district headquarters distance measure. We observe a consistently higher mean h r a l inequalip 32 consumption levels among connected villages in both years. Mean consumption in connected villages was about 10-13 percent higher, according to both measures of remoteness. The results indicate that geographical location contributes to rural inequality through the income gap or consumption disparity between richer connected and poorer remote localities. Table 2.4 Differences of mean consumption between connected and remote households in rural areas (compatible samples), 1997 and 2004 (11 (2) measured by existence of all- measuredheadquarter:to district by distance weather road in the village: with versus without less than versus more than 10 km away 1997 2004 1997 2004 Connected villages 1,651 2,380 1,764 2,470 Remote villages 1,930 2,130 1,945 2,180 Source based on comparable sub-samples from the 1993 sampling frame, CSES 1997, and CSES 2004 Note motor-road is used for 1997 because information on all-weather road was not collected. Poverty rates are higher in isolated localities Rural households can then be classified into villages o f varying sizes and degrees of connectivity. Village size serves as a proxy for population density, whereby large villages tend to be in densely populated areas and small villages in sparsely populated regions. Remoteness is defined as when the distance to the nearest all-weather road exceeds 5km. Table 2.5 presents the poverty rate (proportion of population with consumption less than the national poverty live), for these four types of rural villages. In small or sparsely populated and isolated rural villages, more than half of the population lived below the national poverty line. In larger or more densely populated and well connected villages, less than a third of the population lived in poverty in 2004. These results suggest that isolation due to lack o f infrastructure affects poverty and the existing gap between rich and poor, in turn, underpinsinequality. 33 Sharinggrowth: e p i & and developmentin Cambodia Table 2.5 Percentage of population below the total poverty line, by village size and distance to the nearest all-weather road, 2004 Distanceto the nearestall-weather road Village size Less than 5km More than 5 km More than 1,000 persons 32.5 38.0 (*0.7) (h1.7) Less than 1,000 persons 43.4 52.5 (iO.9) (*1.8) Source: based on the sub-sample from the 1993 sampling frame, 15-month sample, that included both urban and rural areas, based on 15-month sample, CSES 2004. Household-level difference in opportunity and outcome Returns to individual ability As in any society, wealth differences at least in part reflect returns to individual ability and hard work. Cambodian culture is often described by sociologists as being at one level very individualistic: certainly, when asked to list the factors explaining why some households are richer than others, rural Cambodians will credit that for many, their success is due to "good ideas" (khomnuht alohr)-even as others among the rich are seen to have got ahead due to connections,corruptionor inheritance. The perceived importance of individual ability and application is also apparent in the IRL opinion poll. The poll asked the respondents to assess whether their economic situation was better, about the same, or worse than the average people in their village or communities. Some 61 percent repliedthat they were about the same as the average. For those who answered that they were worse off, they were asked two further questions. The first asked them to choose from a list of reasons why others are better off. The second asked why they think they are worse off. In both cases, more than one answer could be given. 81 percent reasoned that others were better off because they "worked hard, had ideas, and took initiative" (Table 2.6). Inequality of opportunity, poverty traps and capital ownership As would be expected, central to the poverty trap is a clear relationship between consumption poverty and poor health. The poor work hard in physical jobs involving greater risk of accidents and injury and have poor nutrition, and have less access to clean water, poorer housing conditions, and less contact with preventativehealth service, all of which make it more likely that they will need medical treatment-which they can less well afford, and which may well push them deeper into poverty. A critical difference between the rich and poor lies in their relationshipto capital (see Box 2.3 for accounts from the MOPS fieldwork). Whereas the poor have to borrow to meet subsistence needs, and frequently then become trapped in a situation in which they work to repay loans and so have no opportunity to save or invest, those with a little security start to lend money to their neighborsat interest, providing a supplementalsource of family income. Table 2.6 Ability, family backgroundand educationwere seen as the most importantreasons why some people can get ahead - answers to the question "If others in your community are richer than you, why do you think they are richer? (no prompts) If others are better off than you in your community, Yo of respondents why do you think they are better off? they work hard, have ideas, and take initiatives to make money 81% they have family members who work in cities or abroad 25% they have education and good skills 63% they have privileges because they're connected to the government, army officials, police etc 17% they are from a rich family 70% they engage in illegal activities 4% they have farms or plantations 10% they have good luck 16% they have good karma 4% they have good health 3% Source: IRL 2007 Box 2.3 Poverty and debt traps It is not easyfor the poor to catch up with the medium because the poor have lots of outstanding debts accumulated over time .,. How can they move out ofpoverty in such a situation? ... In the end, they have to sell their small plots of land to repay their loans ... Losing land is a loss of food security ..,Thepoor have no land or other assets to rely on ... no capital to invest in alternative businesses ... Most poor here take a new loan to repay an outstanding loan, meetfood shortages or cure sick household members ... If someone falls into this vicious poverty, it impossiblefor them to escape ... Unlike the poor, the medium arefree of debt. If they do take a is almost I ~ loan, they use itfor raising a cow that produces a calf; which then can be sold to repay the loan ... There is no hopefor the destitute to escape chronic poverty without helpfiom a humanitarian programme ...I'(Trapeang Prey, Kampong Speu) The have-nots earnjust enoughfor food and consumption and cannot make any savings. Therich can earn more than household consumption... [and] make some savings for investing in new income-earning activities. (moved-out,Krasaing Village, Battambang) Source; CDRI 2007b (forthcoming) 35 Sharinggrowth: egtlity and developmentin Cambodia Investments in enhancing human capability address poverty and inequality To examine the relationship between education and living standards, household consumptioninequality is relatedto the skills ofthe head of household. Two measures of skills are used: literacy and levels of schooling attainment. Table 2.7 presents mean consumption levels according to these two measures of skill levels for 1997 and 2004. Partition (1) presents estimates for literate and illiterate households, while partition (2) presents estimates for those with 7+ years o f schooling and those without. The main findings are that consumption was significantly higher among the skilled, and the differential increased overtime. Table 2.7 Mean levels of consumption are significantly higher for households with skilled heads than those with unskilled heads measured(1by literacy: 1 (2) measured by years of schooling: literate versus illiterate 6 or fewer versus 7 or more years 1997 2004 1997 2004 household heads with-skiIIs 1,906 2,480 2,323 2.824 household heads without 1,740 1,990 1,733 2.137 skills Source based on comparable sub-samples from the 1993 sampling frame, CSES 1997, CSES 2004. Provincial and time variations can be used to analyze the relationship between inequality and the educational stock of Cambodia's population. By pooling provinces in years 1997 and 2004, it is possible to plot provincial Gini coefficients on the y-axis and the proportions of provincial populations with 10 or more years of schooling on the x- axis in Figure 2.7. The result is a statistically significant positive relationship that suggests provinceswith greater shares of highly educated population experiencedhigher inequality. Thus, individual attributessuch as educationalattainment influence inequality through the channel of increasingdisparity of incomes between educated and uneducated workers, especially (as in Cambodia in recent years) during a period of rapid growth and expansion of economic opportunities in manufacturingand services. b r a 1inequakg 36 Figure 2.7 Inequality rose with greater education stocks provincial Glni coefficients versus provincial% popuiatlonwith 10+years of schooling + 40% 10% Pooled Rovinces for years 1997 and 2004 0% 0 5 10 l! % provincial population with lOyrs+ schooling Source; CSES 1997;CSES 2004. Note; Only comparable samples from the identical (1993) sampling frame are used for each survey Thus, the role of Government in providing the services necessary to increase human capability for the young generation, as well as for the adult generation, is critical. An aggressive and inclusive nation-wide universal literacy campaign, with particular attention to out-of-school prime-agedindividuals, should be a pressing priority. Because o f its history, Cambodia has the least educated adult population in East Asia. If it is able to turn this entire generation o f adults into literate citizens, the Government will undoubtedly be able to multiply productivity and growth. As Chapter 3 will show, the returnsto schooling have been high and increasingin recent years. Investment in physical capital will also address inequality In addition to human capital, differences in investmentsin other capital, such as land, are also important factors for income differentials. Investments in land depend on security of tenure. To illustrate this, investments, productivity and consumption are compared between households with and without any paper proof of ownership. Table 2.8 shows different types of households with varying shares of land held with any form o f papers proving land ownership (e.g., application receipts, land investigation papers, certificates), and the average consumptionand yields for these households. Those with at least half of their plots o f land held with some form of papers enjoyed daily real total consumption of 12,609 Riel, which is 18 percent higher than those households with less than 30 percent of their plots held with papers. Households with all o f their plots held with papers enjoyedthe highest yields of 9.05 million Riel of annual outputs per hectare, which is 32 percent higher than households with less than 30 percento f plots titled. 37 Sharinggrowth: egtlig and developmentin Cambodia Table 2.8 Household consumption and crop yields, by shares of land with land papers, 2004 Share of land held with any papers real dally household consumption yields measured by value of output of proof of ownership (In Riel, 1993 prices) per hectare (Riel, current 2004 prices) households with less than 30% of 10,290 6,134,149 plots households with at least 50% of plots 12,609 8,350,061 households with 100% of plots 12,407 9,049,014 Source: CSES 2004 We now utilize variations of investments, crop revenue, and consumption across provinces to analyze the relationship between inequality and investments. Figure 2.3 presents the Gini coefficients of outcomes, either measured in consumption or crop revenue, on the y-axis, and actual investment in irrigation works or potential investments (i.e., land titles) on the x-axis. Panel A shows a robust quadratic relationship between inequality and actual investments in irrigation works. Panel B also plots a significant quadratic relationship between inequality and potential investments(Le., landtitles). The results suggest that inequality tends to rise as the economy is just starting to have small proportions o f households investing in their land, but as the proportions of households with investments increased, inequality tapered off. In other words, when the economy has little capital stock, the income gap between the investors (capitalists) and non- investors (the poor or net consumers) is high, and inequality would keep rising with additional investments when the capital stock is low. However, once the economy has sufficiently large capital stock, the income gap begins to close as the proportion of non- investors (net consumers) is dwindling (as we see in the downward turn of the quadratic part o f the curve). The Poverty Assessment (World Bank, 2006) found that households with secure land tenure tend to invest, which resulted in higher yield, larger cultivation revenue, and greater income and consumption. Thus, another measure to close the disparity of rural income is to create incentives for households to invest in their land. b r a l inegualig 38 Figure 2.8 Inequality and investments (actual and potential), 2004 Panel A. Inequality of crop revenue increases with additional actual investments (in irrigation) when the economy has very small capital stock (of irrigation infrastructure) but inequality begins to falls off with additional investments when there is sufficiently large capital stock in the economy c .-.520% :30% - - (310% - 0 %households with o v e r half of their Dlota lrriaated in a t least Iseason Panel B. Inequality of consumption and crop revenue rose with additional potential investments (secure land tenure) when the capital stock is small but inequality tapers off with additional investments once the capital stock is sufficiently large 40% 80% v: 70% 35% i 7 ** 60% a 9, 50% * * '30% 2 40% L u) 25% * .-530% 0 20% 50" -z 20% 10% 15% 4 0% proportion of household! with land tiles5C 0 .lo 20 40 0 20 40 61 for 100%of their plots proportion households with 100 percent plots held with land titles Source: Cambodia Socio-economic Survey 2003-04 39 Sharinggrowth: equi9 and development in Cambodia Institutions, governance and inequalities in power Both the MOPS and the PPA generated detailed discussion about the nature of power in village life and identified high levelsof concern about the behavior of local authorities. MOPS suggests that satisfaction with local authorities is greater in poor and/or slow- growth villages: in dynamic villages where there were resources to be allocated and rents to be extracted, there is less confidence that local authorities will act in the public good. Confidence is particularly low, unsurprisingly, in communities in which households are heavily dependent upon access to local naturalresources. The final chapter of this report (Chapter 7) will look at the theory regarding the influence that institutions have on economic outcomes (such as equitable development). It will do this by focusing loosely on two key roles of the state-namely (i)raisingand spending public money and (ii)establishingviable and equitable lawsto provide essential regulation and protection to interaction between citizens, between citizens and the state, and between the state and the private sector. Box 2.4 Individuals with a "strong back" (connections or corrupt influence) enjoy greater-often illegal-economic opportunities Two lunds of power were identified, negative and positive power. The concept of power is essentially paternal in nature, positive power i s associated with the qualities of a good father, often described in terms of the parent child or teacher student relationship ...[whereas] negative power i s abusive and corrupt; self-interested and greedy This kind of power is associated with wealth, ... greater rights and freedom and the ability to control access to resources and decision-making. Power i s also associated with having or buying connections and influence, or "strong back" [khnorng thom, literally "big back"]. In some communities, those with this kind of power included rich households within the village; in other villages power was concentrated in the hands of outsiders, and no villagers were perceived to have this kind of power over other villagers. For example, in Babaong, rich households control access to markets, and in Kompong Thnoat local traders control prices, while in several villages, participants said that money-lenders had considerable power over their lives. In D a n g K d a r and Kanchor, some households with money were able to access rights to logging which were denied to ordinary villagers.. It i s very difficult . for ordinary villagers to question the behaviour of those who have economic power over them or complain about abuse or exploitation. "Itis ea9 to buypower heresince thepockets ofallhigh-ranking o@cials are open ...Thosewithpowerjust make afew tnjx to theforest and cut trees; then thg can earn enough mony to cover their expensesfor aposition ... It would befortunate)r us ifthy didn't use theirpower to reapprojitsfrom us thy threaten other villagersfor mony How can thepoor survive? noma4 the victims and the powefulpeople are those who benej2 ......Unlike ... Thepoor are ... butthat is not the case. Norma4 bgore, thepoweful should not be respected. But what can we do? Thy haveguns and alw9.r make mongfrom our backs.'' (Community timeline focus group, D a n g Kdar) `The nib have more power. They have power to contml the local market price ... Thg have lots ojjobs and have mony to hire us. Sznce there are neither companies norfactones m or near5 thzs village, and there are on4 8 salt- fams in our area, the saltfam workers' wage depend on thefarm owner's words.'' (Moved in focus group, K a m p o n g Thnoat) I I - - - - I I_ - ___ I 3. Urban inequality Summary e Offsetting forces left urban inequality largely unchanged between 1993 and 2004. Trade and globalization tended to widen inequality of earnings, as wages for skilled workers rose rapidly. On the other hand, there were also equalizing forces from structural transformation of the urban economy: with increased trade and foreign direct investment, labor force participation has increased, and employment has shifted from low productivity agriculture and allied sectors to the higher value-added , garment manufacturingand service sectors. The fastest growing sectors are intensive iI in unskilled labor, the most abundant asset ofthe poorestCambodians. I e While urban inequality remained stable, it remains considerably higher than rural inequality, reflecting a more diversified set of economic opportunitiesthat result in a ~ wider spectrum of income. The other reason for higher inequality in urban areas is I the greater importance of capital, which is relatively unevenly distributed in a poor 1 country. . . . . .~ "..... ." _. ^...___._"I_ ._._"II.. . . .. I. Chapter 2 showed that rural inequality rose between 1993 and 1997 and only edged upward modestly between 1997 and 2004. Consumptiongrowth was concentrated in the extreme top end 1993-97, but widely shared among the rural majority 1997-04. The two factors that drove rural inequality were (i) between-quartile inequality and (ii)within richest quartile inequality. The latter rose rapidly, and constituted an increasing share of total rural inequality overtime. The causes of inequality were both spatial and social. Isolation or geography matters because spatial disparity contributes to the income gap betweenthe rich (and connected) and the poor (and isolated). Individuals' initiative and investment matter because social disparity in terms of schooling, skills, and initiative contributesto both inequality within the rich and inequality betweenrich and poor. This chapter examines why urban inequality remainedessentially unchangedbetween 1993 and 2004, and remained higher than rural inequality. The first section explains what took place during the past decade by presenting inequality and consumption growth in sub-periods of 1993-97and 1997-04. The second section analyzes the components of urban inequality, by examining between-quartile and within-quartile inequality. The third section presents why inequality remainedat a higher level than rural inequality. The last section investigateshow the urban economy was transforming in such a way that it resulted in falling inequality during this decade. 41 Sharinggrowth: egz@ and developmentin Cambodia What happened to inequality? This section presents what happened during the 1993-2004 period in terms of inequality and household consumption growth in urban Cambodia. Table 3.1 presents the Gini coefficients for real total consumption in years 1993, 1997 and 2004. Between 1993 and 1997, the Gini coefficient fell from 0.465 to 0.433. It remained unchanged between 1997 and 2004. Table 3.1 Gini coefficients of consumption for comparable samples of urban areas (i.e., Phnom Penh and other urban), 1993,1997 and 2004 1993 1997 2004 total household real 0.465 0.433 0.432 consumption (.006) (.008) (.007) Source: SESC 1993; CSES 1997; CSES 2004. Note: comparable samples refer to sub-samples from the identical (1993) sampling frame for each survey year. The growth incidence curves in Figure 3.1 track the percent change in real consumption between two points in time for every percentile of the consumption distribution. Figure 3.1 presents three growth incidence curves for (i) the decade of 1993-2004; (ii) the sub-period of 1993-1997; and (iii) the sub-period of 1997-2004, respectively. The top panel shows that growth between 1993 and 2004 was widely shared, with the bottom half growing at a more rapid rate than the upper half of the distribution. The growth rate at the median of about 20 percent was significantly higher than the mean growth rate of 8 percent, suggestingan equalizing effect. This fall in urban inequality between 1993 and 2004, as in rural areas, masks two sub- periods of very different changes. The consumption changes for each percentile during the earlier sub-period of 1993-1997 are plotted in the growth incidence curve in the middle panelof while those for 1997-2004are represented in the bottom panel. Sub-period 1993-1997: Negative, albeit equalizing, growth Over 1993-1997,growth was negative for the entire distribution. (See chapter 1 on how this might be explained as an across-the-distributionunder-reportingof consumption in 1997 due to political tensions in the survey month.) However, the losses in consumption were more substantial for the top half than for the bottom half. Starting from the 5th percentile of the distribution, growth decreased monotonically. Growth was about zero for the poorest 20thpercentiles of the urban population, but about minus 25 percent for the richest urban quartile group. The period of 1993-1997 can be characterized on the basis of the SESC and CSES data as one of equalizing but falling welfare. Two striking features for the period of 1993-1997 are: . Urban inepaLiiy 42 Growth was recorded as negative for all households except for a few bottom percentileso fthe urbandistribution. Consumptionlosseswere greater for the upper halfthan the bottom half. The implications of these features are a negative median and a negative mean growth rate; as well as a substantial, statistically significant, fall in urban inequality between 1993 and 1997. Sub-period 1997-2004: sharing growth broadly The changes in the second sub-period appeared to be an (opposite) mirror image of the first period. Despite the fact that growth in consumption rose almost monotonically from the lower to the higher percentiles, this period saw growth being broadly shared. While the bottom half experienced lower growth than the upper half, the differentials were not drastic. Growth at the medianwas about 40 percent, comparedto growth in the mean at 30 percent. Thus, the combination of different experiences in 1993-97 and 1997-04-that is, an earlier period of negative growth with more substantial consumption losses among the rich than then poor, and a subsequent period of broadly shared growth-has resulted in falling urban inequality, and broadly egalitarian growth of 20 percent for the median, and 8 percent for the mean, during the decade. The trend during 1993-2004is promising, and suggests that somethingwent right in the urban sector. 43 Sharinggrowth: egtlizj and developmentin Cambodia Figure 3.1 Changes in real consumption, percentile-to-percentile,for periods (i) 1993-2004, (ii) 1993-1997, and (Hi) 1997-2004. Growth incidence curve Phnom Penh and other urban, 1993-2004 (in 1993 sampling frame) - ----- -- ---- Growth in mean i 2 5 10 7 5 100 Percentile o f the population (ranked by total household consumption) Growth incidence curve Phnom Penh and other urban, 1993-97 (in 1993 sampling frame) - - - - - __ __ - Growth In median Growth in mean $ n Percentile o f2sthe population (ranked by total household consumption) so 7s 100 (irowth incidcncc curve Phnom Penh and other urban, 1997-2004 (in 1993 sampling frame) _ - _~ Y4. __---___- Growth in median ~ ~_ _..... 3 2 5 5 0 75 LW Percentile o f the population (ranked by total household consumption) Source: SESC 1993; CSES 1997; CSES 2004. Note: comparable samples refer to sub-samples from the identical (1993) sampling frame for each survey year. Urban ineatialib 44 The components of urban inequality The previous section discussed the changes in inequality during 1993-2004. This section explores the current state of urban inequality. A symmetric approach to rural inequality is used to analyze this question by dividing urban households into four quartile groups. Urban householdsare ranked by their household consumption within an urban- only distribution. The poorest quartile group comprises households from the 1 percentile through the 251h percentile and the second quartile group consists of households from the 261hpercentilethrough the median. The third quartile group consists of households from the 5 1'' percentilethrough the 751hpercentile. The fourth and richest quartile group contains householdsfrom the 76'h percentilethrough the 1OOth percentile. The disparities among the rich contributed the most to inequality: gaps among the rest also contributed, albeit modestly Table 3.2 presentsthe inequality measures of Gini coefficients for each ofthe quartile groups in years 1993, 1997 and 2004. As in rural areas, inequality in the richest quartile group was the highest in all three survey years. However, contrary to rural areas in which there was practically equality in the two middle rural quartiles, there was modest inequality in the two middle urban quartiles throughout these 11 years. Also in contrast to rural richest quartile, inequality within the urban richest quartile fell over time. The Gini coefficients for the urban richest quartile group were 0.33 in 1993 and 0.35 in 1997, but fell to 0.25 in 2004. In a similar manner to the rural poorestquartile group, inequality was modest within the group, and remainedessentially unchangedthroughout the decade. Table 3.2 Gini coefficients of real household consumption by quartile groups, 1993,1997 and 2004 Urban households In comparable sub-samples 1993 1997 2004 Quartile group 1: 1"- 25'h percentiles 0.156 0.150 0.162 Quartile group 2: 26Ihpercentile through median 0.087 0.068 0.103 Quartile group 3: 51'' -75Ihpercentiles 0.090 0.075 0.091 Quartile group 4: 76'h - loothpercentiles 0.332 0.345 0.247 Total 0.465 0.433 0.432 Source: SESC 1993; CSES 1997; CSES 2004. Note: comparable samples refer to sub-samples from the identical (1993) sampling frame for each survey year. 45 Sharinggrowth: eguig and developmentin Cambodia The drastic fall in inequality among the rich, and a negligible narrowing gap between rich and poor, lowered urban inequality Table 3.3 presents a decomposition of urban inequality using the Theil Index. Total urban inequality is decomposed into within-quartile inequality for each of the four quartiles and between-quartile inequality across the four quartiles. For example, in 1993, of the total 0.449 points, between-quartile inequality contributed 0.289 points (64 percent) and within-inequality of the richest quartile contributed 0.153 points (36 percent).Table 3.3 revealsthree findings: . As in rural areas, within-quartile inequality was insignificant among the bottom three quartile groups (i-e., between the 1'' percentile and the 75'h percentile): their joint contribution to total urban inequality is essentiallyzero. 9 In contrast to rural experiences, however, inequality within the richest quartile group became relatively less important over time. Its contribution to total urban inequality fell from 34 percent in 1993 to 22 percent in 2004. As in rural areas, between-quartileinequality remainedthe largest share in total urban inequality. These results, as summarized in Figure 3.2, suggest that the gap betweenthe rich and the poor groups (i.e., between-quartileinequality) is one of the two determinants of urban inequality. This gap is narrowing only negligibly and gradually over time. The other determinant of urban inequality is the inequality within the richest quartile group. This within-richest-quartile inequality fell drastically, especially during 1997- 2004. This reversalof inequality among the richest during a period of high growth is an empirical puzzle which needs further research, beyond the scope of this report. Some potential explanations include (i) the classic case of truncation, Le., the rich may not accurately revealtheir full income and wealth information even ifthey are sampled in the survey; (ii) the measurement o f welfare (consumptionaggregate) can be sensitive to the imputation of services from durable goods. Consumption aggregates are originally computed to arrive at a basket of basic needs which are pertinent primarily to the patterns of livelihood of the bottom half of the population. Thus, consumption aggregates for the richest quartile may not be a true reflection of their consumption, wealth or welfare; (iii) there may be a divergence between income and consumption. Increasing inequality in income in the richest quartile and a falling inequality in consumption may be due to increased savings. The divergent trends are consistent with each other, provided that the rich are increasingsavingsand investments. Urban ineqmh'& 46 Table 3.3 Theil Indices: a decomposition of within-quartile and between-quartile inequality of household consumption, 1993, 1997 and 2004 Urban households in comparable sub-samples 1993 1997 2004 Quartile group 1: 1" 25Ih percentiles - 0.003 0.003 0.003 Quartile group 2: 26Ihpercentile through median 0.001 0.001 0.002 Quartile group 3: 51" -751h percentiles 0.003 0.002 0.003 Quartile group 4: 761h -loothpercentiles 0.153 0.261 0.068 Between Quartiles 0.289 0.246 0.236 Total Theil Index 0.449 0.513 0.312 Inequality in the Richest Quartile Group 4 as a % of Total 34 51 22 Urban Inequality Between Quartile Inequality as a % of Total Urban Inequality 64 4a 76 Source: SESC 1993; CSES 1997; CSES 2004. Note: comparable samples refer to sub-samples from the identical (1993) sampling frame for each survey year. Figure 3.2 A decomposition of Urban Theil Indices z8 - 0.4 7 between 0 2 4 6 0.3 - 0 289 quartile 5 f 0.2 - 0.230 0 L 0 153 richest quart1e0 261 0 068 Note: comparable samples refer to sub-samples from the identical (1993) sampling frame for each survey year. Why did urban inequality remain higher than rural inequality? The previous section shows that, as in rural areas, urban inequality is due to two components: between-quartile inequality, or the disparity between rich and poor; and inequality within the richestquartile. However,the evolutionof these two components in urban areas differed from that in rural areas. This section examineswhy urban inequality remainedsignificantly higher than rural inequality. Urban inequality is higher due to a larger disparity between rich and poor As a first, descriptive cut on the issue, Table 3.4 presents the mean per capita consumption levels o f the poorest and richest quartile groups, in urban and rural areas, separately. The ratio of richest quartile's mean consumptionto that of the poorest was about 7 in urban areas. This ratio increasedfrom 3.2 to 4.4 times in rural areas, between 1993 and 2004. Thus, the rich-poor gap in urban areas is almost twice of that in rural 47 Sharinggrowth: equip and developmentin Cambodia areas. The following section, with additional analyses, shows that this disparity is a result o f differentials in human capital and the greater diversity of opportunitiesin urban areas. Table 3.4 Mean consumptions of richest and poorest quartiles, 1993, 1997 and 2004 Urban Areas 1993 1997 2004 mean consumption of poorest quartile 1,113 1,201 1,472 mean consumptionof richest quartile 7,953 7,479 10,011 Ratio of consumption between richest and poorest quartile 7.1 6.2 6.8 Rural Areas 1993 1,997 2004 mean consumptionof poorest quartile 987 843 1,041 mean consumptionof richest quartile 3,198 3,570 4,534 Ratio of consumption between richest and poorest quartile 3.2 4.2 4.4 Source SESC 1993; CSES 1997; CSES 2004. Note comparable samples refer to sub-samplesfrom the identical (1993) sampling frame for each survey year Urban inequality is higher because of greater human capital stock Drawing from insights in Chapter 2, this section will examine whether education has a role in increasingthe income or consumption gap between the rich and poor in urban areas. The rural and urban samples of 1997 and 2004 have been pooled to examine the relationship between the education stock as measured by the population shares with at least 10 years of schooling and inequality. Figure 3.3 plots the proportion of provincial populations with at least 10 years of schooling on the x-axis and provincial Gini coefficient on the y-axis. The graph shows that there is a statistically significant quadratic relationship between inequality and education stock (in the population). The quadratic relationship suggests that inequality tends to rise as the economy is starting to accumulate education stock (human capital), as increasing proportions of households acquired higher education. This rising inequality is due to the increasing gap between those with human capital and those without. However, once the economy reaches a level at which it has a sufficiently high stock of human capital, inequality begins to taper off and starts to fall with any additional investments or increase in the proportion of the populationthat is educated. This unchangedor falling inequality is due to the closing gap between mean years of schooling or differential human capital in the population. In Cambodia the level of human capital is still relatively low, and increments of education stock would likely drive up inequality. In particular, urban areas have a larger stock of human capital or higher mean years of schoolingthan rural areas. Urban ineatralitv 48 Figure 3.3 The relationship of inequality and education stock 0.5 J 0.4 % 6 .-.-Crn s 0.3 0.2 OpropordZn with ?8yrs+ szRooling4( Source: SESC 1993; CSES 1997; CSES 2004. Note: comparable samples refer to sub-samplesfrom the identical (1993) sampling frame for each survey year Urban inequality is higher because of a more diverse economy Also drawing from findings in Chapter 2 on rural areas, this section investigates whether a more diverse economy with a larger set of productive opportunities is affecting urban inequality, using population shares (by provinces) employed in the secondary sector (i.e. manufacturing and construction) as a measure of the extent of diversity o f economic opportunities. Figure 3.4, based on pooled provincial data from 1997 and 2004, plots the population shares employed in the secondary sector on the x-axis and provincial inequality measured by Gini coefficient on the y-axis. There is a statistically significant quadratic relationship between inequality and diversity of economy. As the economy is starting to diversify, inequality tends to rise. Inequality continues to rise as the economy is diversifying-a consequence of some people getting richer sooner as a result of their participation in the modern sector of the economy-but inequality tapers off as structural transformation takes hold and gains are spread more widely in the economy. The next section analyzes in greater depth and rigor this structural transformation in urban areas in Cambodia. It is argued that urban Cambodia has reached a level of diversity in its economic activities at which urban inequality has started to taper off. The observed fall in urban inequality is a result of the structural transformation, from an agrarian to a manufacturing- and services-based economy, driven by growth in the garment sector, construction,and tourism. 19 Sharinggrowth: equip and development in Cambodia Figure 3.4 The relationship of inequality and population shares in the secondary sector (i.e., manufacturing, construction) 0.6 0.5 0.4 t .-.-00.3 8 C 0) 0.2 0.1 0 propbtion ed?oyed in @e I secondary sector Source: SESC 1993; CSES 1997; CSES 2004. Note: Only comparable samples from the identical (1993) sampling frame are used for each survey. How did transformation of the urban economy influence inequaIity? Cambodia, with approximately US$450 per capita income, is one of the poorest countries in the world. At this stage of development, its economy is moving along the path of diversification from subsistence agriculture. This transformation implies that the economy should be taking advantage of its abundant factor. Export-led growth in labor intensive sectors was the hallmark of early East Asian experiences that lifted many of Cambodia's neighbors from low to middle income status. As employment and value- added spread throughout the economy, it is anticipatedthat inequality falls (see Gill and Kharas2006 on the East Asian experience over recent decades). As is well-documented, garment manufacturingand tourism services are concentrated in urban areas, and both sectors, driving Cambodia's growth in the last decade, are intensive in unskilled labor. This section will investigate whether some structural transformation has taken place in Cambodian urban areas, similar to the experiences of other East Asian countries in their early stages of development, through promotion of unskilled labor-intensivesectors. Urban hequalip 50 Wage employees: employment in hotel, restaurant, and personal services increased 3-fold This section assesses the extent of structural transformation that has taken place in urban areas. Inequality of real wages (as measured by the Gini coefficients) in urban areas fell from of 0.456 to 0.415 between 1997 and 2004. During this period, real wages rose by 63 percent and the share of female workers in the labor force went up. The proportion of paid employees in hotel and restaurant and in social, community and personal services rose about three-fold (from 1.2 percent to 3.9 percent, and from 6.6 percent to 17 percent, respectively). The proportion of wage workers in agriculture about doubled while that in the civil service declined. Those in construction and manufacturing remained stable. The proportion of workers with tertiary education (i.e., class 12 or above) increased from 15 to 23 percentduring these years (Table 3.5). Table3.5 Summary statistics of paid employees in urban areas, 1997 and 2004 1997 2004 Gini Coefficient 0.456 0.415 Real monthly wage (Riel) 114,310 188,231 Male (%) 72.5 60.7 Industry during the past 7 days (%) Agriculture, hunting and forestry; Fishing 3.2 5.8 Mining and quarrying; Manufacturing; Electricity, gas, and water supply 12.6 17.8 Construction 8.9 9.2 Vvholesale and retail trade; repair of motor vehicles, motorcycle and personal and household goods 5.1 3.4 Hotels and restaurants 1.2 3.9 Transport, storage and communications 7.2 3.1 Financial intermediation 2.8 0.8 Real estate, renting and business activities 1.1 0.7 Public administration and defense; compulsory social security 34.8 22.8 Education 8.1 9.9 Health 2.3 2.2 Other community, social and personal service activities 6.6 17.2 Extra-territorial organization and bodies 2.2 3.1 Highest grade level successfully completed (%) Class 3 or below 13.7 16.4 Class 4-6 17.4 18.6 Class 7-9 30.5 26.0 Class 10-12 23.8 16.3 Class 12 or above 14.6 22.7 Number of observations 1,425 3,267 Source: SESC 1993; CSES 1997; CSES 2004. Note: Only comparable samples from the identical (1993) sampling frame are used for each survey. Many moved from agriculture to higher value-added modern sectors Regression results are presented in Annex 2. Table 3.6 summarized the main and year effects o f education and industry from the decomposition of equation (7) in that 51 Sharinggrowth:equip and developmentin Cambodia analysis. Details o f the specific industry's effect can be found in the Annex. The results suggest that a significant portion of wage variation between 1997 and 2004 can be attributed to education and industry effects. The changes in returns to schooling attainment explain as muchas 43 percent o fwage variation. The changingcomposition o f industry-as the labor force reshuffled from low productivity sector to higher value- added sectors-combined with the changing industry premium account for 22 percent of wage variation. These results suggest that a structural transformationhas taken place in urban areas as employment in manufacturing and services absorbed a larger segment of the labor force; a significant number have also shifted from self-employment in (lower productivity) agriculture to be paid employees in higher value-added secondary or tertiary sectors. Table 3.6 Effects of education and industry on wage differences between 1997 and 2004 for urban employees Main effects Year effects Total (changes in the stock of schooling attainment (changes in returns to schooling or in effects or composition of industry) industry premia) education 0.40% 43.1% 43.5% industry 15.2% 6.9% 22.1% Source: CSES 1997: CSES 2004 Thus the story of industrial and service sector growth in PhnomPenh and other urban areas to date appears to have tracked that o f the East Asian miracle countries in their early stages of development. Because of the phenomenal growth in garment manufacturing, construction and tourism services, notably labor intensive sectors, in Cambodia's urban hubs, workers and especially the unskilled in the lower end of the distribution have benefited disproportionately. The country's general economic reform strategy, which includes globalization or opening up of the economy to trade and foreign direct investment,has transformedthe livelihood opportunities o fthe urban population. Returns to education reflect the relative supply and demand of skills The effects or benefits of reform's consequential transformation of the urban economy are shown in Table 3.7. As recently as in 1997, the rate o f return to schooling was flat across all levels of schooling among wage workers: returns to secondary and tertiary different rates of returns to schooling were distinguishable across categories. Workers with tertiary education experienced significantly higher returns than their secondary educated compatriots, who in turn gained significantly higher returns than their primary educated fellow countrymen. The successively higher returns for higher levels of schooling observed in Cambodia in 2004 were consistent with the stylized facts of functioning labor markets in other developing countries. The CambodianGovernment'seconomic reforms freed up market forces that promote efficient reallocationof resources. Workers in 2004 could self-select Urbanineqttality 52 through intojobs that value their particular comparative advantage (attributes, skills, etc). The higher returnsto greater skills in Cambodiareflects increaseddemand for skills from globalization and technology, as in other open economies. Because paid employees only constituted half of the labor force, the results may not be representativeof the reality in urban areas. Using householdconsumptionas proxy for income for all urbanhouseholds revealsvery similar results(see Annex for details). Table 3.7 Rates of return to various levels of schooling, 1997 and 2004 Highest grade level successfully completed (reference group: Class 3 or below) 1997 2004 Class 4-6 0.196 0.166 (0.080)'* (0.050)"' Class 7-9 0.009 0.195 (0.070) (0.040)"' Class 10-12 -0.091 0.341 (0.070) (0.050)'** Class 12 or above -0.079 0.473 (0.080) (0.050)'*' Source: SESC 1993; CSES 1997; CSES 2004. Note: These coefficientsare obtained from running a standard Mincerianwage regressionwith dependant variable as logarithm of real wages, and other explanatory variables include age, gender, marital status (see details in ANNEX); standard error in parentheses;*** denotes significant at the 1% level;** denotes significant at the 5% level: 4. Addressing asset inequality: a focus on land Summary Without off-farm employment or self-employment alternatives, over half a million landless poor find it hard to improve their situation or participate in the economy. Levels of inequality in land ownership are high in Cambodia, and significantly higher than most other countries inAsia. As in other countries in Asia, small farms in Cambodia are more productive than large farms. In crude terms, a more equitable distribution of land will deliver higher agricultural yields and increase the rate of economic growth, with indirect as well as direct gains for the poor. In contrast, a process that concentrates land ownership with de facto redistribution from smallholders to very large plantation-type holdings (as has occurred under the economic land concession program) may slow the rate of growth. Two policy initiatives appear to offer ways to improve equity and efficiency. First, social land concessions (SLC) that redistributes to poor landless households unused land originally allocated as large commercial concessions would provide a basis for livelihood to a significant number of rural poor households. Drawing on the experiences of other countries, the SLC program would complement land distribution with a package of supporting measures to improve access to markets and social services. Second, systematic titling to improve tenure security would reduce vulnerability and uncertainty of those currently working land, reducing conflicts and encouraging productivity-enhancing household investments in agriculture. Although in much of Cambodia locally-accepted de facto ownership is adequate, ownership proof in any paper form i s still extremely valuable. Empirical evidence suggests a robust and strong correlation between secure land tenure and investments in land. Even among poor farming households, the incentives to invest are as strong as the richer households' provided that they have secure land tenure, manifested in holding some paper proof of ownership. Structured contextual/qualitative research, combined with statistical analysis of economic and welfare outcomes associated with other forms of land title in the CSES 2004 data, indicate a number of benefits of secure land tenure. The challenge in the future will be to extend titling from the densely-settled, stable communities where it has operated to date (and where formalization of ownership i s relatively straightforward) to more conflict-prone, marginal areas where titling will need to engage with carefully-managed reclassification of state land. Irrigation is extremely important for improving productivity and income. While policies such as systemic titling will promote private incentives to invest in capital inputs such as water pumps, public investments in irrigations are indispensable for improving the livelihood of an overwhelming 80% of the rural majority whose primary sector of employment is agriculture 54 Addressing asset inegualip: afOcas on land Why is land so important? In looking at how individuals and households can or cannot command access to factors of production, and the implications of this for equitable development, land constitutes a special case (World Bank 2005 pp. 162-168). As land is to all intents a finite resource, the distribution of land ownership and access rights is particularly sensitive to the operation of formal and informal institutions. The importance of land is exaggerated in a primarily agrarian society such as Cambodia in which the majority of the population makes a living as small-holding farmers, meeting the majority of their food and income needs directly from the land. With low levels of education and constrained access to capital, rural Cambodians have limited off-farm employment or self-employment opportunities. This is changing with economic growth and diversification, but for the near and mediumterm land will remain a critical resource for the poor, 91 percent of whom live inthe countryside. Indeed, economic growth, and the necessary long-term structural transformation from an economy dominated by subsistence-oriented, land-basedagricultural self-employment to a more diversified economy in which manufacturing and 'services play major roles, entail some considerable risks of their own. With investment and growth since the early 199Os, land has for the first time begun to acquire significant market value. Inthe past, a low population density and limited economic potential of land underpinned a traditional "legal" framework based on principles of usufruct: that is, a popular understanding that someone who cleared land and cultivated it was assumed to enjoy title. Since the 1990s, however, population growth and market integration have resulted in increasing competition for land, bringing this traditional concept of usufruct into conflict with commercial, state and environmental interestsand claims. These processes help to explain in part the high levels of inequality in land seen in Cambodia. Surprisingly, given the near-zero inequality in landholding when land was distributed in 1989, the inequality of land distribution in Cambodia is very high in comparison to other Asian countries. This inequality reflects a skewed distribution in which there are a large number (an estimated 46 percent) of rural households that are landless or land-poor (that is, own less than 0.5 hectares); and a very small number who control vast areas of land (Figure 4.1, Box 4.1). Figure 4.1 Land distribution in Cambodia ranks amongst the most unequal in Asia Thailand Indonesia Malaysia India Cambodia Gini coefficient (0 = perfect equality, 1 = perfect inequaldy) Source: WDR, IFAD Sharing growth: equity and development in Cambodia 55 -- -- - - __ __ - ---__ - -__. -- ___I I__ -.._--I_____ Box 4.1 Patterns of landholding:findings from a land survey The inequality o f landholdingis underestimatedby household sample surveys, as much landmay be heldby companies or individuals(generally very richhouseholds not necessarily present inthe community) who are unlikely to be covered by a survey. Oxfam GB recently conducteda survey of landholding patterns in 433 villages, selected through a multi-stage process. Although the sampling design means that results cannot be taken to be nationally representative, the findings are still interesting,in that they suggest that in many localitiesa large proportionof "normal" land (that is, privately-ownedland, not including state landallocatedas an economic landconcession) i s heldby powerful individualsgenerally not resident in the community. Ineach sampledvillage, the researchers sought to identify the owners of all land. The finding was a high degree of inequality: 12 percent of owners with holdingsof greater than 3 hectares each owned atotal of 72 percentof the land. a. 67 percentof owners holdingsunder 1 ha... b. ...collectively own < 8 percent of all land 1 72% <0.5 0.5- 1 1-3 > 3 <0.5 0.5- 1 1-3 > 3 size of landholding (hectares) size of landholding (hectares) The report also found that, as would be expected, intensity of land utilization was relatedto size of holdings: while on average 98 percent of holdings under half a hectare were worked, only 71 percent of holdings larger than 3 hectares were under cultivation. Amongst those owning more than 500 hectares each, 31 percent were described as businesspeople; 23 percent as high-ranking officials bearingthe title "His Excellency", 23 percent were okhna (a social title conferredby the Government in recognition of financial contributionsof US$lOO,OOO or more); 15 percent were military officers referredto as Generals; and 8 percent were members of the National Assembly. These figures should be taken as broadly indicative only, in part because there is presumably considerable overlap betweenthese socialand occupational categories. Source: OxfamGB 2007. Inrespectto land managementas inmany other aspects, development policy issues in Cambodia must be worked out in the context of both dramatically uneven power relations and a pronounced legal pluralism (a concept developed further inchapter 7). Landpolicy issues in Cambodia are complex in part because of the importance of land to a variety of actors pursuing a variety of goals. Cambodia currently has no real consensus on the principles on which land claims should be based. This makes it hard to establish and sustain strong institutions for land managementthat command legitimacy and compliance from all actors. To a significant degree, this situation is a legacy of Cambodia's 56 Addressin2 asset zneaualitv: a focus on land protracted conflict and political turmoil: Cambodia has gone through radically different political regimes, each associated with markedly different principles and institutions for allocating landproperty rights (see Box 4.2). Box 4.2 Competition and confusion in land management reflects Cambodia's turbulent history The current era of land administrationbeganwith the emergenceof a landmarket inthe late 1980s and early 1990s (Ashley 1999; Cooper 2004; Oxfam 2005). Prior to this, Cambodia experienced a rapidsuccession of radicallydifferentregimes of land administration. Before 1970, a mixedsystem of traditional usufruct and French-style administration predominated; between 1975 and 1979, privatepropertyrights were abolished by the Khmer Rouge; and from 1979, Vietnamese-style low- level collectivizationwas introducedand then gradually disappeared. As collectivization began to break down in the mid-l980s, the Government passeda sub-decree allowing land to be occupied and farmed by individual families. In 1989, as the shift from a socialist to a market economy accelerated, a partial returnto private land rights began in April 1989 with a sub-decree allowing ownership rights over residential property. This was followed soon after by Instruction No. 3, which formally ended the collectivefarming systems and reconfirmedhouseholdpossessionrights over agricultural land. The 1992 Land Law consolidated the market for land, providingfor a system whereby ownership could be conferred upon legal possessorsof land. However, it was not longbefore the Government, influencedby spiralinglandvalues, uncontrolledencroachmenton state lands and widespread land conflict, decided that a new land law was necessary (Williams 1999). Prospects of increasing economic growth and investment on the one hand, and social pressures on the other, created incentivesto modernize landmanagement systems.Donors also playedarole inthe developmentof the new law, signaling that large scale support for land administration was contingent on the developmentofa satisfactory legalframework. Work began on re-draftingthe land law in 1996, with technical assistance from the ADB, and the new land law was passedin2001. Some ofthe major changesincludedextending privateownership rights to residential land and agricultural land; establishing a system for the systematic titling of land under the control of the Ministry for Land; creatingan administrativesystemfor the resolution of disputes over unregistered land (the Cadastral Commission); and bringing to an end the possibilityof legallyentering into possessionofvacant state land. This chapter examines land issues and possible policy responses through the lens of equitable development. Firstly and importantly, it notes that there appears to be a strong complementarity between equity and efficiency considerations when it comes to land management: smallholdings are more productive than are large holdings. Secondly, the chapter looks at the magnitude of poverty and landlessness, and the case (both interms of growth and equity) for a carefully-managed program to redistribute unused land to the landless poor. Finally, the chapter examines the issue of land title and the growth and distributional outcomes that might be expectedfrom a program of systematic titling. In respect to land distribution, equity and efficiency are fully complementary Analysis of CSES 2004 data shows-as in many developing countries-a clear negative relationship between size of household landholding and a number of desirable economic outcomes (including crop output / hectare and crop income / hectare). Inother Sharing growth: equity and development in Cambodia 57 words, small farms are clearly more productive, and generate more wealth, than do large farms (Figure 4.2). Figure 4.2 Small farms are a better economic proposition than large farms a. Small farms have higher outputs per hectare... b. ...and higher crop income per hectare (y axis plots log of output value I ha.) (y axis plots crop incomeI ha.) 8 (D e a I I 1 I Source: CSES 2004. This analysis, and other land analysis based on the CSES data, is complicated by a number of other considerations. Firstly, it is important to remember that the CSES is based upon a sample of households; it does not necessarily represent the full picture of land ownership. Very large plots (in theory, limited to 10,000 hectares each, but in practice often exceeding this limit) allocated as economic land concessions (ELCs) are not captured in the CSES data. Inthe particular case of concern here, however-that is, the case of land-productivity relationships-the inclusion of the ELCs would only further strengthen the case for smallholdings, as the vast majority of these concessions have not been planted since they were allocated, strongly suggesting that they are being held as speculative rather than productive investments, and currently make no contribution to the economy. A further qualifying observation would be that an observed inverse relationship between plot size and productivity does not automatically lead to the conclusion that smaller i s always better, and smallest i s best. Higher productivity notwithstanding, there i s a minimum size of landholding below which it cannot support a family. High productivity in this case reflects in part a lack of viable household alternatives: in a process sometimes described as "self-exploitation", additional unpaid family labor is invested in agricultural work even as diminishing returns set in (and a commercial farmer 58 Addressing asset inegualip ajcus on land would cease to invest further in paid labor), simply because the family lacks other, more profitable sources of food security and income. Thirdly, while smallholdings tend to be more efficient inproduction, large farms tend to be more efficient in marketing, where there are returns to scale. There are several ways inwhich smallholders can overcome this scale disadvantage-for example, through contract marketing arrangements or organizing themselves for collective marketing-but in the absence of these institutions, they do face difficulties translating productive efficiency into improved incomes and accumulation. With these qualifications in mind, the point nonetheless remains that small farms enjoy a considerable advantage over large farms in productivity. This implies that in aggregate, a national distribution of agricultural land dominated by smallholdings is likely to deliver significantly higher agricultural yields and agricultural contribution to GDP than i s a distribution in which landholding is concentrated in a small number o f very large plantation-type holdings. A redistribution of land from large holdings to small will increase the rate of economic growth, with indirect as well as direct gains for the poor; a process that concentrates land ownership with redistribution from smallholders to large holdings will slow the rate of growth. Landlessness and poverty Numbering over half a million, the landless rural poor are a significant concern It needs to be remembered that not all rural poor are landless, and not all rural landless are poor. In 2004, the incidence of poverty in Cambodia was higher for rural households with land (35.2 percent) than those without (25.3): see Table 4.1. This pattern i s seen in each zone by urban and rural sectors, except that in rural Plains region, poverty rates were similar between landlessand landed households. However, the static picture of landlessness being associated with higher living standards may give an erroneous picture of trends over time. It is possible that the very poorest farmers are forced to sell their land because of debt or catastrophic illness, and that joining the ranks of the landless leaves them worse off with fewer income earning opportunities. Alternatively, a proportion may become landless by choice in order to pursuebetter income opportunities through migrationand wage work. However, we can examine whether the landless are worse off inother respects. Table 4.Table 4.2 shows that on average the landless poor in Cambodia are not necessarily worse off or more disadvantaged than landed poor. The landless poor were not worse off in terms o f access to primary and secondary schools or availability of amenities in their villages (e.g., bank, permanent market). Furthermore, the landlesspoor and landed non-poor seem to have similar likelihood of renting inlandfor cultivation. Sharing growth: equity and development in Cambodia 59 Table 4.1 Poverty and land-holdingstatus in Cambodia by zone and urbanlrural sector in 2004 Households with land Households without land % Mean daily Poverty rate % Mean daily Poverty rate consumption per (%I consumption per (%) capita capita National 72.2 2 794.8 I 34.8 27.8 5,191.7 19.4 Total Urban 32.6 4,033.8 28.6 67.4 7,593.7 8.9 Urban Phnom Penh 5.1 13,477.2 2.6 94.9 10,445.6 2.0 Urban Plains 41.2 3,902.7 19.9 58.8 7,333.6 4.7 Urban Tonle Sap 40.6 3,646.0 35.4 59.4 5,245,3 16.2 Urban Coastal 44.2 3,746.2 28.5 55.8 5,036.3 10.6 Urban Mountains 56.8 3,082.0 28.8 43.2 4,391.2 23.4 Rural Cambodia 79.0 2706.7 35.2 21.d 3,861.3 25.3 Rural PP 8.4 6,129.5 3.1 91.6 6,553.2 7.5 Rural Plains 80.9 2,957.7 28.7 19.1 3,154.8 28.2 Rural Tonle Sap 79.6 2,447.5 42.0 20.4 3,346.9 30.7 Rural coastal 90.6 2,772.3 29.4 9.4 4,492.0 7.9 Rural Mountains 89.5 2,117.6 51.1 10.5 3,229.6 40.4 Source: Van de Walle, "Land inequality in Cambodia," unpublished manuscript, the World Bank, 2007. Table 4.2 Landless poor are not more disadvantagedthan the landed poor Poor Non-poor landless landed landless landed Village characteristics % with Primary school in village 54.63 48.55 59.97 54.43 Distance to primary school (km) 1.48 2.49 1.44 1.90 %with lower secondary school in village 10.32 8.01 17.98 11.59 Distance to lower secondary school (km) 5.70 7.30 3.73 5.28 Distance to bus stop (km) 15.72 25.87 12.3 23.31 %with all weather road in village 78.54 69.68 81.8 74.84 Distance to all weather road (km) 2.86 4.62 2.7 3.58 % with a Bank in village 11.05 9.27 12.0 11.25 % with Agricultural extension agent in village 7.49 5.64 8.2 6.67 %with Permanent market in village 12.42 4.90 26.4 7.33 % with Development project in village 41.17 39.64 37.4 38.88 % with Land available for rent 82.00 74.18 69.70 76.33 Household characteristics Head's years of education 2.97 3.03 4.84 3.97 Spouse's years of education 1.41 1.63 2.75 2.29 Number of children 2.89 2.75 1.37 1.75 Number of elderly 0.13 0.18 0.43 0.24 Number of able bodied adults 2.48 2.81 2.56 2.60 Observations 650 3303 1908 6120 Source: Van de Walle, "Land inequality in Cambodia," unpublished manuscript, the World Bank, 2007. Landholding data from the MOPS research confirms this finding at the local scale, and seems to confirm the obvious explanation: the richtypically have more profitable and dependable sources of livelihood than farming. By extension, it is true that land distribution is not necessarily the best solution for all landless poor households: some, like the rich landless, may have identified more promising non-land employment or livelihood opportunities. Finally, it i s recognized that Government development strategy 60 Addressing asset inegtlalzp: afocus on land can and should seek to facilitate the gradual long-term transition of the population out of agriculture and into higher value-added manufacturing and service sector activities, and not seek to set instone a pattern of subsistence-orientedlow-input smallholder farming. With all these considerations in mind, there remains a very strong case for using redistributive policy to improve the welfare o f the rural landless poor (and, based on the evidence of smallholder productivity, to boost aggregate agricultural production). While the share of landless poor in the total population seems relatively small at 6 percent, in absolute terms the problem is considerable, and a serious challenge for Government. There were about 11 million people in rural areas, and therefore, over 600,000 people were in the ranks of the landless poor. Cambodia faces a tremendous challenge in ensuring productive employment for its rapidly expanding labor force. With rapid growth in the working age population as the post-Khmer Rouge and post-PRK baby booms grow into adulthood, and to date limited growth in non-agricultural employment or livelihood opportunities, the involuntarily landless are likely to suffer from much reduced opportunities and much increased vulnerability to poverty. Regional experience suggests that even under very favorable conditions, the growing off-farm economy absorbs labor only slowly. The share of employment in agriculture in Thailand and Indonesia in the 1970swas similar to that inCambodia today (c. 65 percent). After thirty years of rapid (if variable) growth that has fundamentally transformed these economies, agriculture still employs 40-50 percent of all workers (Butzer et a12003). Government strategy recognizes the need for land redistribution In attempting to bring land ownership and land management within the rule of law, the Government recognizes the need for a redistributive component, including in the Land Law a provision for "social land concessions" (SLCs) by which the Government ` would allocate state land to the poor for housing and agricultural use, with ownership rights granted after five years of continuous occupation and use. The design of the SLC program attempts to build on key lessons learned from previous international and Cambodian experiments in land distribution (see Box 4.3). Are SLCs the best use of state lands for growth and employment? Social land concessions have the potential to be an important component of Cambodia's overall agricultural development strategy. There is a strong rationale for redistribution from the perspective of employment generation and equity. Given increasing competition for land resources, particularly in the areas most suitable for agricultural development, it is then reasonable to ask how social land concessions can contribute directly to agricultural growth. The first issue to be addressed is the potential for small farms to produce efficiently and profitably relative to larger farms. As mentionedabove, overwhelming international evidence suggests that there is no dichotomy between growth and poverty reduction when it comes to defining preferable outcomes interms of land distribution: infact, small- and medium-sized family run farms are generally more efficient in terms of productivity per hectare than large commercial enterprises (see Heltberg 1998 for a review). The observed predominance of large farms in developed countries is generally a consequence of the high costs of labor relative to capital comparedto developing countries (Hazel12001). In Sharing growth: equity anddevelopment in Cambodia 61 some developing countries, particularly in Latin America, the relatively large proportion of large farms reflects the legacy of policy distortions from the colonial and post-colonial periods which favored haciendas-creating the basis for continuing social unrest in a number of countries, such as Columbia (Binswanger, Deininger and Feder, 1995). Box 4.3 Lessons learned and applied in Cambodia's land distribution program Lesson I: Objectives need to look beyondjust the transfer of land: for land to be effective in improving livelihoods, it requires complementary interventions (access to basic services and infrastructure). Lesson 2: Mechanismsfor identifyinglandneedto be basedon localcontextandwork eflciently, The decisionto focus on redistributingstate land presents several challenges: much is unsuitable for cultivation, under forest, already distributedas economic concessions, or informally occupied. For SLCs to succeed requires finding ways to recover illegally-claimed tracts and reach consensus on reallocation of specific state lands. Policy documents provide the basis for recovery, mapping and allocationof state lands, but it is not yet clear ifthese mechanisms can be implemented in a manner that meets all the desirableprocess criteria. Lesson 3: Information, transparency andparticipation are essential in identifyinglandrecipients. The credibility of land distribution depends largely on its integrity in selecting recipients (who must be both poor and landless or near-landless). This requires clear and relatively simple selection criteria, and engaging communities in a transparent process. Engaging third parties (farmer organizations, NGOs, religious authorities) can help. Finally, dispute resolution mechanisms are needed to deal with inevitable disagreements. The program can draw on mechanisms, developed inrespect to other programs in Cambodia, to addressthese challenges. * Lesson4: Let benejkiariestake the leadin the allocationofplots andlanduseplanning. The allocationof specific plots to households can be a source of conflict. The SLC approach to land reform requires groups, rather than government agencies, to take responsibility for the selection of landand how it is allocatedamongst group membersthrough a participatory landuse planningprocess. Lesson 5: Focus delivery of services andinfrastructure at the locallevel As with beneficiary selection, provision of complementary livelihoods investments and services i s best organized at the local level through locally-based, participatory planning and delivery. For many complementary services, ensuring economic and political sustainability suggests that the best approach is one which emphasizes a proactive guarantee of inclusionand universalism, rather than narrow targetingto beneficiariesofthe landredistribution. Source: World Bank 2002; van den Brink et a12006; Habib2002, cited in Schonberger 2007. There is limited systematic assessment of the issues or evidence related to farm size and productivity and broader agrarian structure in Cambodia. Initial results from enterprise models commissioned by the World Bank with MAFF indicate that even under very difficult conditions for factor markets (e.g. for credit on good terms), for most crops grown in Cambodia there are limited if any economies of scale in terms of land productivity (see Figure 4.2 above). When economies of scale are apparent in profitability for some crops, they are quite small (Figure 4.3), and outweighed in importance by significantly lower rates of employment utilization. Small farms can generate as much as five times more employment per hectare at similar levels of productivity and profitability (MAFF). This holds true not only for traditional 62 Addressin1 asset ineqzlality: a foczls on land smallholder crops such as rice and vegetables, but also for a number o f rapidly expanding cash crops such as cassava. Figure 4.3 For most crops, there is no significant economy of scale effect 1 Small (<3ha) Medium (3 to 10 Large (>IO ha) T 7- Wet season Cassava Soybean Cabbage rice Source: CSES 2004 The primary challenge to a small farm-based approach is to improve the linkages o f small farmers to the rest o f the supply chain. Agriculture i s increasingly moving towards higher value-added products based on closely integrated supply chains to control quality and timing o f production. Small farmers have been able to integrate into these supply chains when they have been effectively organized through farmer associations and/or through contracting arrangements with processors or aggregators. Cambodia has limited but important experience with contract farming, and extensive experience with informal aggregators supplying processors across the Vietnamese and Thai borders. In addition, increasing attention i s being focused on developing farmer associations which can reduce the cost for processors and marketers to engage with small farmers, as well as to better defend the interests o f farmers in contracting negotiations and accessing input markets (ibid). Social land concessions alone cannot provide the basis for improved agricultural growth and employment, but neither can economic land concessions. Both social and economic land concessions are simply mechanisms for intensifyingland use: one i s based on labor and the other on expected capital investments (see Hiyami in SNEC 2005). In both cases, their success will depend largely on their ability to integrate into higher value supply chains. These, in turn, require improvements in the investment climate and infrastructure which will attract additional investment into the downstream links such as processing, transport and marketing. Experience from numerous countries, particularly in Southeast Asia, has demonstrated that either a small or large farm approach can work when the downstream investmentis inplace (Swinnen). Government may therefore wish to focus, as have several countries in Southeast Asia, on facilitating domestic and foreign capital investment into agricultural processing, transport and marketing, rather than focus Sharing growth: equity and development in Cambodia 63 almost exclusively on primary production (which can be implemented as efficiently by small farmers). Alternatively, or as a complement, Government can prioritize investments in transport and communications infrastructure in order to better integrate into the expanding regional production networks which can improve linkages between Cambodian farmers and processors and markets inEast Asia. To summarize, the allocation of land for social land concessions over other, more extensive approaches to intensification of land use is not a trade-off between growth and poverty reduction. Rather it i s a choice between prioritizing the use of Cambodian employment relative to imported machinery and labor in improving land use for agricultural growth. As the review of international experience with land distribution programs highlights, to obtain both the additional employment benefits and growth through land redistribution requires that the Government take a muchmore proactive role in supporting investment and service delivery (particularly during the initial years of redistribution) than would be required if the Government opted instead for leasing land for plantations. However, if complemented with improvements inthe investment climate and linkages to regional production networks for processing and marketing, social land concessions can support the effective integration of land recipients, along with the other 90 percent of Cambodia's farmers with less than three hectares, into a growing agricultural sector, with significantly greater effects interms of growth, employment and poverty reduction. Land titles and agricultural development Conventionally, secure title to land is seen as very important for economic development. Clear property rights backed up by impartial courts are seen as critical underpinnings of a market economy. Without such security of tenure, it i s argued, those working the land will be unwilling to invest in productivity-enhancing improvements (irrigation, land improvement such as leveling, etc.) or diversify production into more profitable cash crops, especially perennial crops such as fruit trees which require considerable investment and a lag of several years before yielding a harvest. There i s also an argument that secure title makes it easier to use land as collateral: thus, formal title may make it possible to obtain a loan when one wouldn't have been obtained before, or it may improve the terms of a loan, as borrowers pay a premium to obtain money against un-titled land. Amongst those owning land, the rich are more likely than the poor to possess papers-although undocumented ownership is high for both Table 4.3 shows that the proportion of plots or percentage of operating area held with some forms of papers as proof of ownership increases steadily with wealth (per capita consumption). Amongst the poorest quintile, 40 percent of land i s held with papers; amongst the richest quintile, just under 60 percent of is held with such papers. Thus, a significantly larger proportion of wealthier households enjoy the greater security of land tenure that comes with documentation. However, it i s also worth noting the even amongst the rich, a significant proportion of land (40 percent) is held without papers. 64 Addressing asset inequalip: afocus on land Table 4.3 The richest are 50 percent more likely than the poor to possess papers supporting land ownership - Distribution of paper proofs of land ownership, 2004 land heldwith some form of land paper as proof of ownership Poorest poorest Next richest Next Richest All % of operating area within each quintile 40.9 50.8 55.0 63.1 59.8 54.5 Source: CSES 2004 (15-monthsample) Note: papers referto applicationreceipts, land investigationpapers,certificates/titles,contracts,others. Land titles and legal pluralism Greater tenure security for small farmers in Cambodia, in the form of universal possessionof formal land titles, would undoubtedly be beneficial. However, the current baseline or control situation-that inwhich land title is available for only a proportion of all plots-may not be at present be as restrictive as theory would suggest. Tenure in Cambodia is arguably more secure than might be assumed given the fact that few households have formal title (Adler et a1 2007 p.7). While land disputes are common and high profile, it is hard to substantiatea pictureof generaltenure insecurity (Box 4.4). --__ __- - - - -- I - ___ -_I- _I Box 4.4 In much of the Cambodian countryside, land markets appear to work I I reasonably-for now-despite weak or non-existenttitles I In CSES 2004, only 1.8 percent of plots are reportedas havingever been subject to a conflict. Of these disputes, two thirds would appear to be minor, having been resolved (presumably at the local level) within three months. Although only 49 percent of plots are reported as having any paper proof of ownership (including titles and receipts for title applications), 77 percent of plots are reported as beingusable as collateralfor a loan. These findings are supported by a variety of qualitative and quantitative sources which show that in much of the countryside, land is bought, sold, rented and used as collateral without formal title. Thus the Tonle Sap PPA found that in much of Cambodia, locally-accepted de facto ownership suffices under normal circumstances: "In most villages, the traditional land titles over residential andfarm land seems to befairly well establishedand generally accepted as collateralfor loans andforJinancia1 transactions, such as sale, lease or other. Many of the villagers do not have land titles, but this is rarely mentionedas an obstacle." (CDRI 2007a p. 156) .- - - ---- -__ - - - -. - - - -.-_ Researchsuggests that, rather than constituting a general problem, tenure insecurity is concentrated among vulnerable groups, particularly poorer householdswho occupy lands outside of core residential or farming zones, such as those which are or were forests, flood plains, seasonal lakes, marshes and informal urban settlements-that is, land contestedby the state (WB/CAS 2006a, 2006b; O'Leary 2005). One of the key characteristics of these vulnerable areas-and the poor withinthem- i s that land use here i s subject to different, conflicting sets of principles and norms regarding land rights. Case study research suggests at least three sets of norms are at work. The first is that of social norms, or "living law": the historical concept of usufruct-namely, the idea that vacant land can be occupied and that the act of farming land gives rise to ongoing usage rights-is deeply embedded (despite civil war and changes in state law) inrural Cambodian perspectives and practices. Secondly, contested Sharing growth: equity and development in Cambodia 65 land i s often the object of neo-patrimonial administrative conventions, deriving from both previous regulatory regimes and from informal state practices that empower government officials at various levels to authorize transactions over lands within their administrative jurisdiction. Finally, lands that are under conflict are, as with all land in Cambodia, subject toformal statute, inthe form of legal principles derived from the 2001 Land Law and other regulations currently inforce. Social norms of usufruct were both reaffirmed and limited by the 2001 Land Law which converted possession rights into ownership in certain cases. At the same time, however, the 2001 law restricted legal possession of other lands, most notably `state public land' and lands which entered into private possessionafter the promulgation of the law. But the 2001 Law i s far from being implemented systematically. As such, it is unable to take on the role which Law plays (theoretically) in a liberal democracy-that of establishing the meta-norm around which all other norms should harmonize. Instead, the Law becomes one of a number of competing norms and practices to which people may turn intheir struggles to secure land rights (CAS/WB 2006a). Where these different sets of norms-social, administrative and statutory-line up, there is little conflict and efforts at formalization run smoothly. Where, however, these sets of norms compete, conflict arises, with conflicting parties basing their claims on different normative orders. Competing claims emerge even where none appear to be `legal' inthe narrow sense of the word. The problem, from an equity perspective, with the emergence of these sorts of `competing (i1)legalities' i s that elites are often better able to `forum shop'-that is, select the normative framework which is most likely to legitimize their claims'. Do titles promote investment in land? One o f the key arguments in the case for systematic land titling is, as outlined above, that households will invest in productivity-enhancing capital improvements if they have the confidence that their land is not at risk of appropriation. This section first examines levels and patterns of household investment in agriculture, before then examining whether it is possible to trace a link betweenthis and possessionof a land title. Household investments in agriculture Amongst the poorest quintile, one third make substantial capital inputs intheir land to improve quality and productivity through terracing, construction o f ditches or bunds, planting perennial crops, or using better seeds, fertilizer, and pesticide. ("Substantial" is defined here as investment per hectare over the median average value.) Amongst the richest households, the proportion i s roughly twice as high. Irrigation may be taken as an indicator o f either long-term household-level capital investment in land (albeit a slightly less definitive one) and / or as willingness to pay for annual inputs (see Box 4.5). The disparity of access to irrigation among quintiles is apparent but not very significant: about 41 percent of households in the richest quintile, compared to 27 percent in the ' The "legal pluralism" seen in Cambodia is far from exceptional: see for example Wardell 2006 on how formal law provides a "significant, though not exclusive" reference point around which negotiationsover rightsand duties takes place in respectto forest law inAfrica; or Moser et a1 20010n interactionsbetween levels in a hierarchy of "rights regimes" (pp. 21- 3 1). 66 Addressing asset ineaualitv: a focus on land poorest quintile, irrigated at least half of their plots. This pattern of investments incapital inputs and irrigations mirrors the pattern of land tenure security as a result of some paper proof o f ownership (Table 4.4). Box 4.5 Interpreting household-level irrigation data The cost of irrigation has both public and private components, and capital and recurrent components. In Cambodia irrigation structures such as canals are largely built by Government, donors or NGOs; as such, whether or not a household is located in usableproximity to a source of water may have nothingto do with householdwealth or spending. (In fact, the causalitymay run the other direction: households fortunate enough to own irrigable land next to a water body are more likely to be non-poor.) However, gettingwater into fields often does require an additional household-levelinvestment in a water wheel or scoop (and the labor requiredto use it) or a pump (and the petrol requiredto run it). Ownership of water pumps is four times as high amongst the non-poor (12 percent) than amongst the poor (3 percent: World Bank 2006 p. 90, basedon CSES 2004). Similarly, while purchase of a pump could be taken as a one-off and long-terminvestment, fuel and maintenance (or hiring a pump) would be regarded as a recurrent cost. Householddata on use of irrigation may thus be taken as in part a capital investment (although, as a movable asset that can be rented out or sold ifthe owner cannot use it, a pump is not a very precise indicator of fixed investment in land); and in part as an indicator of ability and willingness to make recurrent investment. Table 4.4 Secure land tenure is positively correlated with land investments % rural householdswithin each quintile with Quintile of rural households All Poorest Next Middle Next Richest poorest richest % of operating area heldwith some paper proof of ownership within each quintile 40.9 50.8 55.0 63.1 59.8 54.5 capital spending per hectare greater than the average (median) level within each quintile 35.6 43.4 51.O 59.5 67.0 51.3 at least half of the operated plots irrigatedwithin each quintile 27.0 30.3 33.4 38.4 41.2 34.1 Source: CSES 2004 (15-monthsample) Two interesting findings emerge. First, the patterns are remarkably persistent in both measures of investment. Second, the differential likelihood of investment between the richest and poorest quartiles was only 30 percentagepoints. Specifically, the probability (67 percent) that a household in the richest quintile would invest at above-average rates was nearly twice the likelihood (36 percent) of a household in the poorest quartile. However, this is a much smaller gap considering that the consumption of the richest quintile i s 5 times that of the poorest quintile. The contrast with wealth-based differentials in household spending on health and education (in economic terms, investment in human capital) is striking: households in the richest quintile are found to spend on average 19 times as much as the poorest quintile on health care, and 25 times as muchon education (see Chapter 6 below). Sharing growth: equity and development in Cambodia 67 That investment in agriculture is only weakly related to the spending power of households i s somewhat counter-intuitive. One would expect that the marginal propensity to save and invest would be much greater among the rich, given the ratio in average consumption between the richest and poorest quartiles. On the other hand, this empirical puzzle is fully consistent with the hypothesis that secure land tenure is the primary incentive for land investments, regardless of one's wealth or income status. Using the variations across provinces, we find a significantly positive and robust relationship betweensecurity of land tenure and a number of investment measures. Thus, secure land tenure i s positively correlated with land investments. In provinces with greater proportion of households holding some paper proof of ownership which is synonymous with secure land tenure, investments in capital inputs and investments in irrigations are significantly higher. Figure 4.4 The relationship between investments and secure land titles - - I 0% 0 4 households with 100% plots of land held 20 40 60 5 0 1 with s o m form of paper proofs of ownership % households with20 0 10 100%plots of land owned 30 40 i / I I 4 ** 0 20 40 60 80 % households with at least 50% plots of land I held with s o m form of paper proofs of O% hou&%olds w% 50% %? m r e $fits of 50 ownership land owned with land titles Source: Cambodia Socio-economic Survey 2003-04 (15-month sample) A simulation of the potential impact of land titling In this section, we carry out a simple simulation exercise to estimate the impact of land titling on poverty. We only allow the impact to work through increased productivity in crop agriculture, and is based on the effects of titling on agricultural productivity. We assume that land title enhances yields, measured by value of output per hectare, by 73 percent than plots without titles. We further assume that titling affects only income from crop agriculture. Details of the mechanics of the simulations are presented in Annex 3. Table 4.5 shows the estimated impact of titling programs of varying coverage or size. We consider a few scenarios for the progress of the titling program. The current 68 Addressing asset ineatlalitv: a focus on land coverage or size of the LMAP is about titling 650,000 titles. The model estimatesthat the LMAP will lift about 280,000 people out of poverty, which is equivalent to reducing the national poverty rate by between one and two percentage points. Complete titling of 2.5 million would, according to the model, lift between a half and one million people out of poverty, reducing the national poverty rate by 4 to 8 percentage points. Naturally, these effects will not occur overnight. It i s likely to take several years for the effects of titling to work their way through investment and access to credit and trade, through productivity growth, to increased income and consumption. Table 4.5 Simulated effect of titling programs on poverty Number of titles Number of Number of Number of Change in national issued households households individuals poverty rate, affected escaping poverty escaping poverty percentage point changes ~~ 650,000 361,988 50,135 284,924 -2 19 1,000,000 556,904 77,131 438,345 -3.36 2,000,000 1,113,808 154,262 876,690 -6.73 2,500,000 1,392,260 192,828 1,095,863 -8.41 Source:CSES 2004 The model inthis simulation exercises is rather conservative (Le., providing the lower bound of the effects) and ignores many potentially important and positive effects of titling, e.g., multiplier effects, effects on other non-crop income, network and general equilibrium effects (see details inthe Annex). The current pace and coverage of the LMAP would have only titled on average about 125,000 titles a year, and attained titling of 26 percent of total plots in the country by the Project's completion year at the end of 2007. The potential productivity and income gains are sizable that warrants a serious consideration of expanding the systemic titling program more intensively and comprehensively into every part of the country. Nonetheless, to maximize gains for the poor, incentives of land cadastral must be aligned to map, register, and title even in remote or isolated areas. It is also important to resolve the issue that a large tract o f the land operatedby the poor is deemed state land or illegal settlement, and therefore, is excluded from the systemic titling program. Systematic titling appears likely to deliver significant benefits The Royal Government is currently engaged in a large-scale program of systematic land titling under the Land Management and Administration Project (LMAP), supported by the World Bank and other donors2. Having contributed to strengthened policy and regulatory frameworks and institutional development in the land sector, LMAP began to 2 LMAP is conceived as the first phase in the Government's 15-year Land Administration, Management and Distribution Program (LAMDP), which explicitly emphasizes equity goals in its three objectives("strengthen land tenure security and land markets, prevent and resolve land disputes; manage land and natural resources in an equitable, sustainable and efficient manner; promote landdistributionwith equity"). Sharing growth: eauity and develoument in Cambodia 69 distribute titles in 2004. In 2006, 300,000 titles were issued, bringing the cumulative total to almost 1 million. It is too early to assess the impact of this systematic titling approach: the program only started in 2004, and the full extent of behavioral changes i s likely to take more that two years to emerge. That said, it is possible to obtain some early insights into perceptions of impact, and perhaps some preliminary and suggestive indicators of emerging actual impact. The value of LMAP, and a more nuanced understanding of its likely effects indifferent places and over time, can be summarized inBox 4.6. Box 4.6 Systematic titling has positive effects, which can be expected to increase with time and with the extension of the program into more remote and less densely-populatedareas At this relatively early stage in systematic titling and on the basis of a quick and light study, it is possible to draw some preliminary conclusions with regard to impact to date, likely effects over the longer term, and changes needed to increase the benefits to poor households. The LMAP titling process functions well inthe villages it reaches, and appearsto confer clear benefits: . Titles are seen to reduce landconflict and increasetenure securityfor almost all respondents. I = i i ~ i e smaKe 11 easier to sell iana to Duyers rrom oursiae me local community, airnougn m e high tax for registrationis a major concern. Titles probablyhave a moderate, positive effect on landvalues. Some other aspects of post-titlingbehaviour are not as anticipated, and some of the other benefits often attributedto the creation of property rights are notyet apparent: It is uncommonto register titled landwith cadastralauthoritiesfollowing sale: rather villagers are sellingproperty by physically handingover the title under cover of contracts witnessed by I local authorities. The most important reason for the lack of registration is that households I . expect to pay a high and unpredictable tax for registering sales. It does not signal that households do not value titles or that titles do not affect behaviour. Even before titling and in areas still without titles, credit markets have been active, formal lenders play a significant role, and the practice of puttingdown land as collateral for loans is common, both with formal and informal lenders. Thus, receipt of title does not create credit markets where they did not exist before. However, credit is very expensive and titles appear to have a moderatebut definitely positiveeffect on access to credit at reasonableprices. With a few important exceptions, respondents do not yet perceive an effect of titles on agricultural production and welfare. However, given the impact on tenure security and conflict and the more minor but nonetheless positive influenceon credit and landmarkets, it is reasonable to expect these benefitsto feed through into positive effects on productionand welfare outcomes in the longer term. The decision to develop procedures for systematic titling in areas in which the process is least likely to be contentious makes good sense. As a result, however, at present the programseems to reach poor and conflict affected areas last, and leaves significant areas of household-operated, agricultural land untitled because it is claimed by the state. Fees for titling make it difficult for poor householdswith large plotsto participate inthe program. 70 Addressin1 asset inequality: a focuson land Public opinion is favorable As might be expected, there is general enthusiasm for landtitles. Villagers and local authorities perceive a number of benefits and limited disadvantages to land title (Figure 4.5). Although the numbers o f respondents are very small, these findings are at least suggestive o f how titles are perceived, both by those who have them and those who do not yet have them. Figure 4.5 The benefits of title are clearly seen to outweigh any problems a. % of respondents who mention the impact in response to the open question "Do you see any benefits of land titles?" (no prompts provided) 0 0 0 0 . . 1 Villagers in titled villages (n = 24) Villagers in non-titled villages (n = 13) B aVillage leaders (n 13) U0) 0 n In E! w- s0 Increased security Reduced conflict Access to credit Access to land of ownership market b. % of respondents who mention the impact in response to the open question "Do you see any disadvantages of land titles?" (no prompts provided) @aVillagers in titled villages (n = 24) aVillagers in non-titled villages (n = 13) Village leaders (n = 13) Increased cost of A land tax might be land sales irrposed Source: Markussen 2007 pp. 14-15 Beyonda general enthusiasmfor titles inprinciple, householdstllat hadreceivedtitles over through systematic titling were also very positive of the LMAP process. While there were minor grumbles about the accuracy of the boundary mapping in a few cases, there was no evidence of elite captureor corruption. Amongst the general public, too, perceptions of the benefits of a land title are overwhelmingly positive, with almost everyone expressing a belief that title conferred Sharing growth: equity and development in Cambodia 71 benefits (see Table ). This reflects opinion amongst both those who have already acquired title and those who have not. The primary concern is that the process of obtaining title was difficult and expensive. Given the numbers who perceived this problem, this perception probably reflects experience of previous titling schemes and the expectation of difficulties, rather than actual experience (direct or relayed by others) with the contemporary process of systematic titling. Table 4.6 Public perceptions of land titling are unambiguously positive Statement (read out to respondent) % respondents who agree It's better to have land titles than not 98.7 Land titles signifies ownership and nobody can take it away against hidher will 97.6 Land titles help secure loans or credit 98.5 Land titles can be passed on to your children 99.4 Land titles facilitate leasing out land 94.4 Land titles increase sale prices of land 97.6 Land titles are difficult and expensive to obtain 69.3 Land titles complicate sales or transactions 5.6 Source: IRL 2007 D. Titles are seen to improve tenure security and reduce the likelihood of conflict For most recipients, land title appears to further strengthen an already robust feeling of ownership security. For a minority of households, however, it may have a more significant effect in conferring security when before there were reasonsto fear that the land could be claimed by someone else (see Box 4.7). Many recipients emphasize that the titles would ensure that inthe future they can pass their land on as inheritence to their children. _- .- __I_ Box 4.7 Among recipients, titles are perceivedas improving tenure security Case 1: A rich household in Tuol Lveang village, Battambang, owns 9 ha of agricultural land. The land was acquiredalready in the early 1990s, and the household has constantlyworried that some of the plots would be confiscated by the state, because the holding is unusually large. After receivingtitles, the householdno longer worries. Case 2; A woman from Kandal recently bought land in Tek Thear village, Kompong Speu. She only lives there for a few months each year. She is on good terms with the villagers, but is still very keen to receive titles as soon as possible. Land prices in the area are going up, and she is worried that powerful outsiders may try to take her land. She expects that titles will protect her "even against powerful people, against whom localauthoritiesare ofno use". Case 3; A household in Sla village, Kompong Speu had a longstanding conflict over a plot of land with a cousin of the wife in the household. The conflict could not be resolved during the LMAP titling process, and has been sent to the national Conflict Resolution Committee. The ~ household is happy about the titles they received for other plots, and are keen to get a title for the plot in dispute. They expect titles to prevent such disputes inthe future. Source: Markussen2007 p. 16. 72 Addressing asset inequalig:afocus on land Conflicts may actually rise during the process o f systematic titling, as the mapping that precedes distribution o f titles may bring latent conflicts to the surface. However, the impression i s that (i) o f these appear to be resolved at the local level with assistance most from local officials and L M A P staff; and (ii) once titling i s completed, the level o f land conflicts has fallen to significantly below the level seen before titling. It i s too early to tell whether titles will prevent conflict in the long term as competition for land increases further, especially given ongoing weaknesses in the judicial system. At this point, however, titles are clearly seen as much stronger proof o f ownership, with the critical advantage that the titles specify plot boundaries in a way that other documents do not. Recipients perceive land titles to improve access to credit As mentioned above, credit markets do function in rural Cambodia and land can be used as collateral even without documents proving ownership. However, rural households interviewedperceived titles as improving the terms o f access to credit: and collateralized loans (on terms that are better than those for loans obtained without collateral) appear to have increased in the wake o f titling inthe villages he visited. Title may enable a household to obtain a loan from a formal lender, when in the absence o f a title the household might have been able to borrow only from an informal lender, at higher interest rates. Alternatively, title may enable households to borrow from an informal lender who might not have loaned money without a land title as collateral. Titles do appear to have advantages over existing documents Given that the benefits o f "title" derived from the CSES 2004 should probably be read as effects o f application receipts rather than fill titles3, this raises the question: if Cambodia already had a reasonably effective system o f "qua~i-formal~'property rights to land before the L M A P titling program, does it actually needthese new titles? It would be cheaper to issue documents similar to application receipts to those 52 percent o f plots which, according to CSES 2004, are not heldwith any paper. It seems likely that the new titles do in fact mark a significant improvement over earlier forms o f "ownership" document (and application receipts in particular) in several respects. Firstly, several households may have application receipts for the same plot. Secondly, application receipts do not specify plot boundaries and so will not help prevent or resolve boundary conflicts. Thirdly, in transactions with people from outside the local community, application receipts will sometimes not provide sufficiently strong documentation o f ownership for the buyer (although under some circumstances they will). Fourthly, in legal terms application receipts represent evidence but not proof o f ownership. This i s o f minor concern to most rural households, as using the courts to resolve disputes i s generally not considered a realistic option: nonetheless, case studies suggest that people perceive that application receipts have a weaker deterrent effect than titles against would-be land grabbers from outside the local community. Finally, it i s easier to obtain loans from formal lenders with titles than with receipts. By the end of 2004-the year in which the CSES was conducted-only one percent of all plots in Cambodia would have been covered by titles allocated through the systematic process. As such, the odds of LMAP-titled plots being recorded in CSES was extremely slim. Sharing growth: equity and development in Cambodia 73 These observations suggest that the older application receipts do not adequately prepare Cambodia for a future in which land market interactions between strangers (people not from the same community) will become steadily more important. Land titles o f the kind now being distributed through the systematic titling program do have the potential to fulfil this role (Markussen 2007 pp. 7-8). Trade-offs in process: pursuing equitable development through interim institutions To begin with, it needs to be recognized that land titling alone cannot be an adequate solution to the challenge o f promoting equitable, efficient land management that promotes pro-poor growth. On its own, titling could be argued to be inherently regressive, in that households benefit in proportion to the amount o f land they own, and the program has no direct effect onthe welfare ofthe landless. A number of pieces of evidence point to LMAP and the systematic titling process it supports having a positive impact in the areas it has covered to date. Such an approach, which sets out to title all land in a Commune at one time (hence "systematic"), seems to be clearly superior to earlier attempts to title land through "sporadic", on-demand titling, which suffered from capacity constraints and high fees, such that only a little land (predominantly that held by the rich) was titled. In that the landed poor tend to be the losers in conflicts over land ownership, providing them with strong titles i s likely to improve their ability to retain their key productive asset in the face o f increasing competition over land resources. A minor concern i s that the area-based formula for the calculation o f titling fees, while not excessive, may deter applications from poor households with large plots. However, the more important consideration i s whether the success o f the program so far in part reflects some deliberate choices regarding the process o f implementation. These choices have arguably limited the impact o f systematic titling on the most vulnerable landholders-and thus on equity-to date. In most Provinces, LMAP has started out inthe more long-settled, more productive and probably better-off rice-growing communities close to the provincial town; and seems likely to reach remote, recently- cleared areas, which are poorest and most prone to land conflicts, last. This sequencing reflects incentives in the administration o f the titling process: fees are paid to the titling teams based on the area o f the plot, leading to a preference to work first in areas with a large number o f small, easy-to-measure plots and a low incidence o f land conflicts (as titling can only proceed once conflicts are resolved). Adler et a1 argue that systematic titling has so far succeeded largely as a formalization o f reasonably well-established local understandings o f land ownership. In other words, it has succeeded in bringing local "legal" understanding and practice into alignment with formal, Government law through a process o f transparent and structured negotiation, in which inclusion o f the poor (Le. equitable process) i s supported by the provision o f information and processes to facilitate their participation. This approach makes sense from a long-term perspective that, informed by considerations o f the political economy o f land and the policy process, emphasizes the importance o f creating improved (transparent, rules-based) institutions for formalizing land rights. These institutions need to be given time to develop technical capacity, 74 Addressing asset inegualig: afocus on land independence and legitimacy. This i s best achieved by focusing first on the relatively more tractable, densely-populated core regions o f the country, before taking on the contentious areas on the forest frontier where widespread encroachment o f local populations, based on traditional concepts o f usufruct, are in direct contradiction to (i) state law that classifies these lands as "state public" or "state private" land and (ii) national and sub-national authorities' assumption o f the right to distribute these lands to commercial interests, generally on an informal, neo-patrimonial basis only tenuously rooted in the law. A somewhat sobering conclusion i s that reform initiatives that either consciously or unconsciously seek to create sustainable institutional improvements as a precondition for equitable development in the long term-the approach described by Adler et a1 in terms o f a focus on "interim institutional arrangements"-probably do not target the poor intheir initial stages. Some recommendations for building upon the gains achieved by L M A P to date include changing the fee policy so that it lowers the rate charged for poor quality land, reducing the cost barrier for poor households seeking to register large plots with low productivity. changing the compensation schedule for L M A P titling teams to create incentives for them to visit remoter, less densely populated areas with lower productivity (for example, by basing pay on the area titled as well as the number o f plots). And, engaging with the reclassification o f state lands in order for titling to benefit the poor living in remoter areas on the land frontier. A comprehensive policy framework for land management, and one that would serve growth and equity objectives simultaneously, i s therefore likely to require a complex process o f integrating titling and redistribution processes. Other policy recommendations would include to conduct more rigorous review o f the status o f the economic land concessions, so that those which are not in compliance with the law (because they are too large, or have not conducted environmental or social impact assessments, or overlap with a cancelled forest concession, or are not active) are cancelled and reallocated to alternative, productive uses; and, related to this, that Government and partners work together to identify ways o f putting the social land concessions principles into operation. 5. Addressing inequality in access to infrastructure: a focus on rural roads Summary Therehas been substantialprogress innarrowing the gap inaccess to roads, water, and reliable lighting, during 1997-2004, across the board. Despite the progress, considerabledisparities remain betweenurban and rural sectors, and betweenrichand poor within each sector. Isolation or geographic remoteness is a key factor inperpetuating poverty and income inequality. Provision of infrastructure such as roads improves connectivity, and helps reduce both poverty and inequality. Chapter 2 found that remoteness due to poor access to infrastructure such as rural roads is a correlate of poverty and hence a factor behind disparity and inequality. As explained in Chapter 2, one measure to lower inequality i s to reduce poverty, through raising income or consumption of the bottom segments of the population to close the income gap. This chapter focuses only on road investments. Roads are publicly provided in Cambodia. By helping to connect the isolated segments of the population, the Government can enhance the ruralpopulation's opportunities to participate inthe broader and more diverse economy. Roads integrate communities into regional and national networks of production and trades. Especially in rural areas, additional and improved roads reduce transaction costs, expand access to markets and enhance rural income. Investments inroads can also enhance the vitality (e.g., growth intrading and commercial activities) and expand diversity (e.g., growth in services) of the rural economy itself.It i s estimated that nearly two-thirds of African farmers are effectively insulated from national and world markets because of poor roads. In contrast, substantial investment in Indonesian roads throughout the previous 30 years enabled the poor to participate in the market economy (World Bank 2005 p. 169). Economic opportunities are shaped by access to infrastructure and especially roads (the focus of this chapter). As shown in Chapter 2, lack of access to roads is a factor contributing to poverty, and hence, inequality Access to roads also has a well-being dimension, i.e., a non-money metric aspect of living standard, as roads bring people closer to goods and services and improves one's daily existence. Furthermore, many of the ruralroads indeveloping countries are labor intensive public works, creatingjobs for unskilled labor. This Chapter first analyzes the distributions of infrastructure such as access to (i) market; (ii)roads; (iii)electricity; (iv) piped water; and (v) lighting, in order to understand the extent of disparity or inequality within urban and rural areas and between these two domains. The second section presents the distributions of access to (i) access 76 Sharinggmwth: equig and development in Cambodia markets; (ii)roads; (iii)piped water; and (iv) lighting for years 1997 and 2004 for comparable samples, based on the 1993 sampling frame to assess progress, if any, in closing the disparity between the rich and the poor, and between rural and urban areas. The third section shows the importance o f rural roads for economic outcomes, measured by total household revenue, net cultivation income or crop profit, non-agriculture revenue, and wage income. Distributions of infrastructure This section presents the distributions o f access to infrastructure for Cambodia's population in 2004, separately by regions and for rural and urban sectors, starting with a presentation of the distributions o f populations within various zones (regions) and sectors according to their positions or rankings inthe national consumption distribution. These distributions in Table 5.1 provide information about the relative welfare levels o f populations in various zones. For example, 87 percent o f Phnom Penh's population belongs to the top two richest quintile groups o f the national consumption distribution, whereas only 20 percent o f rural PlateadMountain population made it to the top two quintile groups. Between40 percent (of the urban PlateadMountain population) and 64 percent (of urban Plains) o f other urban residents belonged to the country's richest 2 quintile groups. The richest urban zone was Plains, followed by Coastal, Tonle Sap and Plateau/Mountain. Similarly in the rural areas, the richest zone was Plains, followed by Coastal rural areas, and the poorest rural zones were Tonle Sap and PlateadMountain. About 62 percent of rural PlateadMountain populations (c. 720,000) and half o f rural Tonle Sap (c. 1.7 million) belonged to the poorest 2 quintile groups in the national consumption distribution. The following Tables then presentthe proportions o f populations within each quintile with access to various public amenities and infrastructure. The population is ranked by its consumption level, within zone (region) and sector, into respective quintile groups in the zone and sector. Table 5.2 shows the distribution o f access to an improved water source, as definedby households with access to water pipes or public tap. Improved water i s widely available in Phnom Penh, and for over half of the richest quintile groups inurban Plains and urban Coastal regions. Even among the richest quintile inrural areas, such access i s extremely limited. 76 Addressing inequalipin access to infrastmcture: af0cuJ on mral mads 77 Table 5.1 Population distribution by national per capita consumption quintile and by region and region*sector Quintile Region 1 2 3 4 5 Totals Cambodia 20.0 20.0 20.0 20.0 20.0 100.0 Phnom Penh 2.5 2.7 7.5 21.1 66.1 100.0 Plains urban 5.1 10.0 21.2 24.8 38.9 100.0 Tonle Sap urban 16.2 14.2 17.3 20.1 32.2 100.0 Coastal urban 10.3 15.6 18.7 21.3 34.0 100.0 Plateau/mountains urban 14.9 19.6 24.2 19.9 21.4 100.0 Plains rural 16.9 22.0 22.4 23.3 15.5 100.0 Tonle Sap rural 28.3 22.4 20.8 16.3 12.1 100.0 Coastal rural 14.6 23.6 23.6 22.5 15.7 100.0 Plateaulmountains rural 38.0 24.1 16.6 11.6 9.6 100.0 Source: CSES 2004 (15-month sample). Table 5.2 Percentage of populationin dwelling with water piped or from public tap Quintile Region 1 2 3 4 5 Totals Cambodia 1.7 2.5 4.0 10.5 34.7 10.7 Phnom Penh 65.6 55.1 54.8 70.2 89.6 81.3 Plains urban 7.8 9.1 9.2 20.6 57.7 30.8 Tonle Sap urban 3.1 5.4 11.0 18.9 30.4 16.8 Coastal urban 0.0 10.7 16.6 28.2 48.6 27.3 Plateau/mountains urban 3.2 7.7 9.1 15.9 39.4 15.8 Plains rural 2.0 2.7 1.8 3.8 5.8 3.1 Tonle Sap rural 0.4 0.0 0.6 0.6 3.4 0.7 Coastal rural 0.0 0.0 0.0 1.6 4.3 1.0 Plateaulmountains rural 0.0 0.0 1.5 0.1 12.6 1.5 Source: CSES 2004 (15-month sample). Table 5.3 presents the distribution of population with access to battery- or generator- powered lighting in their dwellings. Almost everyone in Phnom Penh and among the richest quintiles in other urban areas has lighting. The proportions of populations with access to lighting fall monotonically as we move from richest to poorest quintile groups. While in Phnom Penh as much as 98 percent of its population had battery- or generator- powered lighting in their dwellings, only 46 percent of the urban PlateadMountain zone had such access; half inrural Plains; and a quarter inother rural zones. 78 Sharinggrowth: equzg and development in Cambodia Table 5.3 The distribution of population in dwellings with lighting, powered by city power, generator or battery Quintile Region 1 2 3 4 5 Totals Cambodia 18.4 30.3 41.7 56.6 80.3 45.5 Phnom Penh 93.5 86.4 91.8 97.4 99.2 97.8 Plains urban 49.7 60.5 82.5 82.6 98.2 84.8 Tonle Sap urban 12.1 34.0 52.8 77.4 94.3 61.8 Coastal urban 18.2 24.0 53.1 67.3 91.1 60.9 Plateaulmountains urban 9.8 25.6 40.7 59.7 81.6 45.7 Plains rural 29.1 40.1 49.8 58.9 72.6 49.8 Tonle Sap rural 9.2 17.5 25.4 35.5 61.2 25.0 Coastal rural 9.8 16.2 21.3 40.4 55.8 28.1 Plateaulmountains rural 14.8 22.9 26.8 35.2 59.4 25.4 Source: CSES 2004 (15-month sample). There is a similar pattern in access to electricity (Table 5.4). In Phnom Penh and urban Plains, about 80 percent o f their populations have access to electricity. Incontrast, only 10 percent o f rural PlateadMountain region, and between 18 and 25 percent of populations in rural Plains, Tonle Sap and Coastal areas had access. Within each zone and sector, the proportions o f populations with access declines monotonically as one moves from the richer to the poorer quintile groups. For example, only a third o f the poorest quintile in urban Tonle Sap population had access to electricity compared to 93 percent o f the richest quintile in the same urban zone. The disparity i s just as wide between the richest and the poorest quintile group in other zones with lower average access rate. For example, 6 percent of the poorest quintile group, compared to 30 percent o f the richest quintile group inrural PlateadMountain region had electricity access. 78 Addressin2 inequality in access to infrastructure: a focus on rural mads 79 Table 5.4 The distribution of population with access to electricity (%) Quintile Region 1 2 3 4 5 Totals Cambodia 13.6 18.8 24.8 36.0 57.9 30.2 Phnom Penh 89.4 85.2 79.3 84.0 83.3 83.4 Plains urban 56.9 61.5 72.3 80.2 94.2 80.9 Tonle Sap urban 32.5 64.4 71.7 86.1 92.7 73.8 Coastal urban 66.5 57.0 71.2 72.6 96.0 77.2 Plateaulmountains urban 32.5 31.7 45.0 52.4 69.5 47.3 Plains rural 9.6 13.2 15.9 23.1 29.6 18.0 Tonle Sap rural 13.9 20.3 24.0 32.0 45.5 24.2 Coastal rural 12.5 17.3 18.1 32.6 45.0 24.6 Plateaulrnountains rural 6.3 6.6 10.0 9.4 29.6 9.6 Source: CSES 2004 (15-month sample). Table 5.5 presents the average distances to the nearest all-weather road. In Phnom Penh and urban Tonle Sap, just about every village has all-weather roads, and hence the average i s zero km in every quintile group. Except for PlateadMountain, all other urban zones have good access to all-weather road, averaging 1.5 km for Plains and Coastal. In urban areas, average distance to roads falls monotonically from poorer to richer quartiles. The average distances to roads were several times higher among the rural populations. The distance differential between the richest and the poorest in rural areas was rather small and in rural Tonle Sap and Coastal regions, the richest quartiles had on average greater distance to travel to the nearest roads than the poorest quartiles. Table 5.5 Average distance to nearest all-weather road (in km) Quintile Region 1 2 3 4 5 Totals Cambodia 4.82 3.55 3.21 3.09 2.03 3.34 Phnom Penh 0.00 0.00 0.00 0.01 0.03 0.02 Plains urban 4.94 2.06 2.64 1.24 0.24 1.41 Tonle Sap urban 0.18 0.47 0.33 0.01 0.01 0.16 Coastal urban 3.32 2.62 2.34 1.54 0.34 1.62 Plateaulmountains urban 14.83 3.55 2.75 0.29 0.36 3.85 Plains rural 6.12 3.66 3.75 3.31 1.86 3.75 Tonle Sap rural 2.46 3.41 3.05 4.86 6.58 3.68 Coastal rural 3.40 4.70 3.58 3.82 8.58 4.71 Plateaulmountains rural 8.20 4.37 3.42 3.60 1.24 5.31 Source: CSES 2004 (15-month sample). With regards to market access, Table 5.6 presents the average distance to the nearest permanent markets, which i s approximately 1 km in Phnom Penh, and 2km in the urban Plains, Tonle Sap and Coastal regions. The urban PlateadMountain region i s more 80 Sharinggrowth: egtrity and development in Cambodia isolated, averaging 6km, with a huge gap betweenthe richest and poorest quintile groups. The poorest quintile group has to travel as far as 12km, compared to 3 km for the richest quintile group, in urban PlateadMountain region, to reach the nearest market. On average, distances are several folds longer in rural areas, even among the richest quintiles. Table 5.6 Average distance to permanentmarket (in km) Quintile Region 1 2 3 4 5 Totals Cambodia 10.86 9.54 8.07 7.13 4.22 7.97 Phnorn Penh 0.68 1.40 1.51 1.32 0.77 0.96 Plains urban 5.19 4.23 4.17 2.76 1.10 2.67 Tonle Sap urban 4.54 3.28 2.55 1.45 0.80 2.21 Coastal urban 5.27 3.75 2.65 1.84 0.38 2.15 Plateau/mountains urban 12.26 5.27 7.32 4.32 3.28 6.19 Plains rural 9.40 8.84 7.77 7.19 5.73 7.85 Tonle Sap rural 12.27 10.32 8.50 9.36 7.14 9.95 Coastal rural 6.04 6.89 7.00 7.26 8.38 7.11 Plateau/mountains rural 14.33 16.61 17.52 17.17 13.43 15.66 Source: CSES 2004 (15-month sample). Access to improved water, electricity, lighting, roads, and markets are distributed relatively equally distributed among residents in Phnom Penh Consumption-based inequalities inaccess to infrastructure appears to be much lower within the capital than in other realms. In other urban areas, access to these infrastructure and amenities is more limited than in the capital city but nonetheless significantly better than inrural areas. In rural areas, some amenities (such as water from an improved source) are non-existent even to the better off households. With the exception of Phnom Penh, in every zone and sector, access to infrastructure and amenities increases monotonically from poorest to richest quintiles. Thus, the disparities between rich and poor, especially in rural areas, and inequality between urban and rural households in accesses to infrastructure, utilities, and public amenities are substantial. Substantial progress in closing the rich-poor gap Despite large disparities remaining, Cambodia had made noticeable progress in closing gaps in access to infrastructure. For example, investments in roads have shortened significantly the average distance to the nearest motor road. Table 5.7 shows the average distance to motor roads by quartile groups in rural and urban areas in years 1997 and 2004. In rural areas, the poorest quartile group in 1997 on average traveled 2.8km to get to the nearest motor road. By 2004, the average distance was less than 1 km. In urban areas, there was also improvement. The poorest quartile group in 1997 80 Addressin2 inequality in access to infrastmcture: a focus on mral mads 81 traveled on average 0.53km to get to the nearest motor road, and by 2004, the distance was halved. While there is absolute progress in better access to roads across the distribution, the rich-poor gap i s actually now larger. It is unclear whether this reflects a bias from the changes of the definition of roads (Le., from all-motor road in 1997 to all- weather road in2004). Table 5.7 Average distance (in km) to motor roads by quartile, 1997 and 2004 Rural poorest quartile second quartile third quartile richest quartile 0-25% 25%-50% 50%-75% 75%-100% 1997 2.810 2.949 4.085 2.625 2004 0.875 0.738 0.499 0.378 Urban poorest quartile second quartile third quartile richest quartile 0-25% 25%-50% 50%-75% 75%-100% 1997 0.535 0.581 0.301 0.129 2004 0.204 0.095 0.008 0.026 Sources: CSES 1997; CSES 2004 (15-month sample) Note: there is no informationcollected for motor roads in 2004, and thus the above estimates of 2004 are derived from informationcollectedfor all-weather roads. AI-weather-road is a more expensive and more durable than all-motor roads. Table 5.8 presents the distributions of utilities and other public amenities, e.g., markets, improved water, and reliable lighting for years 1997 and 2004. The results suggest that there have been remarkable improvements for all income groups, in both ruraland urban areas. Especially inrural areas, the increases inaccess to these amenities and utilities were most substantial among the poorest quartile and thus the disparities between rich and poor have shrunk during this period. Among the poorest quartile in rural areas, the proportions of population with access to a permanent market in their villages rose 3-fold from 3.4 percent to 9.1 percent; those with access to improved water jumped 20-fold from 0.1 percent to 2.2 percent; while the percentages with access to reliable lighting increased 4-fold from 4.7 percent to 18 percent. With increases in the numbersof permanentmarkets in villages, the average distance to market also improved during this period. In rural areas, the average distance fell from 7.13km in 1997 to 6.87km in 2004. In urban areas, the average distance improved from 2.26km in 1997 to 1.45km in2004. 82 Sham'nggrowth:eqtrip and developmentin Cambodia Table 5.8 Distributions of utilities and amenities by quartile, 1997 and 2004 Rural Quartile Groups poorest second third richest 0-25% 25%-50% 50%-75% 75%-100% %with a permanent market 1997 3.4 7.1 9.8 10.6 in the village 2004 9.1 8.5 9.8 15.2 distance to the nearest 1997 8.36 7.45 7.08 5.65 permanent market (km) 2004 8.55 8.24 6.85 5.77 %with clean water 1997 0.1 1.7 5.3 2.7 (piped or public tap) 2004 2.2 1.5 1.9 5.1 %with reliable lighting 1997 4.7 7.8 11.9 9.5 (by city grid, battery, generator) 2004 18.2 29.0 39.3 55.9 Quartile Groups Urban poorest second third richest 0-25% 25%-50% 50%-75% 75%-I00% %with a permanent market 1997 16.9 22.3 27.2 31.7 in the village 2004 16.1 25.7 29.3 29.9 distance to the nearest 1997 3.26 2.81 1.65 1.31 permanent market (km) 2004 2.78 1.38 0.77 1.41 %with clean water 1997 13.4 26.4 45.4 68.1 (piped or public tap) 2004 19.5 45.1 67.4 78.1 %with reliable lighting 1997 35.8 56.0 75.0 87.2 Sources: CSES 1997; CSES 2004 (15-month sample) Note: Only comparable samples, from both surveys, based on the identical (1993) sampling frame are used in the analyses. The most remarkable public investments involved the construction o f schools. Across the country, increased access to schools, at all levels, has been notable between 1997 and 2004. Table 5.9 shows that the distances to primary school and secondary schools have fallen precipitously, by bringing schools closer to the populations. Inrural areas, the average distance to the nearest primary school fell from 5.6km in 1997 to Ikm in2004; and the average distances to lower and upper secondary school fell 10-fold from 49km to 4km, and from 118km to l O k m , respectively. Similarly in urban areas, the average distances have fallen about 9-10 fold as well. In 2004, there was little disparity in distances to primary school betweenthe richest and poorest quintile groups in many regions of the country. For example, the average distance to the nearest primary school within rural areas in the PlateadMountain zone was 2.5km for the poorest quintile, and was 2,Okm for the richest quintile. In urban areas within the coastal region, the distance was 0.52km for the poorest quintile, and was 0.50km for the richest quintile (Table 5.10). 82 Addressinm inequality in access to infiastmcture: a focus on mral roads 83 Table 5.9 Average distances (in km) to schools, 1997 and 2004 Rural distance toschool nearest primary distance to nearest lower distance to nearest upper secondary school secondary school 1997 6.59 49.0 118 2004 1.02 3.99 9.55 Urban distance to nearest primary distance to nearest lower distance to nearest upper school secondary school secondary school 1997 4.72 19.5 28.5 2004 0.47 1.73 3.01 Table 5.10 Average distances to primaryschools, 2004 Region Next Poorest poorest Middle richest Next Richest Cambodia 2.25 1.79 1.94 1.60 0.93 Phnom Penh 0.35 0.32 0.42 0.39 0.54 Urban 0.46 0.42 0.55 0.43 0.48 Plains urban 0.34 0.53 0.33 0.44 0.45 Tonle Sap urban 0.45 0.31 0.32 0.34 0.22 Coastal urban 0.52 0.73 1.15 0.38 0.50 Plateau/mountains urban 0.43 0.29 0.40 0.39 0.26 Rural Plains rural 1.22 1.22 1.73 1.04 0.67 Tonle Sap rural 3.23 1.80 2.17 2.09 2.38 Coastal rural 5.05 2.99 3.58 5.80 1.49 Plateauhountains rural 2.49 4.59 3.69 4.73 1.96 Sources: CSES 1997; CSES 2004 (15-month sample) Disparity of outcomes from the disparity of access to roads Chapter 2 showed that there are noticeable differences in welfare due to differential access to infrastructure. Mean real consumption in remote areas was lower than that in connected areas, and poverty rate was higher in isolated villages than better connected villages. This section presents the differences in additional economic outcomes between rural households with access to all-weather roads and those without. We focus on rural areas because despite significant progress in closing the gap during the last decade, the rural majority are still considerably less well-connected, compared to the urban population. In addition to consumption differences which were observed in Chapter 2, there are also significant differences in other outcomes. Table 5.1 1 presents provincial averages o f annual total household revenue, crop profits, non-agriculture revenue, and wage income for 2 categories o f households. One category consists o f households with all- weather road inthe village. The other category o f households do not have an all-weather 84 Sharinggrowth: equity and developmentin Cambodia road in the village. Except for Kompong Chnang and Pursat, the means of total household revenue of all other provinces are significantly higher' among households connected to an all-weather road in the village. An examination of profit from crop cultivation shows that mean averages are not necessarily higher among the connected households in every province. About half of the provinces have higher average crop profits among households without all-weather road in the village than the average of households with all-weather road inthe village. The bottom half of Table 5.1 1 presents provincial averages of non-agriculture revenue and wage income. With the exception of Kompong Chnang and Pursat, all other provinces have significantly higher average non- agriculture revenue among householdsconnected to a road. Similarly, with the exception of Kompong Speu, wage incomes on average are significantly higher among connected households inall provinces. The results suggest that access to infrastructure such as roads raises household income primarily through the provision of better access to non-farm opportunities. Total household revenue is higher among those with access to roads than those without. However, revenue from cultivation or crop profitability is not systematically higher among those with better road access. It would be expected that roads facilitate access to input and output markets, lowering costs and enhancing revenue. However, there i s no systematic pattern of higher agriculture revenue linked to access to roads. On the other hand, access to roads improves access to other non-farm opportunities, e.g., facilitating small enterprise businesses, as seen with systematically higher non-agriculture revenue among those with access to roads. Lastly, access to road facilitates connectivity to wider employment opportunities in other villages or in urban areas. Waged income among households with access to roads i s significantly higher. 1 The significance refers to 1percent level or 99 percent confidence interval. 84 Addressing ineaualitv in access to infrastructure: a focus on mral roads 85 Table 5.11 Provincial averages of total household revenue, crop profits, non- agriculture revenue, and wage income, by existence of all-weather road in village, (annual, in Riel) average total household revenue average profit of crop cultivation connected to without differences connected without differences road road in mean to road road in mean Kornpong Cham 7,057,633 3,477,119 3,580,514 568,475 284,463 284,012 Kandal 9,412,447 3,811,877 5,600,570 236,998 -241,174 478,172 Prey Veng 3,829,271 3,106,895 722,376 406,470 145,524 260,946 Svay Rieng 2,904,427 2,696,480 207,947 263,736 918,992 -655,256 Takeo 5,481,193 4,629,359 851,834 474,940 -281,745 756,685 Banteay Meanchey 8,674,462 3,447,005 5,227,457 560,648 363,837 196,811 Battambang 11,594,940 6,112,980 5,481,960 1,407,561 1,572,887 -165,326 Kompong Thorn 4,918,124 3,056,538 1,861,586 476,226 606,672 -130,446 Siem Reap 12,598,708 2,863,314 9,735,394 462,508 286,844 175,664 Kompong Chhnang, 7,447,693 9,409,218 -1,961,525 372,531 388,189 -15,658 Pursat Kampot 5,166,382 2,334,716 2,831,666 550,960 282,502 268,458 Sihanoukville, Kep, Koh 22,189,763 8,698,686 13,491,077 270,592 334,010 -63,418 Kong Kompong Speu 4,030,935 3,109,317 921,618 419,710 568,778 -149,068 Other Plateau Mountain 5,064,413 2,687,803 2,376,610 562,839 633,388 -70,549 provinces average non-agriculture revenue average household wage income connected to without differences connected without differences road road in mean to road road in mean Kompong Cham 4,091,138 1,100,222 2,990,916 796,789 641,101 155,688 Kandal 5,098,629 1,136,973 3,961,656 1,910,707 1,440,174 470,533 Prey Veng 1,441,I 1,096,272 57 344,885 475,443 412,523 62,920 Svay Rieng 773,121 208,697 564,424 526,925 223,327 303,598 Takeo 21740,933 2,426,512 314,421 615,540 479,726 135,814 Banteay Meanchey 5,229,444 1,489,450 3,739,994 1,268,491 375,283 893,208 Battambang 7,071,883 2,110,211 4,961,672 1,119,217 407,162 712,055 Kompong Thom 2,224,397 913,975 1,310,422 840,040 348,801 491,239 Siem Reap 8,639,408 602,423 8,036,985 1,256,086 299,704 956,382 Kompong Chhnang, 4,881,231 5,161,625 -280,394 520,959 374,913 146,046 Pursat Karnpot 2,242,902 486,869 1,756,033 535,698 101,203 434,495 Sihanoukville/Kep/Koh 15,474,791 4,272,164 11,202,627 2,841,474 595,305 2,246,169 Kong Kompong Speu 1,549,583 282,366 1,267,217 678,726 754,396 -75,670 Other PlateaulMountains 2,375,261 463,780 1,911,481 571,756 186,098 385,658 Sources: CSES 1997;CSES 2004 (15-month sample) Figure 5.1 plots the provincial average of distance to the nearest all-weather road and provincial proportion of households with heads working in the secondary sector. The plot shows that there i s a negative relationship between the distance to an all-weather road and the proportion of household heads being engaged in a secondary sector of employment. Thus, with closer proximity to roads, one has lower transportation costs and greater ease to access other economic opportunities beyond agriculture and outside o f his village. 86 Sbaringgrowtb: equip and developmentin Cambodia Figure 5.1 Access to road and employment in the secondary sector I. c s 0 0 1 2 & average distance to roads (km) Sources: CSES 2004 (15-month sample) 86 6. Promoting equity through service delivery Summary 1 . Education and good health are important intrinsically (as components in the definition o f a I I good standard of living) and instrumentally, in facilitating economic development. For I individuals, households and nations, there are strong connections between (i)human development (or human capital) in the form of knowledge, skills and health; and (ii) economic development, in terms o f opportunities, incomes, living standards and growth. These causal relationships are two-way: skills and health affect income and consumption, which in turn affect access to education and healthcare. Such mutually-reinforcing ties can work in either poverty-reducing or poverty-creating, poverty-perpetuating directions. A virtuous spiral occurs when healthy, educated individuals are better able to secure good jobs or succeed in profitable self-employment; higher, more stable incomes in turn enable them to live better, save and, in turn, invest in nutrition, healthcare and their children's education. Many households, however, become caught in a downward spiral into poverty. Low and variable income and a hand-to-mouth existence make it more likely that the poor will become ill(incurring extra costs, debts, and possible loss ofproductiveassets); and less likely that their children will be able to build up the skills and knowledge they need to diversify into new activities or obtain higher wages in non-agricultural employment. 8 Given that skills, knowledge and health are critical determinants o f individual opportunities, policies that shape equality or inequality in delivery o f basic services (schooling, healthcare, water and sanitation) obviously play a major role in shaping the degree o f equality or inequality of economic opportunity within society. 8 Despite recent gains, pronounced inequalities still exist in educational opportunities and outcomes. The strong two-way relationship between education and living standards emerges from both the CSES 2004 and qualitative research: this link is critical in explaining how inequality of opportunity, and poverty / wealth outcomes, are now reproduced from one , generation to the next. 8 Poor householdsthat depend on hard physical labor and lack adequateaccess to food or clean water are more likely to fall illand less able to afford early, effective treatment. I11 health in turn creates or reinforces poverty: the poor depend more than most on manual labor, and the illness of an economically active household member implies reduced production and income as well as increased costs when treatment is finally sought. The economic impact o f illness includes the distress sale o f productive assets (e.g. land) and / or taking on interest-bearing . debts. It is clearly a disadvantage for poor individuals and families that, because they suffer ' unnecessary ill health and cannot obtain an education that would enable them to make use of their innate talents, they have limited opportunity to participate in productive and profitable economic activity. However, it is also a problem for the economy as a whole. An ' uneducatedworkforce vulnerable to illhealth is not productive or competitive. Cambodia is seeking to make the transition from a dualistic economy (in which narrow sub-sectors of I manufacturing and services-Le. garments and tourism-coexist with a large mass o f ' subsistence-focusedsmall farmers) to a diversified, robust, modern economy. To achieve this transformation, it will need to dramatically improve the health and education o f the ' population. I 88 Sharinggrowth: eqtlig and developmentin Cambodia Human development, equity and growth Human development-taken to mean, broadly, good outcomes in education and health-has both an intrinsic and an instrumentalvalue (Anand and Sen 1994; Sen 1999). Philosophically, education and health are generally seen as desirable as ends in their own right and important components in any definition of a good standard of living. The case is clearest with regard to health-no-one would choose to suffer illhealth, disability or early death when these can be avoided-but most people accept that education expands an individual's ability to participate in society and enjoy a broaderrange of experiences. Equally, however, good outcomes in education and healthare a means to an end: they constitute human capital, improving the contribution of labor to productivity and economic growth. The knowledge and skills that are conferred by education (starting with literacy and numeracy) set a child up for greater or lesser opportunities in the labor market in adulthood. Meanwhile, there is a well-known mutually-reinforcing relationship between poverty and illness which makes the poor more likely to fall ill,less likely to be able to manage the economic consequences of that illness, and more likely to become further impoverished as a result. These interactions between human development and household wealth reproduce from one generation to the next: particularly striking is the influencethat the level o f educationof a mother appears to have on her child's chances of staying healthy, receiving an education, and accessing opportunities. In Cambodia, this link between parents' socio-economic status and the unequalhealthand educationoftheir children is particularly acute because households account for an unusually high proportion oftotal spending on healthand education. Defining equity in human development: service delivery and outcomes In looking at the concept of equity in human development, it helps to distinguish between (i) equity in access to social services, in which different groups enjoy the same levels o f access, utilization and financing of education and healthcare services; and (ii) equity in human development outcomes themselves, defined as the absence of systematic and potentially remediable differences in health or education across population groups (men and women, rich and poor, urbanand rural, and so on). This distinction has implications for policy responses to problems o f equity in human development. Whereas equity in outcomes is a multi-factorial issue, the various dimensions of equity in access to services can be addressed through reform and developmentof servicedelivery systems. Health and wealth Good health status is an important component of development, firstly as an end in its own right. Enjoying good health and avoiding unnecessary pain and suffering is an intrinsic element in defining an individual's quality of life. As such, equity in health, defined as equality of health achievements (and correspondingcapabilities and freedoms) is central to concepts of social justice. In many countries, this i s seen as sufficient justification for a wide range of social policies designed to ensure equality o f access to Promoting equity through sewice delivery 89 essential preventative and curative healthcare (Sen 2002). This view of the intrinsic value of health, and the corresponding obligations of governments to ensure provision of health services to their citizens, is reflected in concepts of rights to health, and in the inclusion of healthtargets in the MDGs and national development strategies. However, good health is also critically important in shaping individual and household opportunities and welfare; and, by extension, as a factor enabling or constraining a nation's economic development. A healthy population is more productive than one that suffers from high rates of preventative illness, unnecessary expenditure on low-quality healthcare, and early death. Healthy children learn better and have better hture earning potential in adulthood; amongst adults, days lost to disease result in reducedearnings and diversion of scarce income to healthcare. In Cambodia, as in many societies in which there is limited protection against catastrophic out-of-pocketcosts to treat illness, people often find themselves in trap in which poor health makes it difficult to escape poverty, and poverty makes it difficult to obtain the diet and services requiredto improvehealth. Internationally, for many years the focus of health policy debates has been on improving overall efficiency and achieving aggregate targets. Recently, however, attention is once againreturningto healthequity as an objective o f health system reforms, recognizing that targets couched in terms of improving averages may mask increasing divergence between the outcomes of the rich and poor. This concernwith equity reflects increasing recognition of important intr-sectoral links between health, education and monetary (consumption, income or asset) aspects of wellbeing and development. Cambodia has in many ways been a leader in policy experiments (for example, health equity funds and the contractingout of service delivery) that are consciously designed to achieve equity as well as efficiency objectives. Health outcomes are poor, but improving The health status of Cambodians remains far below that of their more prosperous neighbors in Thailand, Malaysia, Vietnam or Indonesia. However, when data from the 2005 Cambodia Demographic and Health Survey (CDHS) is compared to data from the 2000 CHDS, it is clear that considerable progress has been made over a five-year span (Figure 6.1). Contact with professional healthcare services during pregnancy and childbirth has expanded dramatically. Far more children now receivethe full set of basic vaccinations by the time they are two years old'. Willingness to seek treatment when ill or injured seems to have risen modestly; and, as importantly, the cost of healthcare (average spend over last 30 days by those seeking treatment) has fallen by aquarter. The health practices of households have also improved over the last five years, particularly with regard to early childhood feeding practices (most notably in the importance attached to breastfeeding). Three in four babies aged less than two months are now exclusively breastfed, compared to just one in six in 2000. This has almost certainly contributed to the fall in child malnutrition and rise in child survival. However, malnutrition rates increase once solid and semisolidfoods are introduced at six months of age, probably due to poor feeding practicesand childhood illnesses. I The diseases covered by basic vaccinations are TB (one BCG injection); diphtheria, whooping cough and tetanus (three PT injections); polio (three injections); and measles. 90 Sharilzggrowth: eqzlity and development in Cambodia Figure 6.1 Availability and use of healthcare have both expanded significantly a. antenatal care b. immunization of children -+-Received2or 54 more doses of tetanus toxoid 67 injection during 44 last pregnancy 3 --a- Last birth - delivered at 30 health facility (public or private) Last birth assisted by -_____ trained v personnel 2000 2005 (doctor, nurse 2000 2005 or rridw ife) c. utilization of healthcare d. cost of treatment Care $20.7 CI TransDort U J " 2000 2005 % E a, c .1 .r 69 1st treatment 2nd treatn-ent ~3 3rd treatment E 2000 2005 Source CDHS 2000,2005 Reflecting these improvements in service delivery (and in average incomes, which have expanded food security and access to care), a number of health and nutrition indicatorshave shown rapid gains, includingchild nutrition (see Box 6.2 for definitions), anemia amongst women, and rates of some of the commoner childhood illnesses(Figure 6.2, Box 6.1). Promoting equity throughservice delively 91 Figure 6.2 While some health outcomes have improved, others have not a. numerous health indicators-especially b. for child nutrition-have improved ... ...while others-notably the maternal mortality rate-remain static -m- fever in last 2 w eeks years) (% children 4) - 45 1; -stunting % l (% children e 5) -0-diarrheainlast 2 weeks Syrrptom of acute (% children 4) u- respiratory infection s (YOchildren 5) -x- w omen below 145cmin height (%women 15- 49 years) +NMR c (deaths I 100,000live 2000 2005 births) Source: CDHS 2000 (pp. 117,155,158,174,177,181):CDHS 2005 (pp. 120,156,158,159,168,185) Box 6.1: Rural healthcare has improved in many respects Health conditions in this village have improved over the past ten years...The number of patients has decreased and TB is almost forgotten due to vaccination programs and training provided since 1993. Villagers now have access to the health centre in Kampong Preah Commune [ 2 h away] and to private doctors in the village...In 1993, hospitals were very far from the village. (Battambang, MOPS) People in the village have had health education about sanitation and clean drinking water. Moreover, children under 5 years of age have been immunized against diseases and pregnant women have also been given injections and education about how to take care of their babies. These services are provided by the health centre staff every month, and they have also educated local midwives to send women to hospital immediately when they have problems during delivery. (Female-headedhousehold, Kompong Champrovince, PPA) Source: CDRI 2007a, 2007b, forthcoming. Box 6.2 Technical note on selected health indicators Stunting. Below-average height-for-age reflects growth defects due to cumulative effects o f long- term chronic malnutrition (inadequate calorie intake) and recurrent and chronic childhood illness. Wasting. Wasting or markedly below average weight-for-height is a measure o f acute current malnutrition, reflecting the inability o f a child to obtain adequatenutrition inthe period preceding the survey, due to inadequatecalories andor a recent episode o f illness resultinginweight loss. Underweight. Low weight-for-age (WFA) i s a composite indicator reflecting both chronic and acute malnutrition: underweight children are stunted, or wasted, or both. Maternal mortality ratio (MMR). The MMR is derived by dividing age-standardized maternal mortality rates (maternal deaths per woman per year aged 15-49) by the age-standardizedgeneral fertility rate (annual births per thousand women aged 15-49). Inthe CDHS, this rate i s calculated for the period 0-6 years before the survey (Le., for 1999-2005). 92 Sharinggrowth: equity and developmentin Cambodia Finally, there has been remarkable success in reversingthe rise in HIV/AIDS. There i s a good consensus that HIV incidence i s now at or below one percent. Transmission through high-risk behavior has been tackled with considerable success: a new challenge is to address transmission from husband to wife (now the major mode of transmission) and from mothers to children (responsible for one third of new infections). Cambodia has also achieved considerable success with ARV treatment, and has committed to expandingthis provision, includingto HIV-positive pregnantwomen. Other health outcomes, however, remain stubbornly bad (Figure 6.2 above). Rates of fever and diarrhea amongst children under five years have not changed; the proportion of women under 145 cm tall (an indicator o f poor nutrition during childhood and adolescence, and a risk factor in giving birth) appears to have gone up. Most notably, there appears to have been no improvement in the maternal mortality ratio (MMR)2. This is out of line with improvements in other indices (e.g. falling fertility; rising proportions of deliveries that occur in health facilities or with the help o f trained personnel; and the increase in antenatal visits), all of which imply that maternal deaths should be declining. This disjuncture may be partly due to the limits o f the measure. As the MMR averages results over the preceding seven years, it is not very sensitiveto change: it takes several years of safer births to shift the MMR. Nonetheless, it seems safe to say that, at the very least, MMR has not registered the kind of improvement seen in other indicators, and that this remains a problem and priority for the Ministry of Health. The lack o f change seen to date is partly because achieving safe pregnancy and childbirth depends, to a greater extent than is the case for other aggregate health indicators, on the quality dimension of health services. To reduce the MMR significantly requires that complications are appropriately diagnosed and treated; which, in turn, depends on staff skills, diagnostic tools and treatment possibilities, and a properly functioning referral system. Dramatic improvement has however occurred in other aspects of mortality. Since 2000, infant mortality rates (IMR) and under-five mortality rates (U5MR) have fallen by about a third. Although the methodology is problematic given the large gaps in the demographic data, careful analysis of the 2004 CambodiaInter-CensalPopulationSurvey (CIPS) suggests that life expectancy at birth has risendramatically(Figure 6.3). However, there are sizeable inequalities in health care and outcomes Within the overall picture of low-but-improving health outcomes, there exist very substantialdifferencesbetweendifferent groups in Cambodian society in terms of access to services (water, sanitationand healthcare) and resultinghealth status. Individuals who are poor or born of motherswith little or no educationare far more likely to be ill,and to die beforetheir fifth birthday. There are also pronouncedgeographical differencesin the patterns o f illness and death, not only between the rural and urban populations, but also between Provinces. 2 The slight deterioration in MMR-from 437 to 472 deaths per 100,000 live births-is not statistically significant: the margin of error is quite large (* 134 in2005). Promoting equip through service delively 93 Figure 6.3 Children born now are more likely to live to adulthood and middle age a. child survival rates have improved by b. estimated life expectancy about a third has risen by around 9 years -E 751 150 1 65 124.4 50. 52- (t9.0) +Under-five mortality rate (t7.5) 25 - * --eInfantmortalityrate `0 D E Q) (deaths before 1 year old) p! .- Source: CDHS 2000 and 2005; ClPS 2005. Geographical inequalities remain considerable Differences in health performance between urban and rural populations, and between Provinces, remain large, particularly for reproductive health (Table 6.1). Above- and below-averageprovincial health performancerecorded in the CDHS 2005 correlate fairly well ifnot perfectly with living standards and proportion of the populationliving over the poverty line (as measured in the CSES 2004). PhnomPenh, with a poverty headcount far below the national average and a much greater density of both public and private healthcare facilities, shows clearly superior performance on almost all indicators. The coastal provinces and municipalities (Kompong Som, Kampot and Kep), however, do rather less well than might be expected for their (fairly good) poverty status, perhaps becauseo f Koh Kong's weak spatial integrationwith the rest of Cambodia. There is some variation between health status and living standards at the bottom end of the ranking, too. Amongst the provinces for which statistically significant CSES 2004 poverty estimates are available, the poorest (Kompong Speu) has, as would be expected, the worst outcomes in terms of child survival rates, female anemia, and babies delivered outside a facility or without trained assistance3. However, the third-poorest province- Siem Reap-registers rates of health-seeking behavior, unmet need for family planning, vaccination, and child malnutrition significantly worse than those in Kompong Speu. Poor female educational status may be part of the explanation for Siem Reap's worse- than-expected performance. The worst health outcomes are found in the upland north-eastern provinces (Mondulkiri and Rattanakiri), where terrain and low population density result in poor physical access, exacerbated by low education and language barriers for the indigenous ethnic minority groups here. These provinces are known to be extremely poor but the absolute numbers are very small, making it impossible to produce statistically reliable poverty estimates. 94 Sharinggrowth: equity and developmentin Cambodia Table 6.1 Health outcomes in rich provinces and municipalities are often twice as good as those in poor areas a. use of services and child health Province I Municipality Health Child health service utilization** IMR UBMR Vaccination stunting coverage ' (moderate or severe) Phnom Penh (richest) 96 42 52 81 27 Kandal (second richest) 94 85 101 79 37 Sihanoukville I Koh Kong (third richest)' 95 88 104 65 53 Siem Reap (third poorest) 82 67 94 43 73 Kompong Thom (second poorest) 87 87 106 55 53 Kompong Speu (poorest) 90 107 122 81 43 Mondolkiri / Rattankiri 76 122 165 35 84 Rural 91 92 111 66 52 Urban 94 65 76 69 40 Total 92 66 83 67 50 Ratio, Phnom Penh (richest) to Kompong Speu (poorest) 1.I 0.4 0.4 1 .o 0.6 Notes: T S E S aggregates Sihanoukville, Kep and Koh Kong; CDHS aggregates Sihanoukville and Koh Kong, linking Kep with Kampot ** % of individuals ill or injured in last 30 days who sought treatment ' % of children 12-23months with all basic childhood vaccinations b. female nutritional status and reproductive health Province / Municipality Female malnutrition Reproductive health anemia in unmet need births at a ANC from delivery women for family facility trained staff assisted by planning trained staff Phnom Penh (richest) 29 16 78 85 86 Kandal (second richest) 46 23 37 78 74 Sihanoukville I Koh Kong (third richest)' 46 20 29 58 57 Siem Reap (third poorest) 56 30 20 69 29 Kompong Thom (second poorest) 57 19 9 59 25 Kompong Speu (poorest) 58 24 9 59 23 Mondolkiri / Rattankiri 43 29 9 28 14 Rural 48 26 17 68 39 Urban 38 22 50 79 70 Total 47 25 22 69 44 Ratio, Phnom Penh to Kompong Speu 0.5 0.7 8.7 1.4 3.8 Notes: 'CSES aggregates Sihanoukville, Kep and Koh Kong; CDHS aggregates Sihanoukville and Koh Kong, linking Kep with Kampot " doctor, nurse or midwife Source: CDHS 2005: province wealth definitions from CSES 2004, analyzed in Knowles 2005a p. 41 , ..but most are closing over time In most key measures, comparison of the 2000 and 2005 CDHS surveys suggests that geographical inequalities in health outcomes have become less pronounced over time. This finding is interestingas a proxy for trends in wealth-based healthinequalities: as it is not possible to disaggregate trends by wealth quintile (as this disaggregation was not better base in urban areas are trends in wealth-based inequalities can be widening existing inequalities obtained from comparisonof spatial data. Thus for example, the gap betweenthe best and worst A: Infant mortality rate performing provinces in terms of child survival (under one year) has narrowed as poorer, remoter provinces have caught up with the gains made earlier in urban areas and more accessible lowland parts of the countryside: Mondolkiri and Rattanakiri posted 40 percent and 39 percent declines in infant and child mortality respectively, while indicators held steady in Phnom Penh. While Pursat experiencedthe second highest rate of all early 50 childhood mortality indicators in 2000, the CDHS 2005 shows that, although still relatively high, infant and under-five mortality rates have 2000 2005 fallen markedly (by more than a third)4. Kompong Thom, where infant and under-five 0ratio:ruralchildrendyingforevery mortality both rose between 2000 and 2005 fl 100 urban children dying (from 65 to 87 and from 99 to 106 respectively), -t-Ruraldeathrate(deathsper1,000 was perhaps a case o f health indicators coming live births) into line with consumption poverty (Kompong -+-Urban death rate (deaths per 1,000 live births) Thom is the second poorest of those provinces for which estimates are available). B: Under-five The picture of narrowing inequalities is not, mortality rate however, universal. Inequalities in child survival between the rural and urban 150 3 146 populations are widening as improvements occur in both, but occur more rapidly-and starting from a much better initial level-in the 100 4 urban centers (Figure 6.4). The very poor Kompong Speu and Preah Vihear / Stung Treng experienced an increase in under-five deaths (from 90 to 122and from 120to 146). 50 Inequalities in wealth give rise to inequalities in health 0 Li 7 2000 2005 Thepoor are more likely fall ill Source: CDHS 2000 p. 125;CDHS 2005 p 126 Central to the poverty trap is a causal relationship between consumption poverty and poor health. The poor are more likely than the richto suffer injury or illness for a variety o f reasons, including poor nutrition; inadequate access to clean water and sanitation; 4 Changes in the way provinces were aggregated between CDHS 2000 and CDHS 2005 may give an under-estimateof 2005 ratios. 96 Sharinggrowth: equip and developmentin Cambodia hazardous occupations; low rates of contact with preventative health services; and, in some cases, lesser knowledgeabout good healthpractice. To begin with, the poor consume less food than other groups. In 2004, 20 percent of the Cambodian population lived below the food poverty line: by definition, this bottom quintile consumed less than the daily 2,100 calories per capita taken to define the minimum intake required for healthy nutrition. This shows up in the rates of child malnutrition. Almost half of all children from the poorest quintile, and 43 percent of children from the second-poorest quintile, showed signs of moderate or severe chronic malnutrition in the form of stunting (Figure 6.5). Poverty significantly increases the risk o f malnutrition; and malnutrition significantly increases vulnerability to illness'. Figure 6.5: Child malnutrition amongst the poor is roughly twice that of the rich Richest Severe pModerate 3 .-3 Next richest s5f Mddle Next poorest Poorest 47 % of children in group suffering stunting (low HFA) Rchest evere ca Moderate 3 =5f Next richest 5 Mddle Next poorest Poorest 43 % of underweight children in group (low WFA) Rchest 0Severe Moderate $ .-C3 Next richest 55 Mddle Next poorest 9 Poorest a % of children in group suffering wasting (low WFH) Source:CDHS 2005 p. 168 CDHS 2005 suggests a stronger correlation between living standards and malnutrition than did the 2004 CSES (World Bank 2006 pp. 113-115). As there are reasons to believe that the CDHS recorded anthropometric data more accurately, these are used in this report. Promoting equig through service deLive9 97 However, anthropometric measures of malnutrition such as stunting and wasting are not only an outcome o f inadequate calorie intake that makes children vulnerable to illness: they are also themselves a reflection of the effects of chronic or recurrent childhood illnesses such as diarrhea. To a degree, then, malnutrition rates among the children of poor families reflect not only that poor children suffer from inadequate quantity and quality o f food, but also the fact that they are much less likely to have access to clean water and sanitation (Figure 6.6). Figure 6.6: The poor are at far greater risk of water-related disease a. Only the rich have acess to piped water 36 Rped water Well, or other improved source 48 Total Unprotected well Other unimproved 33 2 ~ % of quintile I total b. Less than five percent of the poorest have access to safe sanitation Sewerage 23 Water based, 35 a Richest20% improved Total Poorest20% Unimproved or none 95 4 % of quintileI total Unclean water and unsafe disposal of excrement makes children (and adults) much more vulnerable to disease, most notably water-borne illnesses (infectious diarrhea, intestinal parasites and hepatitis) but also respiratory and skin infections. (Extending the perspectiveslightly, the lack of "appropriate" water supply also tends to increase the risk of malariaand dengue: water stored in open wells or unsealedjars increases the breeding grounds for mosquitoes.) 98 Sharinggrowth: eguig and developmentin Cambodia Occupational differences between poor and non-poor also have health implications. The poor work hard in physicaljobs involving greater risk o f accidents and exposure to certain forms of disease. A significant proportion of the rural poor work in peripheral, forested areas where the risk of malaria is endemic (see Box 6.5 below). The poor have less contact with preventativehealth services that might protect them against common illnesses. The best measure of this is probablythat of antenatal care and vaccination rates amongst children (Figure 6.7). Note that these particular forms of preventativehealthcare are in fact likely to underestimateaccess to other health services given that these services are provided free, cost is not likely to be a major barrier, but remoteness is. Figure 6.7 Children from rich families are significantly more likely to be covered by preventative health services Children born in richer households receive ...andare better protected ~ t t e rpreventative care from before they are born ... by vaccinations by age two m Poorest m Next poorest 81 Middle 77 76 Next richest o Richest Mother received Mother took iron Infantfully All basic No vaccinations no antenatalcare supplements protected against vaccinations during pregnancy during pregnancy neonataltetanus Source: CDHS 2005. Finally, the poor are distinguished from the rich in certain aspects o f knowledge, attitudesand behaviorwith regardto health. On some issues-for example, regardingthe cause and prevention of HIV/AIDS transmission-the differences are relatively small (and, when it comes to behavior, wealthier males are actually more likely to engage in commercial sex). Broadly speaking, however, wealthier and more educated groups in Cambodian society have (as would be expected) better knowledge of the causes of ill- health, and are more likely to reflect this knowledge in their behavior, than are the less educated poor. One key behavioral factor predisposing poor families to illness relates to household hygiene practices (Figure 6.8). Whereas epidemics of diarrheal diseases are clearly related to unclean drinking water, the transmission route of endemic diarrheal diseases (particularly in poor communities) may depend heavily on domestic hygiene, and particularly washingof food and hands (see Box 6.3). Promoting equip through service deliver 99 Box 6.3 Sanitation and hygiene may be as important as drinking water quality The second edition of Disease Control Priorities in Developing Countries (2006) notes five types o f evidence to support the view that most endemic diarrheal disease is transmitted by water- washed routes and is not waterborne: Negative health impact studies. Esrey and Habicht (1985) and Esrey et a1 (1991) cite studies in which better water quality has not had a significant effect on diarrhealdisease incidence. Food microbiology. Studies of the microbiology of foods in developingcountries- particularly the weaning foods fed to children inthe age group most susceptibleto diarrheal disease - have shown such food to be far more heavily contaminatedwith fecal bacteriathan is drinking water (Lanata 2003), even when the water is storedinopen pots. Seasonality of diarrhea. In countries with a seasonal range in temperature, bacterial diarrheas peak in the warm season (and viral diarrheas in winter). This suggests that bacterial pathogens show environmentalregrowth at some stage intheir transmissionroute, implying that they have anutritional substrate. Water is thus a less likely vehicle than food. Fly-control studies. Trials in rural Asia and Africa have shown that fly control can reduce diarrheal disease incidenceby 23 percent (Chavasseand others 1999). Hand-washing studies. A recent review of the effect o f handwashingwith soap has shown that this simple measure is associated with a reduction of 43 percent in diarrheal disease and 48 percent in diarrheas with the more life-threatening etiologies (Curtis and Cairncross2003). These five types o f evidence suggest that it is not drinking water quality but domestic hygiene- particularly washing food and hands-that i s the principal determinant o f endemic diarrheal disease rates. Tackling these channels o f disease transmission requires (i)access to a sufficient quantity of water; (ii)access to a toilet or latrine; and (iii)good household hygiene behavior, influencedby health education. Source: Cairncross and Valdmanis (2006) Water Supply, Sanitation and Hygiene Promotion pp. 777-778 inJamison et a1(2006) Disease control priorities indevelopingcountries: Second edition . - Figure 6.8: The rich practice better household There is also a significant hygiene than the poor difference with regard to - % of householdspracticing safe smoking. The CSES 2004 found (contained)disposal of children'sstools that smoking is more prevalent in rural areas; and lower in urban areas, among richer, more Rchest a7 educated groups. (It is also an Next richest 65 overwhelmingly male habit: 42 Mddle 54 percent o f men were daily Next poorest 51 smokers compared to 4 percent Poorest 42 numbers o f women.) As demographic and epidemiologic % of mthers in quintile w ah child 5 years living with her transition proceeds in Cambodia, smoking-related non- Notes: - safe ("contained") disposal of children's stools is taken to communicable diseases (NCDs) encompass the answers "Child used toilet or latrine"; "PuVrinsed into toilet or latrine"; or "Buried". are likely to increase among the - Wealth quintiles based on factor analysis of asset data poor (World Bank 2006). Source: CDHS 2005 p. 164. Significantly, the proportion o f household consumption diverted for tobacco consumption is twice as high in the poorest 20 percent o f households as the richest 20 percent. Amongst the poorest quintile, the amount spent on tobacco was higher than on health care (8.7 percent compared to 6.6 percent). IO0 Sharinggrowth: equip and developmentin Cambodia All of these factors associated with poverty-food insecurity and malnutrition, livelihoods which entail higher risk of accident or injury, significantly lower access to clean water and safe sanitation, and lack of knowledge about the health risks of smoking or unhygienic sanitary practices-make it much more likely that the poor will fall illand need medical treatment. Proving this association with simple questions about the incidence of ill-health, however, is harder than might initially be assumed: analyzing the answers to such questions, as often used in large multi-purpose household surveys, often implies that the rich suffer worse healththan the poor (Box 6.4). Box 6.4 Poverty and illness in the CSES Proving a statistical association between income/consumptionpoverty and poor health on the basis of self-reportedhealth status is often difficult. In most low-income countries, household surveys find, counter-intuitively,that richer andor better-educatedindividualshave more health problems than do poorer/less educated individuals. This pattern seems to be borne out in Cambodia: although slightly more rich than poor report their long-term health as good for their age (which is to be expected), the rich also report more recent illness (20 percent of the richest quintile report a health problem over the four weeks preceding the survey, compared to 15 percent o f the poorest quintile). This over-reporting of illness-in-general by richer groups is usually explainedby (i) the likelihoodthat the rich andbetter educatedare more likely to identify poor health as a problem, rather than accept it as an inevitable feature of their daily life; and (ii) the possibilitythat morefrequent contact with healthproviders (76 obtainedtreatment for a health problem in the last four weeks, comparedto 56 percentofthe poorest)makerichand/or educated individualsmoreaware ofhealthproblems(StraussandThomas 1998). Source: Knowles 2005b pp. 83-4. That the rich suffer more illness than the poor is hard to Box 6.5 Rural Cambodians perceive, and have believe. Firstly, there are explanations for, worse health strong a priori reasons to outcomes amongst poor households assume that the poor are more The PPA in particular found that some common illness vulnerable to illness, for all the (notably malaria, dengue and typhoid) are closely reasons given above (poorer associatedwith poverty andthe poor): access to clean water and The rich are rarely sick because they have enoughfood sanitation, lower rates of to eat and stay in good health... they complete a lot of vaccination, malnutrition, etc.). non-physical tasks and send their children to school for better education., (female youth, Dang Kda, Kampong Secondly, accounts and Thom, MOPS) explanations from qualitative Unlike the rich, the poor work hard in hazardous and research strongly support the tough environments, do not have enoughfood to eat and logical case that the poor, not so frequently fall sick. (FGG, Krasaing, Battambang: the rich, are more vulnerable to MOPS) disease and accidents (Box All the poor and destitute villagers get malaria because 6.5). To a significant degree, they are bitten by daikol mosquitoes when they go to the the explanations provided forest to find firewood and wild fiuit or when they go match those of researchers and fishing. They also drink unboiled water which flows health specialists: the poor downfiom the mountain. (MixedFGD in Khla Kropeau, experience worse health Pursat,PPA) because they do not eat enough Source: CDRI 2007a, 2007b (forthcoming) Promoting eqzrig tbrotlghservice delive7y 101 and because they work Figure 6.9: From birth onwards, poor children suffer at physically strenuous significantly worse health outcomes- tasks in locations including survival rates characterized by endemic disease risk. I m borest E~ e xpoorest t Middle gNextrichest Richest I Finally, examining CDHS data on the Children born in the poorest 20 percent of households are more likely to be born dangerously undersized.... reporting o f symptoms 13 of common childhood ailments such as diarrhea and acute respiratory infection and recordable health outcomes shows that, as expected, the poor Very small Sn-eller than average suffer more than the ...tobe malnourishedduring their first five years ... rich (Figure 6.9). m To begin with, children born into the bottomtwo quintiles are somewhat more likely to be born underweight (less than 2.5 kg) and considerably more Stunted Wasted Vnderweight likely to be reported as undersized. Small ...to suffer from common and serious childhood illnesses ... and/or underweight status at birth is a major 41 determinant o f infant and child health and mortality: in the CDHS, children who were reported as having been "very small" or sm ARI synptom Diarrhea Diarrhea w Rh Nbderateor "smaller than average" blood severe anenia were three times as ...andto die before age five. likely to die within the first month, and 70 percent more likely to 100 die before their first fpI birthday (NIPH et a1 c0 50 2006 p. 127). % Amongst children in -f" o hfant mrtality rate Lhder-fivemrtallty rate the bottom two wealth (deaths before 1year (total deaths before five quintiles, the combined old) years old) Source: CDHS2005pp. 126,152.i56,158,159,167, 179 102 Sharinggrowth: equig and developmentin Cambodia effects of higher incidence of low weight or size at birth, less good initial breastfeeding, less good infant and young child feeding practices, poorer householdwater supply and sanitation and lower rates of immunization results in much higher incidence of malnutrition and childhood illnesses, relative to the children of rich households(NIPH et a1 2006 pp. 169-70; 154; 176-7; Rosenboom et al p. 14). Children from the poorest quintile are 60 percent more likely to have diarrhea; four times more likely to show symptoms of acute respiratory infection (AN); and over five times as likely to suffer from bloody diarrhea (a symptom o f potentially very serious intestinal problems). In terms of nutrition, the children of the poorest quintile are 2.4 times more likely to suffer from chronic malnutrition (stuntedheight); and twice as likely to havemoderateor severe anemia (iron deficiency). Critically, mal-nutrition and illness reinforce each other. Inadequate quantity and quality of food intake makes a child more vulnerable to illnesses, many of which (e.g. diarrheal diseases or TB) in turn impede nutrition uptake, resulting in more pronounced malnutrition, and increased susceptibility to disease. The net outcome is, not surprisingly, much higher death rates amongst the children of poor parents compared to those whose parents are rich. This relationshipbetweenparentalwealth and the prospects of a baby surviving are very pronounced: compared to a child born into the richest quintile, a child born into the lowest quintile is three times as likely to die before either age one or age five. If poverty-induced illness and malnutrition interact to perpetuate each other within the life of an individual child, the interaction of mother's living standards and health, weight and size of children at birth, vulnerability to childhood disease, and childhood malnutrition explain how living standards and physical health interact to transmit ill- health (and poverty) from one generation to the next. As noted above, children with low birthweights are more likely to suffer childhood diseases, which fbrther stunt growth; when small, unhealthy girls grow up into under-sized mothers, they are more likely to give birth to low birth-weight children, perpetuating the cycle. For this reason it often takes several generations of improved nutrition and healthcare to achieve significant improvement in birthweights and subsequent growth from childhood into adulthood. When ill, the poor are less able-and thus less likely-to obtain treatment Ifthe poor are more likely to fall ill,they are also less ableto obtainaccess to quality treatment. The main reason is the most obvious: the poor are less able to afford the costs of medical treatment. However, other factors, including distance from facilities and indirect costs of seeking treatment (i.e. the opportunity costs for the time of the patient and carers) also militate against the poor obtaining access to good-quality curative services. Public expenditure on health care is not notably pro-poor. The 2006 health sector public expenditure review report concludedthat although only 32 percent of total public health expenditures are spent on providing basic health servicesat healthcentre (HC) and referral hospital (RH) level. Partially offsetting this bias towards central and tertiary spending, the allocation of the HC and RH budget does correlate broadly with geographicaldistribution of poverty. Promoting equig through service deiivey 103 More importantly, however, the total level of public financing for health is simply too low. Given this, the impact of the pro-poor pattern in the allocation of primary (HC and RH) healthcare in tackling inequitiesin healthand healthutilization is not clear. In 2006, the total budgetallocation to healthwas 6.9 percentof plannedGovernmentexpenditures. Although budget allocations have risen in absolute terms over time, they are still very low (US$4 per capita per annum in 2006), and have declined somewhat as a share of Government public expenditures in recentyears. Poor execution of the healthbudget has been problematic for many years and has exacerbated difficulties in using public spending to improve access to health for the poor. However, budget execution has recently started to improve. Whereas in 2004 only 83 percent of the operating budget was disbursedby the end ofthe calendaryear, this increasedto 91 percent in 2006. Given the low level of public financing of health services, the cost of health care in Cambodia falls primarily upon the patient, and is relatively expensive. This is true of public health services as well as private care: because the public health service is desperately under-funded,obtaining service requirespatientsto pay high fees at the point of service. Two recent large-scalenational sample surveys-namely the 2004 CSES and the 2005 CDHS-both measured out-of-pocket spending (OOPS) on health6. The CSES estimates that out-of-pocket spending for health care amount to US$l5.48 per capita per annum; the CDHS comes to a higher estimate (US$24.9). This might be explained by differences in how the surveys collected data; and / or by an increase in health utilization driven by falling unit costs (see Figure 6.ld above). Figure 6.10 Frequency and size of out-of-pocket spending on health, 2004 Poorest 0 Next poorest Mddle I!! Next richest CJ Richest @I Average 25 1 s 15 10 5 0 incidence of health OOPS health OOPS relative to high relative OOPS high absolute OOPS (% of individuals reporting other spending (% of persons reporting (% of persons reporting some out-of-pocket heaith (% of total p.c. nonfood OOPS that exceed 50% of OOPS >= 100,000 Riel spending in last 4 weeks) consumption) per capita nonfood during the past 4 weeks) consumption) Source: CSES 2004 analyzed by Knowles 2005 6 At present, disaggregation of health spending by wealth is only possible using the CSES dataset. Although a number of indicatorsfrom the CDHS have been brokendown by wealth quintile (defined in this case by an asset index), this has not yet been done for household spending on health. As such, investigation of differences in health spending by wealth quintiles, and what this might say about the equality of access to health services, will be based on the CSES data. However, for other lines of analysis-e.g. by rural-urbansplit, or by Province-CDHS data will be used, as it often containsadditionalvaluable detail. I04 Sharinggrowth: equip and developmentin Cambodia As Figure 6.10 illustrates, about one in five households surveyed in the 2004 CSES had incurred out-of-pocket spending (OOPS) in the four weeks preceding the survey interview. As expected, the incidence of recent spending on health was lower amongst the poorest quintile (12 percent) than the richest (20 percent). Interestingly, the size of out-of-pocket health expenditures by wealthier quintiles (amongst those who reported spending) was higher not only in absolute terms but also relative to their total spending, suggesting that healthcare (i)remains an important priority as households obtain more disposableincome and (ii)constitutesa significant expense for all groups. Indeed, the richest quintile spends a striking amount, in both absolute terms and relative to the spending by other groups (Figure 6.11). Mean OOPS amongst the richest is almost 19 times that of the poorest quintile; over three times the mean for the population as whole; and 3.6 times higher than mean spending amongst the second- richest quintile. To put it in context, however, this sudden jump in average spending between the bottom four quintiles and the top (richest) quintile simply mirrors the distribution of living standardsthemselves (as measuredby total per capita consumption). Figure 6.11 The richest quintile spends far more on healthcare than do other groups: but this reflects a similar gap in overall living standards total consumption out-of-pocket health spending (mean riels p.c. per day, 199314 constant (riels p.c. in last 4 weeks) prices) 14,857r 6,151 r average, all quintiles: average, all 2,585 r quintiles: 4,761 r 2,604 r + s - 4,178 r 3 - 2 493 r 5 1463 r E 800 r Poorest Next Middle Next Richest Poorest Next Mddle Next Rchest poorest richest poorest richest Source:CSES 2004 analyzed in Knowles 2005. These high point-of-service costs help to explain the 16 percentage point gap between the poor and the rich in seeking treatment for injury or illness (Table 6.2). They also help to explain the dramatic difference in rates of hospitalization (Figure 6.12), which is typically far more expensivethan outpatienttreatment. However, these estimates of out-of-pocket spendingalso suggest that health spending by households, while very high, has fallen by a third over five years: the 2000 DHS reportedmean out-of-pocket health expenditureover the last 30 days at US$24 per capita (NIPH et a1 2005 p. 37). The fall has come mainly in rural areas (down from US$20 in Promoting equip throughservice dehey 105 Table6.2 There is a large gap between the poor and 2000 to US$14 in the rich in rates of health service utilization 2005), while costs in urban areas have fallen Quintile % ofcare person reported illor In need of health only slightly (from in last 4 weeks who sought care US$26 to $24: the Urban Rural Total decline came mainly Poorest 59 59 59 from a fall in transport Next poorest 63 62 62 Middle 60 64 64 costs, with treatment Next richest 70 69 69 costs static). This Richest 81 72 75 implies an encouraging Total 73 65 66 improvement in the Source: CSES 2004 (15-monthsample), analyzed in Knowles2006 p. 62. affordability of rural health care, probably Figure 6.12 There are substantial wealth-based reflecting the combined differences in use of hospital treatment effects of improvements in rural roads, improved 1 coverage by and reduced costs of public health facilities7, and perhaps also increased number of private providers and better availability of medicine (Box 6.6), reducing average transport costs Urban Rural Total and exerting some downward competitive 1rn Poorest 111 Next poorest Mddle Next richest Richest ea TotalI pressure on prices. Source, CSES 2004 (15-monthsample),analyzed in Knowles 2005 p 69 Box 6.6 Private providers have helped to expand access to modern healthcare Although not without problems (e.g. over- or mis-prescription of medicines), the spread of private providers has also playedan importantroleinexpandingaccessto modernhealthcare. We can say that health and treatment services have improved over the past ten years ... The reason is that we now have medicines available,,. f w e have some money we can buy them. Ten years ago, we had neither medical practitioners nor hospitals. At that time we had to spend our rice or gold in exchange for treatment or medicines...Now we have medical practitioners providing services in the village. These changes happened since the second national election. (FGD amongst those who had moved out o f poverty, Krasaing Village, Battambang: MOPS) Source:CDRI2007b (forthcoming) 7 j In 2005, public sector treatment ($15.5 over last 30 days) was cheaper than private sector ($18.6):this is a reverseofthe situation in 2000 ($31 and$29, respectively). 106 sharinggrowth: equig and developmentin Cambodia Previously, out-of-pocket spending was estimated to account for 86 percent of total per capita health expenditure: in the light of recent findings, this may have fallen to about 50 percent. This fall in cost-plus rising average household income and per capita consumption-helps explain the rising trend in health utilization: in 2005, 92 percent of those ill or injured over the last 30 days sought treatment, continuing a gradual upward trend (from 86 percent in 1998 and 89 percent in 2000: NIPHet a12006 p. 32). Financial barriers to the poor are reinforced by the associationbetween poverty and remoteness, which contributes to lower-than average contact rates between the poor and preventative health services, and adds transport costs to the costs of seeking care. Since the mid-I990s, the Government has attempted to ensure geographical equity in the allocation of health services through the Health Coverage Plan (HCP). This attempts to ensure that there is one health centre to serve every 10,000 people; and one Operational District (OD) with a Referral Hospital (RH) for every 100,000 people*. Although significant progress has been made towards implementing the HCP, spatial inequalities remain (Table 6.3), in part because of residual difficulties in expanding coverage to low- density remoterural areas; and in part because between-ODdifferences in rates of natural increase and internal migration have changed the distribution of the population, with the resultthat some ODs now contain morethan 100,000, while others contain less. Table 6.3 The poor have to travel considerably further to obtain healthcare - mean distance (km) to nearest health provider,by consumptionquintile Provider Per capita consumption quintile Total Poorest Next Middle Next Richest poorest richest Public Health center 7.6 6.8 7.0 5.8 6.0 6.6 Referralldistrict hospital 15.9 14.0 12.8 11.8 9.2 12.8 Provincial hospital 41.8 40.3 37.5 34.6 26.0 36.0 National hospital 149.5 129.9 122.7 105.0 87.3 118.9 Private Private hospital 93.4 73.5 64.7 56.3 41.6 65.9 Private clinic 17.1 14.1 12.6 11.6 7.6 12.6 Doctor 18.6 15.5 14.5 11.9 7.6 13.6 Nurse 7.6 6.7 5.7 5.2 3.5 5.8 Trained midwife 6.7 6.4 5.7 5.4 3.6 5.6 Dedicated drug shop 14.5 11.9 10.0 8.6 5.3 10.1 Other shop selling drugs 7.4 6.3 5.5 5.0 3.2 5.5 Traditional birth attendant 1.I 1.5 1.8 2.1 3.4 2.0 Kru Khmer 2.4 2.9 2.9 3.4 3.6 3.0 Other traditional healer 4.4 5.7 5.8 7.2 5.5 5.7 Source CSES 2004 (15-month sample), analyzed in Knowles 2006 p. 71-2. The most significant form of spatial inequality may now lie at the fine-grained level, beyond (or below) the remit ofthe HCP, in regardto how near or far householdslive with regardto their nearest healthcenter. The Ministry of HealthPublic ExpenditureTracking Survey (PETS) concluded that health centers provided effective services to the 57% of the catchment population living within a 4.5 km radius of a health center. Beyond this, however, utilization rates drop dramatically (World Bank and RGC 2007 forthcoming pp. * As o f 2005, there were 966 health centers and 69 hospitals (MoH 2006 p. 1). Promoting equip tbrougb service delively 107 100-102). Given that the poor are more likely to live beyondthis 4.5 km watershed, they are less likely to enjoy benefits from services delivered through the centers. The CSES seems to confirm this, finding that the poor are more likely to live in villages where distance rather than the cost of treatment is seen as a key barrier to healthcare (World Bank 2006 pp. 125-6). It is hard to obtain an easy measure of quality of care, but an effort to cross-match CSES data with data from the MoH database of public health facilities suggests that poor groups in Cambodian society also enjoy considerably more limited quality of public health care provision, as measured by the availability of health professionals and some specialist services (Table 6.4). Table 6.4 There are notable inequalities in the quality of public healthcare Quality indicator Poorest Richest Average Ratio, richest to poorest Health Center (HC) Total staff per facility 8 13 9 1.6 Number of regular staff per facility 6 11 8 1.7 % facilities with at least one medical doctor 5 25 11 5.0 Number of midwives per facility 2 3 2 1.6 Number of beds per facility 4 5 4 1.1 Referral Hospital (RH) Total number of staff 106 137 111 1.3 Number of regular staff 85 118 92 1.4 Number of doctors 12 20 14 1.7 Number of midwives 13 18 14 1.4 % hospitals providing endoscopy services 11 42 20 3.7 Source: CSES 2004 (15-month sample, population weighted) and December 2003 database of health facilities compiled by MOH and NIS; analyzed in Knowles 2005 p. 77. In terms of perceptions, the CDHS confirms that both cost and distance are (i)seen as serious problems by all groups but (ii) seen as particularly severe by the poor. By contrast, there was much less inter-quintile variation in those reporting quality o f care issues as a problem: the level of concern about the lack o f skilled staff or drugs was similarly high across all wealth groups, with the exception of the richest, who expressed significantly less (though still considerable)concern (Figure 6.13). This probably reflects confidence in their ability to deploy their much higher level of health spendingto obtain access to providersand drugs. 108 Sharinggrowth: equzty and development in Cambodia Figure 6.13 For the poor, cost and distance are the main barriers to healthcare a. Cost and distance act as bairiers for the poor v) .-0 L b ) - 8 Gettingmoney for Distance to health Havingto take Not wanting to go treatmnt facility transport alone rn Poorest Next poorest QI Mddle B Next richest Richest Total b. By contrast, concerns about quality of care were the same for all-except the richest Concern no female Concern no provider Concern no drugs Gettingperrrission to provider available available available go for treatment 1rn Poorest IINext poorest Mddle Next richest Richest QI Total 1 Source: CDHS 2005, analyzed in NlPHet ai 2006 p. 149 Strikingly, there are significant variations between Provinces in both cost and distance and quality concerns as problems obtaining access to healthcare: Banteay Meanchey, Preah Vihear and Stung Treng all register very poor ratings across all indicators (Figure 6.14). To summarize: at first sight, it appears that the burden of OOPS is lower among the poor than the rich. However, this is most likely because the poor simply do not use health services: and ifthey do, they use lower cost and lower quality services. As a result o f various barriers to healthcare, the poor either live with illness (reducing their productivity and making full recovery less likely); or seek treatment late, at which stage it is more expensive and less likely to work (Box 6.7). Promoting eguig through service delivery 109 Figure 6.14 There are pronounced spatial variations in barriers to healthcare a. There are significant inter-Provincial differences in cost and distance barriers m 8" W Getting rwney for Distance to health Havingto take Notw anting to go i treatment facility transport alone Banteay Meanchey Reah Vihear / Stung Treng a Konpong Thorn FhnomPenh Sihanoukville I Koh Kong fa Total b. Quality of care also appears to vary dramatically between provinces Concern no female Concern no Concern no drugs Gettingpernission provider available provider available available to go for treatment 1 Poorest Next poorest ~4 Middle Next richest rn Richest rn Total 1 Source CDHS 2005, analyzed in NIPH et a12006 p 149 Box 6.7: The poor often delay using public services until problems are serious Distance, expense, and the perception o f low-quality care (especially for poorer patients) deter the poor from seeking treatment at a public health facility. Very poor families will seek to live with a minor illness; those with some money to spendwill first seek to cure a minor illness with low- cost traditional medicines or medicines bought from a pharmacy. For more serious conditions, those who can afford to tend to turn first to private providers who are seen to offer closer, more convenient, and often better care, and who will often extend treatment on credit (unlike most public facilities). This option is sometimes chosen even by poorer families because private providers are more numerous and closer, so saving on the cost and time o f travelling to the facility: average transport cost per treatment for private sector care was $0.89, compared to $2.36 for public sector care. Often, it is only when a patient deteriorates to a critical condition that they are taken to a public health centre or referral hospital. Source: NIPH et a1 2006 p. 37, CDRI 2007% 2007b (forthcoming) 110 Sharinggrowth: eqtlip and development in Cambodia ...and less able to absorb the economic impact of illness If poverty increases the likelihood of falling ill,ill health (and its inadequate treatment) in turn undermines livelihoods and creates or reinforces poverty. Illhealth may throw a household into poverty by simultaneously affecting both the expenditure and the income sides ofthe householdeconomy. It is obviously hard for poor households to meet high out-of-pocket health costs with the limited money available to them. However, the poor also depend more than the non-poor on unskilled manual labor to meet daily needs for food productionor income: so, on the income side, the illness of an economically active member of the household (which in poor households encompasses notjust adults but also older children) implies reducedproduction and income as well as increasedoutgoings. In the absence of instruments which would enable the poor to manage the sudden imbalance of spending and income, illness in turn leads on many occasions to coping strategies that enable the poor to get through the current crisis at the expense of long-term prospects for escaping poverty. In rough order in which they are adopted, these immiseratingcoping strategies include: running down savings, which reduces capacity to later invest in improvements in productivity (e.g. tools or livestock), livelihood diversification, health (e.g. better housingor water and sanitation) or education(children's schooling); cutting back on current consumption (including consumption o f items needed to sustain or improve productivity in the short or long term, such as food, children's education, or fuel); mortgaging or selling (typically at distress prices) productive assets such as land or animals, reducing the long-termpotential for poor householdsto work their way out o f poverty; pulling children out o f school to help out in the home or earning income (thus reducing their employment prospects and earning potential as adults, reducing their odds o f escaping and stayingout of poverty); taking on interest-bearing debts (with future repayments amounting to a non- productive expenditure and a drag on Box 6.8 Despite formalization of user fees and consumption); and / or (limited) introduction of fee exemptions, cost remains a barrier to care engaging in marginal Thepoor have to bring along with them money to show the livelihood or income- nurse or doctor so that they will give us medical treatment; earning activities-such otherwise, they will not take care of us. (FGD among as working in dangerous women, Kompong Thkoul village, Pursatprovince: PPA) areas or jobs, theft, Whoever we go to, we have topay for the services: without prostitution, or illegal money thepatient willjust be waiting to die...For instance, cross-border migration- theyprovide us with good care only when we give 20,000- that entail high risk of 30,000 riels to each nurse or other medical stafl Normally one or more serious they avoid us after three days when we run out of money. consequences for long- FGD, Kampong Thnoat village, Kampot Province: MOPS. term household survival Source: CDRI 2007a, 2007b (forthcoming) Promoting equip through service delively 111 (due to illegality; risk of illness, injury or death; or moral stigma and resulting erosion of social capital). Not all o f these channels are equally amenable to investigation. However, there are encouraging signs with regard to bullets 1, 3 and 5 in the list above. The use of short- term financing strategies that increasethe risk of long-termpoverty seems to be declining as average incomes and spending power rise and health costs decline, increasing the affordability o f healthcare. The 2000 and 2005 CDHS surveys asked about the source of money used to meet health costs over the last 30 days (Table 6.5). The answers suggest that significantly more are now able to finance health care from immediately available resourceswithout running down savings or assets, or borrowing at interest. Table 6.5 Increasingly, Cambodians can obtain healthcare without needing to resort to measures which cause them to fall into, or remain in, poverty Wages / pocket Gift from Savings Borrowed (no Borrowed (with Sold assets Year money relative / friend interest) interest) 2000 16 na 54 9 11 6 2005 45 6 30 6 7 3 Change +30 -24 -3 -4 -3 Source: NIS et a/ 2000 p. 39; NlPH et a/ 2006 p. 39. Note: The category "Gift from relative / friend" was introduced in the 2005 survey but not present in 2000. It seems likely that those who would have provided this answer in 2000 would probably have had their answers recorded under "borrowed (no interest)" or "other". It is likely that higher living standards and more disposable income have played a role in reducing the need for high-risk strategies for household financing of healthcare costs. However, RGC policy ofthe last few years seems also to have playeda part, in correcting a shortcoming of earlier policy that limited access for the poor. The earlier policy involved the introduction of formal user fees for public health services, in an attempt to enhance staff motivation through fee-financed incrementson their basic salary; suppress widespread (unpredictable and often high) unofficial fees; improve transparency and quality of care; and improve access to public health services for the majority o f the population. Policies to identify and exempt the poor were meant to ensure that user fees did not exclude poor patients. However, a study to evaluate the impact of user fees (Wilkinson et a1 2001) revealed that user fees in referral and national hospitalsdid in fact act as a barrier to the poor; and paying for health care, particularly secondary or tertiary care, was still a major cause of destitutionamong the poorest. The study concludedthat hospital exemption schemes failed to protect the poor because of an intrinsic conflict betweena viable exemptionscheme and a viable salary incentive scheme. To correct this problem, the last 5 years have seen a significant rise in the number of facilities that operate healthequity funds (HEF). These funds pay user fees to healthcare providers on behalf of the poor: in other words, they make it possible to exempt the poor from the user fees normally charged by facilities, without reducing the income received by the facilities, which would create incentives for facilities to avoid treating the poor and/or provide the fee-exempt poor with sub-standard care (Box 6.9). They appear to have had considerablesuccess in reducing financial barriers for the poor, increasinguse o f services and reducingcosts. 112 Sharinggrowth: e+& and development in Cambodia Box 6.9 Funding public health and protecting the poor: user fees and equity funds Health user fee schemes introducedunder the framework o f the 1996 National Health Financing Charter sought to increase operating finance available to facilities and get rid of widespread informal fees. Although successful in many respects, they were found to act as a barrier to the poor. Innovative health equity funds appear to have had considerable success in addressing this problem. Under these arrangements, facilities are compensated from the fund for providing fee exemptions to poor patients. These schemes havetypically involved contracts at the Operational District (OD) level between an INGO or donor, who provides the hnd, and an NGO or community organization which manages it, identifying poor patients according to agreed criteria (Conway and Crossland 2002) and reimbursingthe hospital management for the fees waived in treatingthese individuals. As ofJanuary 2006 22 HEFprojects were in operation, covering21 of the country's 76 ODs, falling in 14 ofthe 24 provinces andmunicipalities. Evaluations suggest considerable positive impact. A review of the SoknikumOD in Siem Reap found that the HEF (which compensated patients for travel costs as well as paying fee exemptions) effectively removed financial constraints to treatment for the poor, increased admissions, was cost-effective, displayed minimal leakage to the non-poor, and contributed to poverty reduction (Hardeman et a1 2001 and 2004, cited in Annear, Wilkinson, Bigdeli and Lo 2006). Evidence from Kirivong OD in Takeo suggests that HEFs can change health-seeking behavior, encouraging the poor to seek treatment at public health facilities rather than private practitioners (which provide services which are often either poor quality and / or expensive). A recent study on improving access to health care for the poor concluded that the incidence of health related debt was 3.4 times less in a squatter community with an Equity Fund than in a community without. Several issues remain. Maintaining the purchaser-provider split (essential to the HEF approach) and pre-identifyingthe poor through some form of census of the catchment population (which is preferable to attempting to identify the poor when they present for treatment) result in administrative costs accounting for a significant proportion of the total fund budget. Overall, however, HEFs appear to protect the poor at relatively low cost. Total monthly subsidies to beneficiaries amount to US$200-300 at the health center level, and US$ 3-6,000 at referral hospitals. A pre-identificationcensus exercise may cost around US$7,500. Reducedoverheads may be possible if pre-identification of the poor occurs as part of a shared, multi-sectoral targeting exercise (currently at the pilot stage). The number of HEFs is set to rise to 31 schemes in 29 ODs in 16 provinces, at which point they will cover about one-third of the Cambodian population. Under the 2003 RGC Strategic Framework for Equity Funds and the National Equity Fund Implementation and Monitoring Framework, MoH will take a more central role in setting HEF policy: to date, equity fund pilots have shown considerable variation, being implemented by a wide range of donor, INGOs, local NGOs and other administrators. Over time, it is hoped to link equity funds to community-based insurance schemes, which serve primarily to prevent the not-so-poor form falling into poverty underthe impact ofhealth costs. Sources: Annear, Wilkinson, Men and van Pelt 2006a, b; Annear, Wilkinson, Bigdeli and Lo 2006; http://www.skv-cambodia.org/ Despite progress, the healthcare cost-poverty link is still clearly strong: a third o f health episodes will lead to reduced savings or borrowing on interest. There is also a significant urban-rural gap in households' ability to meet health expenditures. While 61 percent of urban respondents were able to meet health costs from wages or "pocket money", this figure fell to 43 percent for those living inthe countryside. Promoting equi9 through sewice delivey I13 Conversely, proportionately more rural households had to dip Box 6.10 Healthcare-induceddebt and land loss remain serious problems into savings (32 percent compared to 21 percent in urban areas), Sometimes when we are seriously sick, we have to borrow with interest (8 percent take out a loan at very high interest. If we borrow compared to 5 percent) or sell 10,000 riels, we must pay 1,000 riels interest every month. (FGD of poor women, Kompong Our village, assets (3 percent compared to 1 Kompong Chhnang Province: PPA) percent) in order to cover health costs. It is harder and harder for us because our income is just enough for buying rice. Whenever any family Although weakening, the link member falls sick, mostly we go into debt or have to between poverty-exacerbated ill mortgage or sell our land. Afer recovering, we then health, high healthcare costs, and become workers in the village or have nothing to rely indebtedness remain significant. on any more." (FGD amongst those who havefallen The frequency with which illness into poverty, Khsach Chi Ros Village, Kampong leads to permanent debt and/or sale ThomProvince: MOPS) of assets is confirmed in qualitative Source: CDRI 2007a, 2007b (forthcoming) research (Box 6.10). Box 6.11 Poor households lack the resources The 2004 CSES found that to cope with health shocks illness and injury rank fourth Results from surveys in two villages south of Phnom amongst the most common reasons Penh suggest that while both crop failures and for taking out a loan (13 percent of illnesses are entailing similar magnitudes o f all loans). The burden of health- economic damage (averaging a few hundred related indebtedness is, thousand Riels), households find it harder to cope unsurprisingly, higher among the with illness. The negative consequences from health poor (accounting for 11 percent of shocks are more damaging to both immediate and the value of all outstanding loans, long-run livelihood as they require an immediate compared to 6.5 percent amongst lump-sum of money for urgent treatment. Because the richest). In a recent study, most households do not have savings (and rural credit markets do not operate well), households are illness was the number one reason often forced to resort to distress sale of productive for land sales (see Box 6.1l)9. assets (including land) and/or enter long-term debt, reducing their future income streams and increasing Reproductive and maternal their non-consumption expenditures, respectively. healthcare is improving but This broadly confirms the findings of earlier Oxfam outcomes remain poor studies which found that half of all distress sales, or around 40 percent of cases of once-landowning Reproductive and maternal families losingland, involvedhealthcrises. health issues have been RGC priorities for some time: targets Sources: Yagura 2005; Biddulph 2004; Ballard and So 2004. 9 While this report focuses on how mutually-reinforcinginequalities keep the poor poor (and hold back their potential to contribute to economic development), it is important to acknowledgethat the effects of illhealthcan be seriouseven for the non-poor. Inthe absence of working insurance markets, only the very richest can be considered immune from the economic consequences of serious illness or injury. In the CSES 2004, the 10 percent of population who reporting health status as bad or very bad accounted for 63 percent of total health OOPS. 114 Sharinggrowth: equip and developmentin Cambodia have been incorporatedin the CambodianMillennium DevelopmentGoals (CMDGs), the Health Sector Strategic Plan, and the 2006-2010 National Strategic Development Plan (NSDP), amongst others. A range of initiatives have resulted in significant improvements in relevant service delivery outcomes, which are reflected in change in intermediate outcomes such as the fertility rate. Although it is not possibleto disaggregate trends in contact with antenatal care on the basis of wealth (as the 2000 CDHS data has never been subjected to the kind of wealth analysis seen in the 2005 CDHS), it is possibleto disaggregate improvements in terms o f differences between trends for rural and urban women. This would suggest a mixed picture regarding equity in service delivery. In terms of antenatal care, there is some narrowing o f inequalities: althoughcoverage is still significantly better for urban women, improvements have been faster in rural areas, and the urban-to-ruralratios are shrinking, in some cases significantly (Figure 6.15a). Figure 6.15 Access to care during pregnancy and delivery vary greatly between rural and urban populations a. there has been dramatic improvement in b. However, despite rapid improvement, rural access to antenatal care, particularly in women are still far less likely to give birth in rural areas a health facility 79% I -c- 1 69% urban 62% 34% --eurban 22% -9-rural total 6% i+ pelc:.eillage 11 2000 2005 2000 2005 Source:CDHS 2005 pp, 138,144 Regardingsafe deliveries, however, a lot depends on perspective. Although the rate of improvement has been faster in proportionate terms in rural areas and rura1:urban ratios have come down, in urban areas a similar magnitude of change in percentage point terms (from an already much higher base) means that absolute gaps (percentage point difference between rural and urban rates) are widening over time (Figure 6.15 b). Urban women are still almost three times as likely to give birth in a health facility; and almost twice as likely to be attended by trained health staff. Inequalities are more pronounced between wealth quintilesthan between urbanand rural populations(Table 6.6). Promoting equi9 throughseruice delivey 1I5 Table 6.6 Rural-urban inequalities in maternal and reproductive healthcare are declining; wealth-based inequalities remain considerable Key indicators measure ratio in 2005 urban : rural richest quintile : poorest quintile currently using a modern method of % of all currently married contraception women 1.1 1.5 unmet need for family planning % of all currently married women 0.8 0.5 received antenatal care from trained % of all women who had a health staff during last pregnancy live birth in last 5 years 1.2 1.6 1 received at least 2 antenatal care visits % of all women who had a during last pregnancy live birth in last 5 years 1.2 na 1 suffer anemia (mild, moderate or severe) age 0.8 0.6 delivered baby with assistance of trained % of all deliveries in last 5 health staff (doctor, nurse or midwife) years 1.8 4.3 1 delivered baby at health facility (public or % of all deliveries in last 5 private) years 2.9 10.4 Source: CDHS 2005 pp. 83, 112, 138-9, 144-5, 188 Despite Government efforts and progress in service delivery and intermediate indicators (the fertility rate), there has not yet been a measurable change in the maternal mortality ratio (MMR), which remains very high (472 per 100,000 live births) and unchanged since 2000. In part, this reflects the limitations of the MMR as a measure (see discussion earlier in this chapter). In part, however, it does suggest that there is a real lag, and some bottlenecks, between service delivery gains and corresponding improvements in terms of safer motherhood. Bringing down the MMR will require concerted effort to improve the functioning of basic health services as an integrated system: in particular, it will require further improving the proportion of deliveries that occur in health facilities and with trained staff, and strengthening the referral system to ensure that women with difficult deliveries are transferred to accessible hospitals. Increasing the number, quality and motivation of public health staff, and ensuring the more equitable geographical distribution of fully-staffed facilities, will take time. In the medium term, improvements are also likely to be affected by improved rural roads that reduce the time and cost to reach healthcenters or transfer from health centers to referral hospitals. The MMR will also be positively influenced by better nutrition and, in the long run, by improvedlevels o f female education. Health equity and economic development As mentioned before, there is a strong intrinsic case for pursuing equity in health. Avoiding easily preventable illness and death is seen by most people as central to what constitutesa good life. By extension, that children who through no fault oftheir own are born into poverty are not able to avoid easily-preventable illness and death, and as a result are unfairly disadvantaged throughout their lives, is clearly inequitable, and provides a clear justification for Government intervention. I16 Sharinggrowth: eqzko and developmentin Cambodia Ill health, the labor market, and dependency Figure 6.16 Ill health reduces labor force participation However, there i s also an instrumental, economic rationale for * 4 4 e42 public action to expand the coverage 6 39 - -- e and quality of health services and e 37 break down the mutually-reinforcing association between ill health and poverty. This instrumentalcase hinges on the effect of widespread preventable illness (and high cost and inefficient health services) on national productivity. As would be expected, there is a clear correlation between Very Good Average Bad Very Tota health status and labor market good bad participation (Figure 6.16). Self-reported health status population is ill at any given point in e Hoursworked during the past 7 days Schooling and skills are affected by health status High rates of preventable illness also have knock-on effects on the level o f skills available in the Cambodianeconomy, and consequent effects on future labor productivity and international competitiveness. Illness may affect schooling directly, when children who would otherwise receive an education are too illto attend school (or too illto learn effectively when they are physically present in school). I11 health may also affect schooling indirectly. As described above, a common householdresponseto the illness of any household member is to withdraw children from school: this may be because high out-of-pocket health spending makes it impossible to sustain spending on schooling- related costs, and/or because health costs, possibly exacerbated by a drop in household food production or income when an economically active adult falls ill,make it necessary for children to be put to work earning income or helping out with agriculture or other subsistence activities. The preceding chapter on education demonstrates clearly that schooling has high returns to the individual in terms of increased earnings: children who drop out o f school (because of illness or any other reason) have much less opportunity to achieve success in household enterprises, or secure a wage-payingjob. At the national level, the aggregate effect is to keep Cambodia trapped at the level of unskilled or semi- skilled labor with low value-added. Promoting equip through seruzce delivery 117 Poor health can give rise to disability and impaired opportunity Estimatesof the disability rate in Cambodiavary dependingon definition (both by the enumerator and by the respondent). CSES 2004 found 4.7 percent of surveyed individuals reporting one or more disabilities, while the CDHS 2005 found 2.2 percent reporting according to the more restrictive category of "physical impairment". The most commonly reported form of disability was visual (3 1 percent of all disabilities), followed by impaired mobility (24 percent) and hearing (15 percent). Disability and health exist in a complex relationship with each other, and with other processes and outcomes, including education (see previous chapter). People with disabilities face complex barriers to equality of opportunity and outcome. These barriers include socially-constructedattitudinal barriers (discrimination and social exclusion) as well as potential roadblocksassociated with the physical and/or mental impairment itself. Environmental or contextual accommodations are required to allow full participation. These may include products and technology (e.g. wheelchairs or Braille books); adaptations to the physical environment (e.g. for accessibility); special types of support (e.g.job coaching); and targeted policies and services (e.g. school policies accompanied by training to teachers in the use of sign language). Unfortunately, many of these needed contextual enablers are lacking in Cambodia. While a disabled individual in a comfortably well-off household (particularly in urban areas) may be able to overcome barriers through private spending, poor families (and particularly poor rural families) are much less able to overcome these impairments. Without the adequate provision of necessary accommodations, this vulnerable population becomes opportunity-deprived, and the net result is exclusion from many forms of participation in work and community life. Disability may be either a cause or an effect of other aspects of individual or household circumstances. While some disabilities are congenital and others relate to accidents or violence, the CSES 2004 (asking about total disabilities) and CDHS 2005 (asking about individuals with a physical impairment) producesimilar estimates (roughly one-third) of disabilities originating directly from health status, and thus potentially dependent upon healthcare(Box 6.12). Healthcareand healthstatus may also account for a significant proportion of impairments arising from difficult deliveries (as small and malnourishedmotherswith low rates of antenatalcare and limited access to facilities may result in preventable impairments incurred during pregnancy or childbirth); and some portion of the disabilities attributed to "old age". A survey by Handicap International Belgium found disease identified as the causal factor in the development of childhood disability in 46 percentof cases (VanLeit 2007). This suggests that poor healthcare(inadequatecoverage of preventative care and poor access to and quality of treatment for illness and injury) may contribute to the development of many disabilities which might otherwise be avoided. Conversely, this also implies that improvements in healthcare" plus targeted policies to allow for the 10 Improvements that would reduce the incidence or severity of disabilities include preventative care (antenatal care, safe delivery, postnatal care; further improvements in vaccination); environmental improvements and practices (e.g. improved water supply and sanitation; bed I I8 Sharinggrowth: eqtrity and developmentin Cambodia participation of disabled ~ individuals in schooling and , workplaces could reduce the Box 6.12originateainthird of all disabilities Around poor health incidence of disability and the a. "illness"is the single largest cause of unproductive economic costs it - physicalimpairment entails. m m 1 Disability rates appear not 2- u 33% to vary much by per capita I consumption, although the causes show some differences: old age is a more important quintiles (who live longer), as cause of disability for wealthier i' are traffic accidents; while congenital causes, mine or 5 birth illness iandmine gun road other s accident accident unexploded ordnance (UXO). injury, malnutrition, and , Source: CDHS 2005 p. 30 difficult delivery are all more common among the poorest b. disease, malnutritionand fever account for a quintile than among the richest. third of total disabilities There are some gender differences, too: women report Cause of disability YOof reported disabilities old age 27% slightly more disabilities, with disease 26% old age, disease, malnutrition, congenital 9% difficult deliveries and accident ' 7% domestic violence more fever 6% important for women than men, violence 4% mine I UXO 3% and accident and war-related malnutrition 2% injuries a more important cause difficult delivery 1% among men. bad luck 1% mentaltrauma 1% Early childhood Source: CSES 2004, analyzed in Knowles2005b p. 8. development can help Notes '"Accident"includes (inorder of significance) traffic accident, work accident,chemical accident and burns tackle inter-generational "violence"includes(in order of significance) war injuries, transmission of inequality domesticviolence, violent attack,torture, suicide attempt and rape 8 percentdue to "other" causes; cause "not known" for 5 There is clear evidence that percent. the major damage caused by 1 malnutrition takes place in the womb and in the first two years of life; that this damage is irreversible; that it has consequences for cognitive and physical capacity (which in turn reduce productivity, slow economic growth and perpetuate poverty). It is also clear that malnutrition passes from generation to generation, since stunted mothers are more likely to give birth to underweightchildren. nets); and improvements in the accessibility and affordability of timely, good quality treatment of illness and accidents. Promotingegtlig through service delivey 119 Internationally, several studies have documented a strong correlation between adult height and earnings. Malnourished and sick children are more likely to reach adulthood with reduced height. There is also evidence that the health and nutritional status of children are linked with educationalattainment, both in terms of cognitive capability and school attendance. The effect on adult earnings and productivity is estimated at 10 percent for stunting and 4 percent for childhood anemia. Estimatesof these losses to the economy are estimatedat 2-3 percentof GDP in low incomecountries (ADB 2001). It has been estimated in one recent international study that the present value (using a discount rate of 5 percent) of the benefits of shifting one child at age 3 from malnourished to normal nutritional status is US US$514 (Behrman, Alderman and Hoddinott 2004 cited in Knowles 2005.). If this estimate (or anything close to it) is accepted, the potential benefits from interventions designed to reduce Cambodia's currently very high rates of child malnutrition are about US$84 million annually or about US$6.50 per capitaper annum". The macroeconomics of investments in water supply and sanitation A global study by WHO concludes there are potentially substantial benefits from investment in rural water supply and sanitation. The regional estimateI2of the cost benefit ratio (for urban and rural combined) for water supply alone is estimated at 8.17. For water supply and sanitation combined it is 11.04. This means that for every dollar invested, the economic benefit would be $8 and $11 respectively. Even the most pessimisticestimateof high costs and low benefitsshowed a positive cost-benefit ratio of 1.21 and 2.21 respectively. The results show that the benefits from every $1 invested in sanitation would be $23.48 (using mid-level values), with a range from $11.09 in a pessimistic scenario to $58.75 in an optimistic scenario. The economic benefits usually accrue as the result o f avoiding costs (e.g. for health care, missedwork) and gainingtime. Specific economic benefits associated with improved health arising from improved water supply, sanitation and hygieneare summarizedbelow. Table 6.7 Improved water and sanitation has multiple positive effects Beneficiary Direct economic benefits of Indirect economic benefits Non-health benefits related to avoiding diarrhoeal disease related to health improvement better water and sanitation Health . Value of less health workers - Less expenditure on 1 1 More efficiently managed sector treatment of diarrhoeal falling sick with diarrhea water resourcesand effects disease - - -- on vector bionomics - _- -- II - - - - ---I I I _. 1-1- " Patients Less expenditure on Value of avoided days lost at More efficiently managed treatment of diarrhoeal work or at school water resourcesand effects disease, less related costs . Value of avoided time lost of on vector bionomics Less expenditure on transport parenffcaretakerof sick in seekingtreatment children Less time lost seeking care Value of loss of death avoided Source Hutton and Haller.2004 `I The CSES suggests there were 277,000 children aged three in 2004. Multiplying this by the percentage of children age 3 that are estimated to be moderately stunted and the estimated potentialbenefits per malnourishedchild of US$514 yields the estimate cited in the text. I2 The World Bank Water and Sanitation Program is currently financing an "Economics of Sanitation Initiative" in Cambodia. Country specific figures will be available later in 2007. 120 Sharinggrowth: equip and developmentin Cambodia Educational equity and inter-generational mobility This section examines the distribution of educational opportunities and outcomes in Cambodia and the relationship between education and living standards. Given the advantages conferred by education, it seems there is a strong case for ensuring equal access to schooling, on both social grounds (fairness) and aggregate economic grounds (transforming innate talents into skilled labor). While education indicators are poor, trends have been positive (rising averages and shrinking inequalities between rich and poor, men and women, and rural and urban areas). Nonetheless, a combination o f monetary and non-monetary barriers continue to reproduce pronounced inequities between children in terms of access to educationand the opportunities that come with it. The returns to education: schooling and economic opportunity The amount and quality o f educationthat an individual receives is a critical influence upon his or her opportunities in life (e.g. securing good employment, or succeeding in a family business). Statistical analysis confirms that higher levels of education are clearly associated with higher average incomes and higher standards o f living (see Figure 6.17). An educated individual will also be better informed about how public services work, and how to obtain usehl assistance from the Government. Cambodians, including most poor Cambodians, clearly recognizethis (see Box 6.13). Box 6.13 Preference or constrained choice: do the poor value education? Two dlfferent qualitative research projects by CDRI suggest somewhat different pictures of demand for education amongst the poor. T h e 2005 PPA carried quotes a younggirl: My mother askedme, `whatis the use ofyourgoing to school, if we don't have anythingto eat and I@is so dz@cult?' (boor and destituteFGD,Battamband T h e PPA concludes that Whilethe rich and mediumfamilies see educationas the meansfor childrento have a betterfuture, thepoor and destitute do not see that investing in educationwillprovide opportunities to move out ofpove@ T h e MOPS research, however, gives a different picture, with clear recognition of the role of education in givingyoungpeople a chance to go further than their parents: Education is the most important asset ojpeople in this world Being well educated,someone canjnd agood j o b with a high standard of living. Withgoodeducation,s/he can manage to have a better I@or be able to set a cleargoal. ..For example, a literate woman is able tojnd agoodjob at the garmentfacto7 now. Those who have no education cannotfind suchfortune and become mobile laborers, a more risky I@. (Femaleyouth group, Kampong Speu). Youngmen recognized that their better-educated peers had better prospects than them: School is vey important as it is aplace wherepeopleget some knowledge and train themselvesfor a better I$....Thosepeople who are educatedcan make mon9 more easi4 and do non-manualjobs...Somepeople of our age whojnished high school can seek work in Phnom Penh while we cannot. (maleyouth group, Kompong Thom) Source: CDRI 2007a, 2007b (forthcomind Promotingequig through service deliver 12I Figure 6.17 There is a strong association between educational achievement and the quality of employment and income a. more educated individuals are much more likely to obtain paid employment Unerrployed Waged (private sector) Waged (public sector) 7 No schooling Rimry Rimry Lower Secondary Upper Secondary University (incomplete) (complete) b. ...and, within any given category of paid employment, to receive higher pay = 7,000 m N .g 6,000 v 8 5,000 0m -2 5 4,000 m t L 3,000 n c 8 h 2,000 E> 1,000 All paid employment Waged paid Waged errployment, Waged employment 3ther (non-wage) employment public sector private sector paid errployment I All w No Primary Rimary Lower 0 Upper University 1 schooling (incomplete) (complete) Secondary Secondary c. 7,000 Eu3 - 1 6,000 4 5,000 m co 7 0 0 t @4 m m `c1: 7 Skilled Unskilled Formal sector Informal sector Source: CSES 2004, analyzed in Sakellariou 2007. 122 Sharinggrowth: egtlio and developmentin Cambodia There are some exceptions. In remote, poorly connectedrural areas, there may be no significant alternatives to basic, subsistence-orientedlivelihood strategies based on low- input farming and common property resources. In such communities, where children are likely to end up in farming occupations regardlessof the schooling they complete, there is little economic incentivefor investing in education:when parents and children perceive this to be the case, there is, unsurprisingly, low demand for schooling (Bredenberg 2003, cited in World Bank 2005 p. 68). However, as the rural road network expands, these peripheral rural areas will gradually open up to new opportunities, either in situ or through out-migration, increasingthe potential value of education. Figure 6.17 highlights the implications of increasing levels of educational attainment for labor market activity status, type, and sector of employment for adults (aged 22-65 years). Higher education qualifications constitute a passportto employment in the higher paying wage employment sector; conversely, low education essentially prohibits access to higher-paying formal / waged employment (Sakellariou 2007 pp. 13-17)13. Only 2 percent of workers with no schooling and under 6 percent of workers with incomplete primary education are employed for wages (two-thirds of them in the private sector): by contrast, 52 and 82 percent of workers with upper secondary and university education work for wages (two-thirds in the public sector). Figure 6.17-b illustrates how differences in education translate into differences in earnings from paid employment within various types and sectors of employment. Working for wages in the private sector is the most rewarding labor market activity, with an average hourly wage of 1,874 riels (much higher than the average 1,047 riels for paid employees as a whole). However, there are considerable differences in pay amongst private sector waged employees: those with upper secondary education enjoy wages 3.6 times higher than those with no education, and those with a university education earn rates 5.3 times what those with no education earn. Wages in the public sector are much more compressed (Table 6.8). Table 6.8 Wage rate differentials in paid employment Category of labor market activity ratio of average hourly wages % of working age population no education :upper no education : secondary university a. Waged 11 2.1 3.3 -Waged, public sector 6 1.4 2.2 - Waged, private sector 5 3.6 5.3 b. Other (non-wage)paid emPloYment 7 17.6 na' All paid employment i a 6.5 10.3 Source:CSES 2004, analyzed in Sakellariou 2007. More sophisticated multivariate analysis of the returns to education in waged employment (which, it should be remembered, accounts for only 11 percent of the labor l3 The largest employment category is informal self-employment, followed by unpaid family work. Only 18 percent ofthe labor force is inpaidemployment; only 11 percent earn regular wages. Promoting eguig through sewice delively 123 force) yields somewhat complex findings, with average returns to schooling diverging considerably depending on individual characteristics. The distribution of wages appears to reflect an interaction between educationand ability, which results in different average returns to education at different levels of wage and education, and depending on sex (Sakellariou 2007 pp. 22-35). Similar education-based differentials exist between the earnings of self-employed individuals, who account for 43 percent of the labor force (Figure 6.18). Those earning between 100,000 and 149,999 riels (c. $25-37.50) per month appear to be equally likely to have no education, primary, lower secondary or upper secondary education. Those earning wages below 100,000 riels ($25) per month, however, are significantly more likely to have no education or only primary education; and those earning wages in bands above 150,000 riels ($37.50) per month are more likely than not to have a secondary education (Ridao-Can02004; World Bank 2006). Figure 6.18 Average earnings amongst the self-employed appear to reflect educational attainment B No schooling SJ Rmry E Lower secondary D Upper secondary ab a -- IJ) .c 0 :: L z 49,999 50,000-99.999 100,000-149,999 150,000-199,999 200,000-299,999 300,000-499,999 > 500,000 Eatirratedearnings (riels per month) Source: CLFS 2001, analyzed in Ridao-Can02004 p. 33. Given the apparent importanceof educationin shapingan individual's chances in life, ensuring equality of access to education is fundamental to ensuring equality of opportunity in society. Increasing the level of educated individuals in society also enables Cambodia to compete in a global economy for investmentswhich require skilled workers: thus, broadening access to quality education, and ensuring that those who get education are those who can make best use of it, contribute significantly to the efficiency and diversity of the economy and to the long-term sustainability of economic growth. It i s in the national interest that children with innate abilities can obtain the education that enables them to make use of these abilities to generate wealth, regardless of the circumstancesinto which they were born. The goal, then, is a Cambodian education system that will enable all children equal access to schooling and will enable them to acquire as little or as much education as their innate talents and preferences dictate, regardless of their sex, where they live, or the wealth or poverty of their parents. This is, essentially, the policy position of the Royal Government of Cambodia (see Box 6.14). Cambodia has made significant progress 124 Sharinggrowth: eqtlig and developmentin Cambodia towards this goal in recent Box 6.14 Since 1993, the Government has been years. Despite this progress, committed to equity in education however, the level of inequality in to education services The 1993 Constitutionof Cambodia notes that it is obligationof the state to provide nine years of free and (schooling) and education compulsory education to all its citizens, an obligation outcomes (such as literacy)- affirmedby the RoyalGovernment of Cambodiawhen it and the effect of these ratified the Convention on the Right of the Child and inequalities in terms of the signed the Education for All Framework for Action. In perpetuation of wealth 2003, the Government adopted the Education For All differences and the below- (EFA) National Plan for 2003 -2015, the fundamental potential performance of the thrust of which is to ensure that all Cambodia's children economy-remain very and youth have equal access to formal and non-formal pronounced. A number of key education, independent of economic status, gender, challenges remain in geography, physical disability andethnicity. Cambodia's progress towards realizing the objective of quality basic education for all. Average educational status is low, but improving Standard educational attainment indicators for Cambodia are among the lowest in the world. Amongst its neighbors, only Lao PDR has higher rates of illiteracy; and only China has a higher share of the adult population with no schooling. Repetition and dropout rates in primary school remain very high. The inefficiency of the Cambodian education system represents a significant burden: the public and private cost of education that reduces the private and social returnsto schoolingand thus economic growth. Nonetheless, remarkable progress has been made in recent years. In 2003/04, the number of pupils aged from 3 to 5 has increased by 11 percent and pupils aged over 5 increased by 20 percent. Primary school enrolment has grown significantly, with the majority of growth being in rural areas and an 8% growth in female net enrolment. Enrolment at the lower secondary education level have also increased significantly, reflecting in part the expansion in the Table 6.9 There has been a dramatic expansion in number of secondary basic education in recent years schools, mainly in Performance indicator 2000 2004 change previously un-served Pre-schools 952 1,275 34% rural and remote Lower secondary schools 511 688 35% communes (Quinio Classes 1,628 2,249 38% 2005). Primary enrolment (millions) 2.4 2.7 14% Equality in access Primary enrolment in remote areas 54,000 81,000 50% to primary Primary net enrolment rate (NER) 84% 90% 7% education is Lower secondary enrolments 284,000 460,000 62% improving Source: Sakellariou 2007 citing Quinio 2005. As a result of expanding coverage, equality in access to education is improving as increasing numbers of girls and children of rural and/or poor families are drawn into the school system. Promoting egui'y throughservice delive7y 125 Although there continue to be large inequalities in outcomes (notably literacy) between men and women, rich and poor, and rural and urban populations, these group averages mask considerable progress in recent years. Despite problems with dropout, repetition and quality of teaching, greater equality in access to primary education is starting to result in closing gaps in educationoutcomes. This progress emerges when we look at the variation in literacy rates amongst the younger age groups which have more recently passed through the (improving) education system. Amongst these age groups, literacy rates are higher across the board, and the inequalities by sex, location or socio-economic status are shrinking rapidly (Figure 6.20). In many aspects, however, inequality remains pronounced Despite the slow but consistent progress towards increasing participation o f poor children, the level of inequality in the distribution of education in Cambodia remainsone of the highest in the Southeast Asian region (Holsinger et al 2004). There are large and statistically significant differences in average years of schooling completed by sex, employment type, and various types o f geographicaldifference(Figure 6.x, Table 6.19). Figure 6.19 Although shrinking over time, educational inequalities remain large a. between urban and rural areas d. between landscape / agroecological zones Rural Rateau / mountain Urban Tonle Sap Coastal Rains FhnomFenh 7 7 b. between men and women years schooling Female Male e. between self-employed and waged years schooling employees Famly business c. between the capital and the remote NE highlands Unemployed Self-enployed Rattanakiri 1 7 Paid errployment Fhnorn Penh Wage employment years schooling years schooling Source: SES 2004, analyzed in Sakellariou 2007 126 Sharinggrowth: equip and developmentin Cambodia Figure 6.20 Increasing equality in primary enrolment is closing gaps in literacy rates amongst the younger generation 100, a6 a2 4 82 + 71 E 7 0 0 42 2 5 1 , , , , , , , 0 2 5 m 1 ~ 0 g , -.e E 0 15-19 20-24 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-64 65-6970+ age (years) loo 91 1 + a9 + a3 4 a4 4 a4 a3 4 83 4 ao 0 74 4 77 4 75 0 67 70 66 60 60 '62 57 4 60 4 55 4 Urban 0 45 39 4 36 2a 0 4 15-19 20-24 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-64 65-69 70+ age (years) 100, 94 -* 93 aa 4 * - 4 87 a7 *a-i +-a5- a2+, 76 0 a 75 71 ' '67 ns 62 7 62 0 . + 53 53 0 0 p! 46 0 49 E x Next richest 41 -PE 25 BI Mddle t Next poorest 0 Poorest 0 4 15-19 20-24 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-64 65-69 70+ age (years) Source: CSES 2004. Promoting egui9 throughservice delivey 127 Table 6.10 Dimensions of inequality in education The Tonle Sap PPA Ratio of average years schooling for: ratio concluded that three main urban : rural 1.72 factors determined the level of Phnom Penh : plateau I mountainProvinces 2.26 education received by children. Phnom Penh : Rattanakiri 4.53 These were (i) family financial men :women 1.64 status; (ii) the educational family business :wage employment 2.50 background of the parents; and Source: CSES 2004 analyzed in Sakellariou2007 (iii)the distancefrom hometo school. It also noted that girls are much more likely to drop out of school at an early stage. Investigation of the data from CSES and CDHS confirms the importanceo f these four factors and the implication that the children of poor families-families with limited finances, low level of parental education, and/or living in remoter areas-are likely to complete far fewer years of education than their richer contemporaries; while girls remaindisadvantagedin education relativeto boys. Reflecting these various barriers to access, inequalities in participation and completion rates feed through into educational outcomes (or individual capabilities). Although closing over time, considerable inequalities remain between the average literacy rates of men and women (20 percentage points), rural and urban populations (16 percentage points) and the richest and poorestquintiles (3 1 percentage points). Inequalities in access to education reflect inequalities in wealth Education costs money, and part o f the costs o f education must be met by the household. Poor parents are less able to afford this money and so their children are less likely to obtain either the quantity or quality of education available to the children of richer parents. While this is the case in any country, Cambodia is unusual in the proportion of total education costs that must be borne by households. (The same point is made above with regard to out-of-pocket spending on health.) In the late 1990~~ households and communities in Cambodiawere providing an estimated 59 per cent ofthe total resources for primary schooling, even in the public system: the Government provided an estimated 12.5 percent, while the balance (28.5 percent) was met by politicians, NGOs and external agencies (Bray 1999). The proportion of costs met by households is thus very large, and considerably larger than in most countries. This has implications for urban/ruraland socioeconomic inequalitiesin Cambodia. These direct costs of education have clearly come down in recent years. From 2000, the enrolment fee for primary education was abolished: focus group discussions suggest that this was a major factor in increasing primary enrolment amongst poorer groups (CDRI 2007b, forthcoming). Nonetheless, there are extremely striking wealth-based differences in the amount spent on education per child enrolled in school. In the 2004 CSES, per child household education expenditures amongst the richest 20 percent of households was found to be 25 times greater than in the poorest 20 percent of households (Figure 6.21). 128 Sharinggrowth: eqtriij and development in Cambodia Figure 6.21 The richest parents outspend the poorest parents by a factor of 25 Poorest Next poorest average, all quintiles. 118,034 riels Mddle Next richest Richest 387,995, bkan spending on education per child per year (riel) Source: CSES 2004, analyzed in Knowles 2006 p. 82 This gulf in spending represents the cumulative effect of a number of phenomena, including not only that the rich spend more on like-for-like education (e.g. the rich will spend more on a year of primary education than will the poor) but also that the rich are more likely to educate their children through to lower and then secondary school, and the annual costs o f schooling rise with each of these transitions (Le. it costs considerably more to put a child through a year of secondary school than it does to put a child through a year o f primary school). Nonetheless, these figures are telling as a summary measure of the much greater capacity of rich parents to invest in the education of their children. As with household out-of-pocket spending on health, Figure 6.22 reveals that the increase in household spending on education does not increase in a smooth linear fashion but in a series of increasingly big steps (the richest spending on average four times as much as the next richest). And, as with health spending, it seems that the very large gap between the bottom four quintiles, on the one hand, and the top quintile, on the other, broadly reflects the distribution of wealth (measured as total consumption per capita per day). That the richest can spend so much more on their children's education reflects in part that they have so much more to spend, with levels o f total consumptionover twice that of the next richest quintile. The underlying gap between the living standards and disposable incomes of the richest quintile and those of the remaining four quintiles helps explain why, even with significantly higher absolute spending, the relative burden of educating children is less for rich households. As a proportion of their much higher total non-food consumption, the rich spend less on education(per child) than do the poor. While the relative burdenof schooling is roughly similar for the richest and poorest households at Grade 1 (8 percent and 6 percent of total non-food consumption, respectively), it is significantly harder for poor households to meet the costs of schooling to higher levels. Those few households from the poorest quintile who educate their children to Grade 12 pay on average 58 percent of their annual per capita non-food consumption for this year: the comparable figure for the richestquintile would bejust 35 percent. Promoting equig through service deLive9 129 Figure 6.22 Richer parents find it can afford (much higher) education spending Poorest Next poorest Mddle Next richest Richest Total Grade 1 Grade 6 B Grade 7 E Grade 9 Grade 12 Source CSES 2004, analyzed in Ragatz 2005 p. 26. However, indirect costs remain a significant burden for poor households. Indirect costs (such as foregone earnings) constitute a heavy burden for some households, particularly the poorest, and can be a significant determinant of whether or not a child attends school. As in other low-income south-east Asian nations, child labor in is Common (seen in 23 Box 6.15 For poor households, demands on percent of households) and children's time is a key reason for over- begins at an early age. In age enrolment or dropout both ruraland urban settings, Poor Because we are poor, our children quit school at an early often age or after only one or two years in order to help their rely heavily on their parents earn a living. Unfortunately, they cannot go asfar children to help with a as the rich do in obtaining skills to earn a living. Because variety o f tasks (generally we are trapped in illiteracy, we have poor knowledge and household-based) essential are without ideas, remaining short-sighted and powerless. to the wellbeing o f the (KomPong Thorn: Mops) family (Box 6.15). Source:CDRI2007b (forthcoming) These forms of child labor include both income-earning or subsistence-oriented productive or trading activities, and household reproductive tasks (caring for younger siblings, cooking, cleaning, carrying water, etc.)-much of which, of course, is necessitated in order to free up both parents for economic activity, including in some cases long absences from the home in pursuit of income opportunities elsewhere in Cambodia or in Thailand. The CLS in 2001 found that children contributed on average 28 percent of total householdlabor inc~me'~. For many Cambodianchildren, work and school attendance are largely substitutesfor one another, with this trade-off becoming more pronounced as children get older. The burden of housework and productive work is a particular problem for rural children, l4 This figure coincides remarkably withthe estimatesoftheir contribution to householdincome providedby youths, both male and female, in MOPS focus group discussions. 130 Sharinggrowth: eguig and developmentin Cambodia children of families in the poorest quintile, and girls. From age 12 onwards, household work is significantly more likely to interfere with girls' education than boys. Children who work more than 14 hours a week-who are more likely to be poor, rural, and female-start to fall behind in grade attainment comparedto their peers who only attend school. This difference is insignificant at age ten, but pronouncedamongst 17-yearolds (Ragatz 2005, p. 25-6). The high proportion of children reported to both attend school and work suggests that the householdneed for child labor not only contributesto outright dropout from school, but will also make it hardto increase the currently very short school day. Direct and indirect household costs reinforce each other to producea critical barrier for the poor in upper primary and lower secondary schooling(Bray and Bunly 2005). Distanceto school may also be a barrier to educationfor the poor: althoughat present less important than other barriers, if may become more pressing as a binding constraint on expanding access to secondary education. The poor are more likely than the rich to live in remote parts of the country, with lower population densities and fewer schools. However, the rapid pace of school construction since the mid-1990s has significantly increased the number of primary schools. As a result, roughly 50 percent of Cambodians live in a village (or urban equivalent) that has its own primary school; and the average distance to the nearest primary school (assuming a distance of zero for those households with a school in their own village) is under two kilometers (Figure 6.23). Neither figure varies much between the richest and the poorest households: both are equally likely to live in a village with its own primary school, and the difference betweenrich and poor in terms o f average distance to school (2.25 km compared to 1.66 km) is not a serious barrier. (The upland north-east may be an exception, and may Figure 6.23 Distance is not now a major provide one-though maybe not constraint to primary education- the most important-explanation but is likely to act as barrier to for the low educational secondary education for the poor attainment of the indigenous ethnic minority groups living in 1 this region: see Box 6.16). rnRichest Primary Next richest Among the general school Mddle population there is however Next poorest significant inequality in physical access to lower and upper Lower secondary schools, with the secondary school children of the poorest families needing to travel considerably further than the children of richer Upper parents. secondary school Disability and access to , education mean distance (km) to school Disability often acts as a Source: CSES 2004, analyzed in Knowles2006 p. 74 severe limitation on the educational opportunities available to children and, thus, on their employment and incomes opportunities. Promoting equip tbrougb service delively 131 Box 6.16 Inter-ethnic inequalities in education The Cham, indigenous ethnic minorities and ethnic Vietnamese, Lao and Chinese groups, which collectively make up 4 percent o f the Cambodian population, have very different rates o f participation in the education system. The educational performance o f these different groups broadly reflects the relative wealth o f these groups, with one notable exception. The Chinese have enrolment rates higher than that o f the Khmer majority, corresponding with their higher- than-average living standards(56 percent of the Chinese population fall into the richest quintileof the CSES sample); the Cham, whose living standards are very close to those of the Khmer, have rates lower than but similar to those of the Khmer; and the indigenous ethnic minorities, who are significantly poorer than the Khmer (60 percent are in the lowest two consumption quintiles), have the worst enrollment and drop-out rates of all. The group that does not fit this correlation between overall living standards and education is the Vietnamese, who have living standards slightly above those of the Khmers (58 percent fall in the top two consumption quintiles, compared to 40 percent o f Khmer) but notably poorer school enrolment. a. with the exception of Chinese, non-Khmer ethnic groups have significantly lower rates of current enrolment... 0 R s Chinese Khmer Cham Vietnamese Other local group b. ...andsignificantly higher rates who never enrolled E Be N e C a, 0 0 P * e 0) m m e 0 8 Chinese Khmer Cham Vietnamese Other local group Poor educational outcomes of the indigenous and Vietnamese minorities reflect, to varying degrees, the effects o f remoteness; low population density; the lack o f a fixed abode; language barriers and the lack o f ethnic minority teachers; and social and political marginalization. O f the local indigenous minority populations, 27 percent live in Mondulkiri, 26 percent in Preah Vihear, and 13 percent in Kampong Speu; the Vietnamese tend to live in Phnom Penh, Kampong Chhnang and Preah Vihear (43,20 and 10percent respectively). Source: CSES 2004, analyzed in Ragatz 2005 pp. 10-12; Phnom Penh Post July 14-27 2006. 132 Sharinggrowth: equity and developmentin Cambodia The CSES 2004 finds adults with disabilities have much lower average rates o f educational attainment. Rates of low or no schooling are particularly high amongst individuals with problems in vision, hearing and speaking, suggestingthat relatively low- cost interventions could eliminate a significant proportion o f impairment-related educational under-achievement A survey of households with disabled children in Siem Reap and Takeo Provinces conducted by Handicap International Belgium found that 45 percent of school-aged children with disabilities never attended school, and most of those who did attend dropped out at very early grade levels. Reasons given for non-attendance included problems with transportation to school; lack of special equipment to overcome learning impairment; expenses associated with attending school; problems with teachers who did not have training or experience in teaching a child with disabilities; health concerns; and discrimination on the part of teachers or other children. lnequalities in education perpetuate themselves across generations If education is associated with wealth, and parents' wealth is associated with their children's educational achievement, it is not surprising that parents' education is itself a good predictor of the education of their children. Just as poor women who are undernourishedand undersizedtend to give birth to small babies who are more likely to be illand more likely to grow into undersizedadults, so too parents' educationalstatus is passed on to their children, with obvious implications for equality of opportunity. Inthis case, however, it appears that it is the father's level of educational attainment that is the stronger influence on their children-albeit with some exceptions (Figure6.24). Female education has cross-sectoral and intergenerational effects Improving education affects household welfare through a number of channels other than employment and wages. In particular, there is a strong argument that there is a strong correlation between the level of education of a mother and improved childhood health. As described in the previous section, the children of poor families are born at a severe disadvantage to their richer peers in terms o f access to healthcare and probability of poor health outcomes. However, there are also strong suggestions that a woman's education is a major determinant of both child and maternalhealthoutcomes. The advantages of children with educated mothers start even before birth. Nearly half (46 percent) of all uneducated women who have given birth during the last five years went through their last pregnancy without any contact with an antenatal care (ANC) provider; the comparable figure among women with secondary education is a fifth of that rate (9 percent). Promoting epio through service dehey 133 Figure 6.24 Parents' education appears to be a strong influence upon a child's educationalachievements a. the level of education achieved by fathers shapes the years of schooling achieved by children... -3 I 3 15 -131- ~ > No schooling Rimary Rimary Lower Upper University incomplete complete Secondary Secondary Level of father's education b. ...and the level of education completed by children ..-6 h m .-VCc Rimary s2 complete University -m r Rimary I0 incorrplete L I E .-c I 0 -0 No Upper fA C schooling secondary 42 L - - Lower secondary 3 -0 i-- No Rimary Rimary Low Upper University schooling incomplete Complete Secondary Secondary Level of father's education Source: CSES 2004, analyzed in Sakellariou 2007 p, 6. This difference in contact with health services is reflected in the likelihood of the unborn child receiving important aspects of antenatal care. The child of a mother who has completed secondary education is almost 50 percent more likely to be protected against neonatal tetanus, and 66 percent more likely to have benefited from his or her mother taking iron supplements during pregnancy, than the child of a woman with no 134 Sharinggrowth: equi4 and developmentin Cambodia Figure 6.25: A mother's education influences a education (Figure 6.25). child's chances of obtaining health The child of an educated care, from conception onwards woman thus enters the world with a head start in protection The children of educated mothers receive better against one of the major care from before they are born... causes of neonatal deaths; and in the micronutrients required for healthy growth 7n 83 over the first six months of his or her life. Women with secondary education are three times more likely to deliver a baby with a trained health hrather received no fvbther took iron hfant fully professional; and five times 'c antenatal care supplements during protected against more likely to deliver at a during pregnancy pregnancy neonataltetanus health facility. Both factors No schooling 0 R i m r y 0 Secondary or higher 1 make childbirth much safer, for both the mother and her ...are better protected by vaccinations... child. These inequalities in 83 access to basic healthcare are compounded during a child's first year of life. Once again, maternaleducationappears to be an important influence. Although very few Cambodian children-seven All basic No vaccinations percent-now have no vaccinations vaccinations, this rate varies ...and considerably between those receive treatment for common and potentially serious illness. children whose mothers have a secondary education (one percent) and those who have 62 no education (13 percent). A 50 baby whose mother has a 39 42 r secondary education is 60 percent more likely to receive all of the basic vaccinations to protect her or him against common, and dangerous, childhood diseases, relative Acute respiratory Fever Diarrhea to the child of a woman with infection (ARO no education. Source: CDHS 2005 pp. 138, 140, 142, 154, 160 Promoting equity throughservice delivery 135 Figure 6.26 A mother's education strongly shapes The better standard of her children's heath outcomes healthcare enjoyed by children of educated Children born of uneducated mothers are more mothers extends from likely to be born dangerously undersized .... preventative services to curative care. Compared to mothers with no education, mothers with a secondary educationare 25-40 percent more likely to seek treatment for their children Very small Smaller than average when they display symptoms of acute No schooling Rimary Secondary or higher I respiratory infection (ARI), fever or diarrhea. ...tobe malnourished ... As Figure 6.26 shows, g aB 1 46 44 the disadvantages faced by the children of uneducated mothers in terms of unequal access to and use o f basic healthservices feed through into health outcomes. When compared to the sez3 Stunted Wasted Underweight child of a woman who has completed secondary ...tosuffer from serious childhood illnesses ... education, the child o f an - . . uneducated mother is ten times more likely to be born very small; twice as likely to suffer from chronic malnutrition in their first five years, L n "' sO* ' 5 E Z g ARI Diarrhea Diarrheawith fvbderate/ resulting in stunted growth; synptom blood severe anema one-and-a-half times as likely to suffer from an acute respiratory infection . ...and to die before age five. ro or anemia; and twice as 136 likely to suffer from bloody diarrhea. Reflecting these health problems, the child of the uneducated mother is two-and-a-half times more likely than the child of the m o 7 7 o rg Infant mrtality rate Under-fivemrtality educated mother to die rate before reaching his or her Source:CDHS 2005 pp. 123,152,156,159,167,179 fifth birthday. 136 Sharinggrowth: egtlig and developmentin Cambodia Policy recommendations Health policy needs to prioritize equity objectives Equity goals need to be a more explicit objective of health sector reforms. (Health equity-or "improving access and coverage for the poor"-is included in the current policy framework, but as one among three elements in the strategy of "improving service delivery", which is in turn one among the 8 priority (and 20 total) strategy strategiesthat comprisethe Health Strategic Plan (HSP) 2003-7.) This will require some changes to the ways in which health sector outcomes are monitored. At present, monitoring concentrates mainly on aggregate indicators. More disaggregation-by location, by sex, and by consumption or wealth status if possible- would ensure that policy-makers and planners have the information needed to formulate and revise policies designed to achieve health equity. Data on inter-Provincial differences are already available in the health information system, and should be used to direct programsand resourcesto laggingareas. Increased attention should be paid to disability prevention, both in the health care system and in other sectors. A functioning primary care system that meets the needs of pregnant women, provides fast, inexpensivecare for basic childhood illness, and screens for children with early developmental problems could go far to prevent avoidable disability and the socio-economiccosts (to individuals, householdsand communities) that this entails (VanLeit 2007; Knowles 2004). More ambitiously, the Government should seek to provide people with disabilities with special equipment and training that allows participation in daily activities at home and in the workplace. Glasses, canes for the blind, sign language and wheelchairs can make a profound difference in terms of allowing increased activity and function. Cambodia already has good insights into policy innovations that promote health equity (along with other health goals). Health equity funds have proved effective in reducingthe financial barriersto the poor (see above). Contracting out arrangements for the management of District-level services has also proved to have a positive impact on equity, with health utilization increasing faster amongst the bottom rather than the top half of the wealth ranking (see Box 6.17). Both experiments are now being expanded: this can be expected to have a positive impact on health outcomes for the poor. Over the long term, investments in water and sanitation and child health (encompassingantenatalcare and nutrition, safe delivery, and early childhood healthcare and nutrition) can have a profound effect on the learning capacity of school-agechildren and health, productivity and earningsas adults (Knowles 2004). Finally, more needs to be done to build on what is known about critical cross-sectoral linkages (in one or in some cases both directions) between health and other development outcomes (for example, better roads, civil service reform, female education, rising household consumption, or water and sanitation). These linkages are important firstly as factors facilitating or constrainingthe achievementof health sector goals (suggestingthat MoH and it's partners should, at a minimum, advocate for progress in areas of public actionthat have a bearingon progress in health). Promoting equip through service deliveZy 137 1 Box 6.17- Government-NGO partnerships have reduced out-of-pocket costs for -- -- -______I_ _---I-___ "l__llll__l-l- ~ the poor and delivered improvements in health equity I I In 1998 the MoH, with support from ADB, beganan experiment inpublic-privatepartnershipsfor the delivery of basic health services at the Operation District (OD) level. Contracts to manage service delivery in five districts were awardedto NGOs on the basis of international competitive tendering. Prior to the start of the contracts, a baseline household survey was conductedin nine districts. Data on eight asset and housing quality variables was coded and analyzed using principalcomponentsanalysisto constructa wealth index andrankingofhouseholds. Each of the nine districts was then randomly assignedto one of three categories. Two districts were contracted out, givingthe NGOcomplete line managementresponsibility,includingpowers to hire, fire and set wages, procure and distribute drugs and supplies, and organize and staff facilities. Three others were contracted in: inthis case, the NGOworked within the normalMoH systems to strengthen district management,were not able to hire and fire, andreceiveddrugs and supplies through normal MoH channels. The remaining four districts servedas a control, being managed as normal by an MoH District Health Management Team. Precisely defined and objectivelyverifiableperformanceindicators were definedand targets set: one of these relatedto equity, defined interms of deliveringservicesto the pooresthalfo fthe population. Two-and-a-halfyears after the baseline, a second survey was conductedto assess performanceof the three models. All districts were found to have achieved considerable progress, reflecting national reforms in the public health system. Contracted out districts clearly performed best, 'iincludingon the equity goal: concentrationindices for each of the target services established(and probit regressionconfirmed)that contractedout districts hadachievedmost in changing an initial patterno f service delivery that favored the non-poorto one that was more equitable or positively pro-poor. The 2001 survey also found large declines in average out-of-pocket health spending amongst the poorest half of the population in contracted-out districts. With donor support, the Governmenthas now expandedcontracting-outto cover 10percentofall Cambodians. Source: Schwartzand Bhushan2004; Kellerand Schwartz 2001. _ _ _ _ _ _^_I - ~ I. . Exploring these cross-sectoral links is also important in making the case that investments in healthcare promote other aspects of economic and social development. This latter argument is particularly important with regardto large and unpredictable out- of-pocket health care costs for serious injury or illness. Improvements in health service provision and financing could make a major contribution to social protection by (i) protecting against such shocks and (ii) providing affordable and effective treatment when these shocks do occur. Education sector strategy needs to consolidate gains at the primary level, and extend them to higher levels To build on the equity gains already achieved in primary education, the Education Sector Plan needs to continue to tackle both supply- and demand-side bottlenecks in the educationsystem. On the supply side (which is particularly important at the primary level), progress entails sustainingthe stock of school buildings (upgradingthe remainingprimary schools that are incomplete, lacking such facilities as toilets, drinking water or libraries; and building more lower secondary schools) and improving the quality of education and learning. Improving the quality of teaching, especially in poorer and under-servedareas, 138 Sharinggrowth: eguizj and developmentin Cambodia calls for better training that is more directly relevant and better prepares teachers for classroom tea~hing'~.As in health, achieving significant improvements in teacher performance will also depend in the long term upon a substantial across-the-board improvement in salaries, combined with improved human resource management and rewards for good performance. Pre-schoolearly childhood education may also have a significant role to play in reducing over-age enrollment and better preparingchildren for school, so reducing drop-out rates. Finally, focus on including children with disabilities in school. This means in particular that schools must be physically accessible for children with mobility impairments (as most newly-built schools are, but many older ones are not); and that attention is paid to special needs of children with visual, hearing and speaking impairments, particularly when low-cost responses would have a considerableimpact. Demand-side interventions become increasingly important from upper primary through secondary levels, as families are under increasing pressure to take older children out of school to help with livelihood or income-generating activities. Suitable approaches would include expandingthe national scholarship scheme, which targets poor families with cash or in-kind benefits conditional on children's attendance at lower secondary school (grades 7-9). The other main issue with regard to demand-side constraints is the persistence of informal fees levied by teachers in grades 1-9. Eliminating these, however, will require fundamental reform of teachers' salaries and benefits(and better performancemanagement). Finally, there are a number of education system reforms which would help to facilitate improvementin aggregate and equity outcomes. Apart from a service-wide rise and merit-baseddifferentiation in salaries, there is a need to build Government capacity to conduct rigorous impact evaluation o f different initiatives, and so identify priorities and sequences o f policy and program responses to critical binding constraints to further improvement. '' In the Educational Quality ImprovementProgram(EQIP), of a variety of interventionsput on trial in Takeo, Kampot and Kandal, money invested in teacher development was found to have the greatest payoff in terms of student retention and promotion and learning outcomes (World Bank 2005 p. 4). 7. Progress and challenges in gender equality 'Summary Economic growth and change over the last decade has had distinctly gendered effects on employment and incomes. One of the main drivers has been the emergence of the garment sector, where women represent over 80 percent o f the workforce. There is parity inthe number of economically active women and men, and women's income frequently covers a significant , proportionof householdexpenditures. Female-headedhouseholds broadly defined are not worse off than male headedhouseholds in terms of consumption, although their capacity to earn and save income and invest in assets is often constrained, resulting in vulnerability in other respects. This is particularly true for households without an adult male and a highnumber of dependents. As in many other countries, access to education in Cambodia has traditionally been an issue in which gender differences are pronounced, with a strong bias against girls' schooling and literacy. In Cambodia, however, rapid progresstowards greater equity has been achieved over the last decade, as the expansionof primaryenrolment has beenpro-girl. Despite this progress, because of the past, the median grade of schooling completed varies considerably (2.3 female compared to 3.6 male). The gender gap widens at higher levels of education, with three male students to every female student in tertiary education. As returns to education become significant above the primary school level, shortfalls in both overall enrollment rates and gender equity in enrollment at the lower secondary school level and above will tend to widen gender inequities. There are strong arguments for promoting women's education as it is closely associated with higher levels of their own and their children's health status. Severalkey indicators for women's health have shown improvement, such as contraceptive prevalence, a reductionin unmet need for family planning, improved access to antenatal care, and more deliveries by trained professionals and in health facilities.However, the maternal mortality ratio remains highat 473 per 100,000 live births and highlights the importance of continuing the efforts to upgrade both the accessibility and the quality of healthservices, particularlyreproductive health services. The participation of women in decision making positions at various levels is very low but slowly improvingand in some cases approachingthe CMDG targets. However more efforts are ' neededto increasethe number of women Ministers andGovernors. Traditional attitudes combined with rapid socio-economic change have resulted in prevalent behavioral patterns which constitute a serious problem for women's health and opportunities. These include the prevalence of domestic violence, sexual harassment and violation. Trafficking in persons and the commoditizationof women is a on-going concern. There are some indications of a positive shift in norms and attitudes towards support for increasedgender equality, related to the importance of girls' education in particular, as it is linked to increased employment opportunities as well as improvementsin maternal and child health. 140 Sharing growth: equip and development in Cambodia Gender inequalities in Cambodia, in common with much of south-east Asia, have typically been seen as serious but less extreme than in other world regions. Key "traditional" dimensions of inequality in the past have been with regard to girls' education; the double burden faced by women in managing domestic tasks while also contributing to household subsistence or income-generatingactivities; and the low level of female participation in positions of public authority and decision-making. With economic growth structural change and improvements in human development outcomes over the last fifteen years, some of these disadvantages have become less pronounced. At the same time, however, other aspects of rapid economic and social change have resulted in the emergence of new forms of gender inequality. Inequalities exist in access to income generating opportunities and control of household assets and men and women experience different forms of vulnerability to violence: the incidenceof violence against women, inside and outside the home, is high and, apart from constituting a critical development issue in its own right, also has negative effects upon household welfare (assets, income and consumption). Structural inequalities (e.g. lower average levels of female educational achievement) and gender attitudes which support gender inequality persistand are mutually reinforcing. Economic growth has created factoryjobs for women and constructionjobs for men In broad sectoral terms, patterns of male and female work in the dominant informal sector are roughly similar: the vast majority are occupiedin agricultural self-employment, around a fifth work in services, and less than 10 percent are employed in manufacturing. Figure 7.1 Women are more likely than men to be engaged in services in the informal sector, and manufacturing in the formal sector a.informal economy (self-employed, family b. formal economy (employees and employers) workers) - 80% of total employment (83% of female - 20% of total employment (18% of female employment, 76% of male employment) employment, 24% of male employment) o Services o Industry Agricuiture Males Females Males Females Source: CSES 2004. Progress and challenges in gender equality I41 Although growing, wage employment remains limited, accounting for only 20 percent of total employment (24 percent of male and 17 percent of female employment). In this growing labor market there are some clear differences between male and female trends and employment status. Construction, public sector jobs and "other services" are of considerably greater importance to men than women. The vast majority (80-90 percent) of the jobs created in garment manufacturing since the late 1990s are by contrast female: these jobs, now numberingaround 350,000, account for 3.4 percent of total employment and 5.5 percent of female employment. They were also, by comparison to informal and even many other formal sector jobs in Cambodia, "good", being linked through trade quotas and an ILO project to core labor standards. The impact of garment sector jobs on the welfare of employees is however more limited than a simple analysis of their wages would suggest. Around 90 percent of garment factory workers remit a portion of their wages. The proportion that is remitted is typically very high: Dahlberg (2006) found workers remittin on average 35 percent of their earnings, which conforms well with other estimates . F The impact of these remittances is very important to the mainly rural recipient households, in some cases sustaining current consumption (food or health expenditures), in some cases enabling investment in fertilizer, livestock or the education of younger siblings (Dahlberg 2006; AD1 / CCC 2005). To remit this much, however, garment factory workers typically spend very little on their own welfare (food, housing and healthcare). Figure 7.2 In wage labor, women are concentrated in garments; men dominate construction, public sector employment, and other services 1,000000 Garmnts c 5 750,000 Construction - 500,000 Other industry U Ea `kblic Sector' (proxy) 250,000 m Rivate households Other sewices Agricutture Sources: Census 1998, Labor Force Surveys 2000 and 2001, CSES 2004 The distribution of paid employment can be further disaggregated by wealth quintile (Figure 7.3). Employment in public services is dominated by men from the wealthiest quintile. Nearly 60 percent of paid workers in agriculture are fiom the bottomtwo quintiles with nearlyequalnumbersofmalesandfemales. ' AD1 / CCC 2005 found a mean of 27 percent; in the largest survey to date (CRD 2004), 40 percent o f workers remitted 3 1-50 percent o f earnings and 32 percent more than half. 142 Sharing growth: equity and development in Cambodia Figure 7.3 Men, particularly men from the richestquintile, dominate public sector employment rn Poorest UJ Next poorest MI Mddle Next richest Richest 4E LL LL LL Agriculture, Manufacture Construction Other industry Public services Rivate Other services foresty & of wearing households fisheries apparel N ith errployed persons Source CSES 2004 There is still considerable lack of clarity about the distributional nature of garment sector employment, in terms of whether those who receive these jobs come from poor or non-poor backgrounds. Data from the CSES 2004 provides some insights, but does not record enough information on migration or remittances to provide a complete picture of how factory workers now found in Phnom Penh relate to sending households in rural areas. Obtaining access to thesejobs is easier with a certain level of education and some household savings to fund the move to Phnom Penh (and, in about one in six cases, a bribe paid to secure the job), which suggest that the poorest are indeed likely to be underrepresented amongst factory employees. The CSES data finds 8.7 percent of households to have one or more garment worker (21 percent in Phnom Penh, and 8 percent of rural households. Relatively few (12 percent) of the poorest quintile have a family member working in the sector. Findings from factory-basedworker surveys and detailed case studies which link workers via remittances to their home villages suggest a somewhat greater distributional impact of garment sector jobs (ADB 2004; Dahlberg 2006). In terms of equity effects, however, it seems that the greatest impact of the growth of the garment sector may be in promoting gender equality rather than consumptionor income equality. Expanded service delivery has closed gender gaps, particularly in education Access to schooling and education outcomes such as literacy have traditionally been one of the aspects of Cambodian gender relations in which inequality was most pronounced. As described in the previous chapter, contact with public services and Progress andchallenges in gender equality I43 human outcomes Figure 7.4 Educational outcomes amongst have improved considerably over recent school leavers show the last decade. Progress has been much less inequality particularly good in regard to education, where the increase in primary enrolment has been 5.8 5.7 markedly faster amongst girls than 1 , boys. The result is that educational % attainment-as measured in, for - example, age-specific median E . values o f highest grade completed-is near equal for those in or recently leaving school. While inequalities emerge amongst those aged 20 or above, and the ratio of male to female median 6 9 10-14 15-19 20-24 25-29 30-34 grades for the population as a age (years) whole is high at 3.6 to 2.3, female Source:CHDS 2005 p. 11-12. educational status in the next generation should be much higher than in their parents' generation, and much closer to that of their male contemporaries. The next challenge then is to improve aggregate outcomes and gender equity above primary school level, where participation, particularly of girls, drops off markedly. At the lower secondary school level, the ratio of girls to boys is 77:lOO; at the upper secondary school level, the ratio has slightly exceeded the target o f 59:lOO for 2005. Net enrollment was however only 9 percent o f the age group and the vast majority of the students were from higher income groups (74 percent o f upper secondary school students are from households in the highest two income quintiles). Finally, at the tertiary school level, women are outnumberedby men 340-1. While this marks an improvementover the 3.3-to-1 ratio in the current work force, it implies that gender disparities in professional and management occupationswill persist in the medium to long term. As higher returnsto education are discernable primarily above the primary school level, shortfalls in both overall enrollment rates and gender equity in enrollment at the lower secondary school level will tend to perpetuateboth income andgender inequities. A notable health concern for women-and a lagging indicator amongst the Cambodian Millennium Development Goals-is the maternal mortality ratio (MMR). Improvements in preventive and curative healthcare have benefitedwomen as well as men. However, while a number of service utilization and intermediate outcome indicators (e.g. the percentage o f pregnant mothers who receive antenatal care, and the percentage of deliveries that occur in a health facility and / or with the assistance of a trained health professional)have improvedin the last five years, this has yet to result in improvement in the MMR. In part, this may reflect the limitations and lags in the MMR as a measure. Evenaccounting for this, however, it is clear that progress in ensuring safe motherhoodis beingachieved at a slower ratethan is the case for other healthindicators. 144 Sharing growth: equity and develoument in Cambodia Men and women face different forms of insecurity Prevalence and tolerance of gender-based violence remains high In common with many post-conflict countries, Cambodia has a high rate of violent crime. Men tend to experience public violence from male gang members, peers, and authorities, while women are more likely to experience sexual violence and assault, including in particular domestic violence, from men (MoWA 2005). A series of studies conducted over the past decade (Zimmerman 1995; MOWVA and PADV 1996; MOWA 2005), including most recently the CDHS 2005, have estimated the proportion of ever- married women who have ever experienced violence at between 13 to 23 percent (Table 7.1): it is safe to assume that, as in most countries, the incidence may be higher given the tendency for domestic violence to be underreported. Table 7.1 Estimates of domestic violence, 1996-2005 Year Organization Sample % of respondents who have ever experienced physical sexual violence emotional, violence physical or sexual violence 1996 MOWA 1,374 women in six provinces 16 16 2000 CDHS 15,351women aged 15-19 16 4 25 2004 MOWA 1,580 women aged 15 and over 23 2005 CDHS 16,823 women aged 15-19 13 3 22 In most cases of violent crime but particularly in the case of violence against women, institutional responses do not deliver effective justice. Domestic violence is often not seen as a criminal or serious offense by police and local authorities with the result that perpetrators are not punished. The Ministry of Women's Affairs (MOWA) made respondingto domestic violence a high priority, with a new law adopted in 2005. The impact of domestic violence, particularly on the poor, is severe. In the 2005 survey, 24 percent of the poorest group in the sample (those earning less than $50 a month) who had experienced domestic violence said they lost two months income as a result of violence in the 12 months preceding the survey. A fifth of respondents experiencing violence reported that their children had missed school between five to twenty times in the past year as a result of the violence (MoWA 2005: 37). While tolerance is high, there are some signs that attitudes may be shifting with domestic violence coming to be seen as a public rather than a private issue in at least some communities (Box 7.1). The main factors which appear to contribute to changing attitudes are media education, NGO training programs and, in some cases, greater intervention by local authorities, including fines, arrests, and `advice' to the victim and perpetrator. Progress and challenges in gender equality 145 Box 7.1: Qualitative research suggests a decline in domestic violence in some villages In recent CDRI studies (MOPS and the PPA), there are suggestions of a decline in domestic violence in some communities. In five of twenty four PPA villages, domestic violence was reportedto be declining: it was seen as increasinginthree others, and static inthe remainder. In six of the nine MOPS study villages, both male and female participants reported that the , incidence o f domestic violence had declinedsignificantly in recent years (with slower change or no change in the other three villages). This improvement was attributedto the impact of social and legal awareness campaigns, facilitated by the spread within the villages of television and radio. In both MOPS and the PPA, the decline was credited to humanrights trainingworkshops on domestic violence delivered by NGOs. Also important was the perceptionof stronger laws against domestic violence and of local authorities increasingly willing to intervene(whichis itself a product o f public education and legal awareness campaigns). The situation i s still far short o f ideal; the emphasis of local police andauthorities, as inmuchof "living law" disputeresolutionat the local level in Cambodia, is on reconciliationand relatively symbolic punishment through a few days confinement injail, rather than the strict application of the law. Nonetheless, in some communities the combination of new laws and active rights education appears to have led to an improvement upon the previous situation, when domestic violence was clearly seen as a family matter in which it was difficult or impossible for outsiders-both other villagers and local authorities-to intervene. Before, when a husband hit his wife, the wife didn 't dare complain at thepolice station, but now, after people supporting women's rights have been here, the wife claims her rights at the police ~ station to have him arrested and warned, (KhsachChiros, KompongThom Province:MOPS) Now, we have much less violence in our area than the past Jive years because some actions1 Participation in decision-making Most household decisions are madejointly Joint decision-making is the norm in Cambodian households: for most of the sample issues on which questions about decision-makingwere asked in the CDHS, husband and wife jointly make the final decisionjointly (Figure 7.5). When decisions are not made jointly, it is more likely to be the wife rather than the husband who has the final say. The notable exception, where more conservative attitudes are more apparent, is with regardto decisions on whether or not the wife should work to earn money where, once again, the decision is madejointly in the majority (54 percent) of households: but, when it is not a joint decision, it is more likely to be made by the husband (21 percent) than the wife (16 percent). Amongst the subsample of currently married women who had one or more living children at the time of the survey, three-quarterswould make decisions relating to their children (e.g. on schooling, what to do if a child was sick, and whether or not to have another child) jointly: but once again, if the decision was made solely then it was muchmore likely to be made by the woman rather than the man. 146 Sharing growth: equity and development in Cambodia Figure7.5 Most household decisions involve both husbandand wife 1001 ?6 C l e purchases 3 75 own healthcare 0 s5 sE 50 Q visits to famly or relatives L?! whether w f e shouli work to earn money * 5 25 s large household purchases 0 whether to use Wfe only Wfe and husbandjointly Husband only 1 contraception Source: CDHS 2005 p, 269. Box 7.2 Women tend to be responsible land or livestock) are often joint or led While investment decisions (e.g. on ,' for managingborrowing traditional practiceswith regardto credit ' by the husband, research confirms that In the Khmer household, the wife and the husband have to discuss with each other before still prevail: men are generally agreed to , going to borrow moneyfiom the creditors. The find asking for a loan embarrassing, and wfe's responsibility is to borrow money and leave these negotiations to their wives, the husband is responsible for earning money who are Seen as more competent in to repay the loan. The husband doesn`t want to managing such financial affairs (BOX do the borrowing at all because he is ~ 7.2). However, this does not necessarily embarrassed. (Village report, KOah Ta ensure that she has a say in how the I Village, KompongThom Province:PPA) I money is used; and in some cases Source:CDRI2007a. I I women become liable for debts owed by ; I I-- their husband. Women`s participation in public life is increasing, but still low Women's participation in public life in Cambodia is extremely limited (Table 7.2). There are some positive trends-following the 2007 Commune / Sangkat elections, women hold 15 percent of seats in Commune / Sangkat councils (up from 8 percent in 2002) and 4 percent of commune councils have a woman commune chief (up from 0.8 percent in 2002). However, overall rates of political participation are low (Table 7.x), giving it one of the lowest gender empowerment scores in Asia (Urashima 2007, UNIFEMet ~12004). Research on the responsiveness of local institutions found that women's participation in Commune / Sangkat councils is explained by their low levels of education, family responsibilities, and social attitudeswhich regardpolitics as somethingfor which women are not suited-attitudes which clearly undermine the confidence of women councilors who do participate(Rusten et a1 2004). Progress and challengesin gender equality 147 Table 7.2 Women are under-represented However, women's participation in in all aspects of government community level meetings (such as the commune planning process) tends to Public office % of posts held by be equal to or greater than that o f men women (Rusten et a1 2004, Brereton 2005). Legislature Separate meetings for women and the NationalAssembly member 16 opportunity to participate in Senators 13 discussions about local issues which Communecouncilors 15 are immediately relevant to women's Communechiefs 4 Executive lives appears to support their Ministers 5 involvement. The fact that women are Secretaries of State 6 more likely than men to be in the Under-Secretariesof State 12 village during the day-which makes Provincialor MunicipalGovernors 0 it harder for them to run for office, Judiciary Judges 12 take leadership roles or travel to take Prosecutors 2 part in meetings-does make it easier Civil service for them to participate in meetings at Civil service employees - of which 31 the village level. Women do not senior Dositions 9 usually participate in planning processes above the commune level: thus for example in 2004 only 17 percent of participants in District Integration Workshopswere women (Brereton2005). Women's participation in civil society and community decision-making outside of elections is generally low (albeit somewhat higher among older women who no longer need to look after for children). Although women entrepreneurs do exist and women often play a key role in managingfamily businessesamongst the wealthier quintiles, men dominate not just public but also private sector institutions involved in economic management, including the private sector forum, chambers o f commerce, market management, and the judicial system. There are few women in management positions in either the public or private sector. Civil society development: from gender training to a women's movement Levels of associational activity are rather low in Cambodia, for both sexes: only 6.4 percent of women are a member o f any association, club or organization. Cambodia traditionally had few permanent village-level groups (Ebihara 1968; Delvert 1961; Conway 1999; Aranvind 2004). However, since the early 1990s there has been a rapid if uneven expansion o f civil society and there are now a number of organizations working on women's issues who specifically target women as beneficiaries. To date, however, much o f this has targeted women primarily as simply recipients of training. For the most part, activities are focused on gender awareness training; women in politics, particularly commune councils; HIV/AIDS; women's legal rights; and trafficking and domestic violence. Although relevant, this set of activities has largely reflected donor and INGO priorities. There are no grassroots movements advocating for gender-responsive action on other key gender concerns such as maternal healthservices or female education and there are few organizations supporting the economic advancement of women-or men for that 148 Sharing growth: equity and development in Cambodia matter-outside o f credit, handicrafts and traditional vocational training. There remains resistance to the term `women's empowerment' as this is generally interpretedas taking power away from men and against traditional culture. 8. Institutions and equity in the development process Summary .Contemporary economics emphasizes the importance of institutions-both formal structures and the values and practices in which they are embedded-in explainingnational development outcomes, including equity and equality outcomes. Internationalexperience suggest that state institutions that reflect the interests of poor and promote equality of opportunity foster more stable politics and more sustainable long-termgrowth than those which concentrate power and ' wealth. Building such institutions is often hard in post-conflict states. In Cambodia, surveys identify informal fees as a critical burden on firms, and reveal corruption and lack of trust in .state institutionsas major problems. The state can foster equitable development in the way it raises and directs public finances. Cambodia's public financial management (PFM) system has many serious problems; however, progress in PFM reform has opened up possibilities for more effectively and efficiently , directing public spendingtowards poverty reduction. Priorities includeputting in place strong, ' transparent measures for using revenue that is expected to flow from oil and gas within a few years; reallocating spending towards priority goals; raising capital spending, emphasizing '' economic sectors (roads, irrigation, energy) that shape distribution of opportunities and the pattern of growth; and increasing recurrent spending on maintenance (to sustain the value of capital investments) and, critically, on public sector wages. Building RGC institutional capacity through increasedcivil service salaries and better managementmay be as importantas .investingin infrastructure. Besides spending, the state guides growth and poverty reduction by formulating and implementing rules that establish the conditions for interaction between economic and social actors. In pursuing equity objectives, liberal democracies rely heavily on the rule of law-that is, governing society with clear rules applied fairly to all. In Cambodia, state law i s not yet established as the primary source of authority: often, it must compete for legitimacy and influence with other rule systems. Cambodia's legal andjudicial systems are not currentlyvery effective or supportive of equity: the law is often applied selectively to enforce decisions of dubious legitimacy or protect influential individuals from prosecution. Building the rule of law involves fundamental change in how society is ordered and power exercised. Incentives to reform may reflect international processes (e.g. trade negotiations) or political considerations (e.g. concern regarding popular mobilization around land issues). Legal and judicial reform strategies need to understand that the rule of law emerges as much from processes of social .contest as it does from technical change. Progress in Cambodia has often been achieved in enclaves, aiming to develop robust institutions and later build outwards, rather than attempting to implement comprehensive institutional reform in one round. Thus, ties to trade quota agreements created incentives for improving working conditions in the garment industry; and principles of participatory and accountable local planning, introduced via donor-supported "project law" under the Seila program, have taken root and come to influence RGC thinking on decentralization and deconcentration policy. In both cases, outcomes fall short of the ideal envisaged under concepts such as the `rule of law', but do mark a significant advance on what existed before, and may open up possibilitiesfor future stepwise-_progress. " - - - -- - - - __- - - __- - -- - -- I I - __ II_I_ I__ -I--I -.__ - __ -" 150 Shan'nggrowth: epio and developmentin Cambodia Institutions and development: state responsiveness, legitimacy and the policy process- Market institutions are Box 8.1 What do we mean by "institutions"? embedded in and depend for their The term "institution" is awkward, as it is used in effectivenessupon a much broader English in at least two different ways. In academic set of social and political terminology (e.g. inthe New InstitutionalEconomics institutions. (Box 8.1 discusses literature), there is a commonly accepted distinction what is meant by "institutions".) A . between institutions as the "rules of the game" and wide variety of organizational organisations as the "players"'. In daily usage, actors interact to shape economic however, the term institution is often used and social relations, and help interchangeablywith organisation, so that a Ministry determine the degree of equality or o f Health, an NGO, or an inter-Ministerialcommittee inequality of opportunity that are also described as institutions. Bratton and de different individuals and groups Walle explicitly accept that the term can be used in bothsenses: face in participating in marketsand obtaining access to services. political institutions can be highly abstract notions, These include the courts and such as constitutional principles, or they can be police; the media; the executive expressed concretely in actual organisations, such as and legislative bodies; the civil trade unions, political parties, or the military. They include key aspects of formal politics, like the service; political parties; civil judiciary, but also informal customs such as society organizations based on patronage, clientalism, seniority principles, or principles of collective action or lobbying'. charity; and private sector associations that seek to lobby Huntingtonlinks institutionsand organisations in an interestingmanner, defininginstitutionsas "expected policy-makers on the basis of patterns of behaviour", which helps to establish the collective business interest. The distinction between organisations that are orientation and effectiveness of institutionalised(ie. predictable in their procedures) these bodies are in turn rooted in and those that are not. O'Donnell argues that institutions (practices and norms), consolidated democracy is institutionalised such as the constitution and laws; democracy-ie. political actors work on the the rules (formal and informal) of assumption that democratic processes will continue the policy process; socially-defined to be the major means o f managing conflict in gender roles; national, ethnic or society. He also draws a distinctionbetweenformal regional identities; and values and institutions (elections, parliaments, courts) and expectations with regardto vertical informal institutions (patron-client networks, influence of elders)'. This helps to frame an analysis (patron-client, class, administrative o f the policy process, to the extent that regularised hierarchy) and horizontal political frameworks are more accessible than (solidarity, cooperation) socio- discretionary andunpredictableones. economic relations and forms of social organization. Institutionsand their organisational expressions help to provide a concrete focus for the examination of Institutional arrangements that policy choices, in both the short and long term. By are conducive to long-run setting the rules of the game (and determining which economic growth and structural actors can play) institutions both perpetuate and transformation tend to be those transform the natureof state-societyrelations. Institdons and equity in the developmentprocess 151 which exert checks and balances and limit the natural tendency for economic and political power to reinforce each other. To put it another way, economieswhich succeed in achieving stable long-run growth tend to have institutions which put limits upon the growth of inequality in political voice and power. By contrast, when economic and political power are both concentrated in the hands of a small and unaccountableelite that lacks either the vision or capacity for sate-directed development, the result is generally "bad" institutions which protect the property rights of in-group members but not others; are inconsistentin enforcing contracts; focus upon extracting rents rather than identifying new investment opportunities; and ration market entry and access to capital and information. This is reflected in the World Development Report 2006, which argues that institutions tend to evolve in vicious or virtuous cycles. Less equitable social structures lead to the formation of institutions that perpetuate inequalities in power, status and wealth; conversely, equitable institutions only emerge in situations when the distribution of power is not highly unequal. In these terms, one dilemma of development is how to bump a society from the vicious cycle of self-reinforcing inequality to the virtuous cycle of self-reinforcing equity (World Bank 2005 pp. 8-9). Inequality in the ways institutions treat different groups of citizens also matters for legitimacy and governability. If it is widely believed that wealth differences result from unfair rules, this undermines people's faith in politicians and officials, making it harder for Government to retain people's loyalty and ensure their compliance with law and policy (see Mekong Economics 2006 for a discussion of inequality and legitimacy in Vietnam). High inequality also undermines the trust required for citizens to work together to solve their own problems (Labonne et a1 2006); and is clearly related to the incidence and severity of crime (Demobynes and Ozler 2005). Internationally, there is a relationship observed between high levels of inequality and weak (unfair) institutions associated with these levels of inequality, on the one hand, and the increased likelihood of social and political instability, crime, and violence, on the other (Bourgignon 2004 citing Rodrik 1998). This is especially the case when the inequalities in question are horizontal or group-based inequality (e.g. between regions or ethnic groups: Killick 2002). Conflict also imposes opportunity costs, in the form of resources (public and private) diverted from more profitable activities into preventing and fighting crime. Beyond a certain threshold, all these act as deterrents to investment, lending and economic growth. Post-conflict recovery and institutional evolution in Cambodia That Cambodia demonstrates many of the kinds of institutional shortcomings described above (see Box 8.2) is unsurprising, given its legacy of conflict. At the collapse of the Khmer Rouge in 1979, all political institutions were destroyed. The People's Republic of Kampuchea (PRK) thus had to rebuild the institutions of the state from the ground up, in the context of a civil war against foreign-backed insurgents; extreme shortage of material and human resources (amounting in the first years to economic collapse and incipient famine); geopolitically-inspired exclusion from internationaltrade and investment (and most of the internationalaid system); and a weak foundation of popular legitimacy. In this context, it is unsurprising that the political 152 Sharinggrowth: egtlig and developmentin Cambodia institutionsthat emerged after the Khmer Rouge prioritized the basic task of forming and protecting the state over more advanced aspirations to transparency, accountability or democracy (Hughes 2003; Gottesman 2003). These institutions have evolved considerably since 1979 with the shift to market economics (beginning in 1989) and, in the Paris Peace Accords of 1991, the end of Cambodia's civil war and agreement on arrangements for the transition to multi-party democracy'. The abrupt transition to market economics and insertion of Cambodia into regional and global economies enabled officials ofthe PRK (from 1989 to 1993 re-named the State of Cambodia or SOC) to establish an early privileged position in the marketplace, bringing political position and economic power together to a degree that had not been possible under the enforcedslow-growth isolation ofthe 1980s. Box 8.2 The quality of local governance is a major, generally negative influence upon opportunities, which will probably require national solutions "The MOPS demonstrates that good governance is hndamental to development and poverty reduction. Growth, developmentandpoverty reduction can still take place inthe context of poor governance and weak institutional capacity, but are slowed, or even halted by corruption, impunity and weak rule of law. This point was made again and again in study villages; unpredictable, unreasonable, and crippling corruption costs are actively preventing many householdsfrom going abouttheir daily businessandearninga sustainableincome. "The RGC is committed to strengthening local government through the de-concentration and decentralizationprocess. The MOPS suggests that strengtheningthe financial and administrative capacity of local institutions is important as there is more that local authorities can do, for example to manage local conflicts, involve community members in decision-making, reduce domestic violence and so on. But this will only be effective up to a point, as local authorities are effectively powerlessto act once powerful, external influencesare involved. "Significant governance failures, in particular in relation to natural resourcemanagement, cannot be addressed at the local level and require a national response. Political will is required to implementgovernment policies and impose consequences onthose who breakthe law." Source:CDRI 2007b, forthcoming. Acting as a partial countervailing centrifugal force to the power-centralizing effects of unregulatedeconomic liberalization,the 1993 Constitution laid out a framework vision for a liberal democratic state. Over the last decade reform-minded actors in Government and civil society, together with the UN, donors and internationalNGOs, have sought to use this framework to create political space for the evolution of an independent civil society, free press, and the separation of powers. The RGC has made significant commitments to principles of equitable development, as seen in the Rectangular Strategy (RGC 2004), the NSDP 2006-2010 (RGC 2006), and the CMDGs (many of which relate explicitly to equity targets). Through allocation formula for the communehangkat fund and an ongoing process to develop a national methodology for low-cost identification of poor households, it is also now developingpracticalmechanismsto meet NSDP goals for better targeting poor areas and poor households. I The Khmer Rouge subsequently reneged on this peace agreement and returned to armed oppositionto the state. The last pockets ofthis opposition finally surrenderedonly in 1998. Institutions and equip in the developmentprocess 153 Legal pluralism and governability A striking feature of Cambodia is the prevalence of what may be described as legal pluralism: a situation in which several, often contradictory and competing, rules and norms influencesocial, economic and political relationships. This situation is not unique to Cambodia: many developing countries are characterized by the existence, in parallel, of traditional, religious, administrative, legal systems, each with different (and unreconciled) definitions of rights, duties and principles and institutions for mediating disputes (Moser et a1 2001). Legal pluralism typically becomes less important at later stages of economic development: a well-resourced state, sustained by a broad tax base and kept in tune with popular concerns and moresthrough democratic elections and other channels o f accountability, has a greater capacity to project a legitimate definition o f law (and to define law in such a way that aligns popular and formal concepts of rules and norms). Box 8.3 'Politicalwill' and challengeof institutionalreform Difficulty with governance and particularly andjustice sector reform in Cambodia is regularly ascribedto a "lack of politicalwill" (e.g. Global Witness 2005; US Embassy 2006; AHRC 2007). This concept is at once illuminatingbut also limiting. On the one hand it correctly identifies that reform of these key sectors is an inherently political process in that it involves a redefinitionof how power is exercised within society. On the other hand, the concept of a lack of "will" is reductionist in that suggests that reform could occur through a simple change of sentiment. This underestimates the extent to which current arrangements are maintained by deeply interwoven and self-reinforcingsystems of norms and incentives. The difficulty of achieving governance reform in a country like Cambodia is thus better described as a problem of systems and the powers that support them than a problem of will. The question in terms of how more equitable institution of governance could be developed is thus not so much about how will is created but how complex systems are transformed. Low income developing countries, by contrast, often demonstrate a degree of legal pluralism, in which national laws are a significant but not exclusive reference point when actors negotiate rights and duties. Legal pluralism is particularly pronounced in countries, such as Cambodia, which have in a relatively short span oftime been subject to a succession o f legal frameworks (pre-colonial, colonial, socialist and liberal- democratic); and / or where the reach of state law has been weakened due to war, revolution and rapid re-insertion into regional and international markets and political- legal frameworks (e.g. the international human rights system built around the UN; WTO rules on trade and commercial law; and so on). In Cambodia, constitutionaland statute law often compete both with (i)social norms (e.g. a historically deep-rooted concept that vacant land can be clearedand occupied and that the act of farming confers land use rights) and (ii)neo-patrimonial administrative conventions, in which the formal distinction between public and private property is blurred, and underpaid Government officials assume the right to use their office to generate rent (e.g. from informal charges for services that are officially free) or to authorize transactions (including lease or sale) of state assets (e.g. land) that fall within their jurisdiction. Ina situation of legal pluralism, parties will typically engage in "forum shopping": that is, selecting the legal forum and underlying normative principles which 154 Sbaringgrowth:eguig and developmentin Cambodia best legitimizes their claims. Unfortunately, the wealthy and educated are better positioned to shop for a forum supportive of their case, and to deploy the power of the state to back up their claims. Qualitative researchclearly revealsa widespread perception of an imbalance in power, and the consequences of this in terms of inequality of opportunity (Box 8.4) ." -- - - ._ I ___I___-_I__II___I_--l-_l______ I ,Box 8.4 "Rights in practice": power confers greater opportunities for the rich I ,Cambodiansociety has traditionallybeen orderedin a hierarchical manner, even ifrelativestatus and power in any given relationship was a composite of many different factors. Asymmetry in rights and the pointlessness of resisting this inequality are reflected in many traditional Khmer proverbs, such as "The egg cannot fight the stone". While recent qualitative research reveals increasingpopular understandingof the concept of equality of rights that is the central principle o f a liberal state, it also reveals widespread agreementthat this situation has yet to be established in Cambodia, given that the richandthose with connectionshavemorerightsinpractice. Thepeople are poor and ignorant. That is why they [the powerful] grab land and abuse rights. Even if the people know the law, but they are poor, there is still nojustice in dispute resolution, because the others have power, money and they also know the law. Pagodacommitteemember, Prasat Sambo District,quotedinCAS andWorldBank2006 p. 31. At that time the villagers did not complain because there was no way. They did not know to whom they should complain when the village chief refused to help. Informal leader, Prasat Sambo, quoted in CAS and WorldBank 2006 p. 20. , Iampoor: ifa problem arises with a wealthyperson Iprefer todrop the issueso thematter will not go too high, because I know I will lose. I try to avoid problems with the rich; they have , power. Woman, 50, Svay RiengProvince, quotedinLuco 2002 p. 98 Sources: CDRI 2007a, 2007b, CAS and WorldBank 2006 p, 16.- _ _ - - ~ - - - - ~ I _ I I_ I I_ I -I This theme of legal pluralism, and the challenges and opportunities it creates for moving to equitable, pro-poor institutional formations, is addressed later in this chapter in the context of labor law and local governance reforms. Promoting equity through public spending Public financial management (PFM) is not the only tool by which Government shapes economic and social outcomes: the state also regulates economic and social activity through legal and regulatory functions, through monetarypolicy (setting interest rates and otherwise controlling inflation), and through influencing identities, values and ideology. Nonetheless, control over the level and allocation of public spending (and how it is financed) is typically the most important lever by which to influence outcomes, including equity, both directly and indirectly (see Box 8.5)2. Correspondingly, while there are lenses other than PFM through which to examine a state's effectiveness in achieving goals, analyzingpatternsand trends in public financial management is the logical starting point. 2 The role of state spending may be particularly influential when state revenues are large relative to the real domestic economy as a whole: this is clearly not the case in Cambodia at the moment, but this may well change in the near future once oil and gas revenues start to come onstream. Institutions and eguitv in the develobmentbrocess 155 , , Box 8.5 Core concepts in public expenditurepolicy In theory, any productive and efficient expenditure within an economy should, all things being ' equal, contribute to higher growth: in practice, however, the link between public (i.e. Government) expenditureand growth is a topic of debate. The two key issueso f interest are (i) the level oftotal publicspendingand (ii) the composition ofthat spending, interms ofthe balance betweendifferentsectors andcategoriesof spending. To assess whether the level of spending is appropriate requires comparing aggregate spending to the size of the economy and analyzing the sources from which this spending is financed. The spending-to-GDP ratio is important because high levels of public spending can fuel inflation, I affect the value of domestic savings, and potentially crowd out private investment. The sources ' from which spending is financed (domestic tax and non-tax revenue, borrowing, and ODA) i matters because the benefits of taxation-fundedpublic expenditure may or may not exceed the 1 costs (economic and political) of taxing, if high levels of taxation deter economic activity and ' , impede the efficient operation of markets, resulting in a slower rate o f economic growth for which tax-fundedpublic spending only partially compensates. Spendingfinanced by borrowing, ' meanwhile, stores up public debts which may come to undermine the sustainability of public finances and macroeconomicstabilityin the future. The composition of public spending, meanwhile, needs to be guided by a different set of trade- ' offs. The first i s between general types o f sector (e.g. economic sectors vs. social sectors vs. i general administration), sectors (e.g. health vs. education, or roads vs. irrigation)and sub-sectors (e.g. vertical disease control programs vs. primary health care systems). Here, the balance of 1 spendingneedsto reflectanalysis of the returnsto differentpotentialuses of limitedpublic funds: I for given economic and social objectives, is the Governmentbetter investingan additional dollar in healthcare, or in irrigation? The second set o f trade-offs is with regard to the balance to be struck betweencapital(new investment)and recurrent(wageandother runningcost) expenditure. Public expenditure policy can be an important influence upon equity and efficiency within a given economy. As discussedearlier, there is a broad spectrum of policy options within which the goals of increasingefficiency and increasing equity are compatible and even on many grounds complementary. There are also, at the extreme ends of the spectrum, positions which may achieve short-term equality at the cost of suppressing markets and slowing growth; or which focus exclusively on maximizing short-term aggregate output, with no regard for the inequality of the distribution of that wealth. Neither of these polar positions is likely to be sustainable, or desirable, over the long term. Gains in revenue collection would enable higher, pro-equity spending Cambodia enjoys considerablepotential to expand spending The fundamental problem with Cambodian public expenditurepolicy is that the level of public expenditure is extremely low, severely limiting the potential for using spending to influence economic or social development. Public spending as a percentage of GDP has ranged from 17.6 in 2002 to 14.4 in 2005, averaging 16.3 percent between 2002 and 2005. This is low by any standard, including comparison with other low income countries: from 1995-2004 Cambodia's total expenditure averaged 15.2 percent of GDP, 156 Sharinggrowth: equig and development in Cambodia compared with the low income country average of 25.1 percent3. Of 49 low income countries for which data is available, Cambodia ranked sixth lowest in total public spending proportional to GDP, out-performing only the Democratic Republic of Congo, Haiti, Sudan, Liberia and Myanmar (see Taliercio 2006 p. 4). Over this period, inflation has remained low and there is no evidence that public spending has crowded out private sector investment. In other words, there remains considerable "fiscal space" within which the Government could significantly increase public spendingbeforethere is any risk ofjeopardizing macroeconomicstability. Indeed, it appears that the over-riding priority in Government's fiscal policy is to sustain macroeconomicstability, and that it pursues this-to the detriment of other objectives- first and foremost by containingexpenditurewithin a (very restrictive) revenueenvelope. To spend more, however, the Government must first raise more revenue For public spending to be an effective tool of pro-poor, pro-equity Government strategy, then, requires that significantly more revenue is collected. Improved revenue collection is emphasized because increases in the other sources of finance potentially available for public spending(namely ODA and borrowing) are presently either not likely or not advisable (Box 8.6) and because revenue collection in Cambodia is very low in comparison to other low income countries. From 1995 to 2004, revenue-to-GDP averaged 6.8 percent in Cambodia, comparedto an average of 14.2 percent for other low income countries, placing Cambodia third from bottom amongst these 5 1 countries (Figure 8.1). From 2003 to 2005 Cambodia's total revenue averaged 11.1 percent of GDP (7.1 percent tax revenue, 2.8 percentnon-taxrevenue). 3 In these calculations, revenue figures do not include ODA, with the exception of budget support. Institutions and equip in the developmentprocess 157 Box 8.6 Non-revenue financing of public spending I I Official development assistance (ODA) Cambodia currently makes heavy use of ODA, which finances nearly half of total spending. Given the highly concessional nature of assistance available to Cambodia, it makes sense for the Government to seek and utilize as much ODA as possible to finance its spending program.In the short term, however, Cambodia has little control over the level of aid flows, which are largely determned by exogenous factors (e.g. aggregate amounts of ODA made avadable by donors). Some sources (viz., those from the World Bank, ADB, and the US) are influenced by RGC performance indcators, and as such could be influenced by the Government if it further improvedits policy and institutionalenvironment. However, there i s a problemmth lagged measures of performance, as current levels of ODA are determined by performance in previousyears. As a result, it seems likely that at best, Cambodia's total "tradiuonal" ODA would likely remain about the same inreal terms over the next five years. Camboda's amount of ODA per capita (c. US$ 46 in 2006) is already very high in comparison to other low income countries (e.g. the Sub-SaharanAfrica regional average of $28 per capita). However, it i s possible that aid from "non-traditional donors," such as China, could increase. In 2006 China committeda relativelylarge sum of US$600 million to I Cambocfia (although this amount includes quasi-concessionalproject financing over a multi- year period as well as private investment). While it is not possible to estimate with any precision the amount given in grant aid from China and other non-tradtional donors on an annual basis, it i s reasonable to assume that this amount d 1 increase over the next three to five years. Debt The third source of financing available to the Government is foreign and domestic debt. Cambodia's relativelyhigh debt stock is largely due to a historicallegacy of borrowingfrom the US (1 970s) and USSR (1980s): the recent debt policy of the Government has been prudent. All new external debt is concessional,and domestic debt, at only 3.4 percentofGDP, is contractedon amounts andterms conduciveto sustainability. Overall, Cambodia's debt is sustainable:although the risk of debt distress is high, a concessional rescheduling with both the US and the Russian Federationwould be likely to bring sustainability indicators back within safe limits. There is no evidence that debt policy is inappropriate in terms of public expenditure policy. In fact, the Governmenthas recently startedto consider borrowing small amounts on less concessional terms for highreturninfrastructureinvestments. -- __ __ - - __ __ - - _- ___ -- -- - - -- -- ___I I I This comparisonwould suggest that, rather than worrying that excessive taxation may generate economic costs, Cambodia in fact can and should increase revenue collection in order to finance higher levels of spending. A package of administrative and policy measures (e.g. introducing a modern property tax, raising the VAT rate, and improving tax and customs administration) is needed to boost Cambodia's revenue to GDP ratio beyond its currently stagnated level. Similarly, the current structure o f tax revenue is unlikely to be overly distortionary and while Cambodia's non-tax revenue structure (composed mainly of user fees and concession charges) is not notably well designed, the problems with this structure (particularly the concession charge component) is that it tilts 158 Sharinggrowth: eguig and developmentin Cambodia in favor of under-charging (often generating inefficiencies in the overexploitation of natural resources due to low pricing structures), not that it constitutesan excessive charge upon the private sector. In summary, neither the magnitude nor the structure of revenue collection rank as serious problemsfor economic development. Figure8.1 Cambodia ranks 49 out of 51 low income countries for revenue collection - revenue collectionas a percentageof GDP (average, 1995-2004) r ~Mo0o.a o e *sian ) 1 Congo Rep j 1 ZmDaDw L 1 Mongolia Papua New Guinea SolomonVietnam Islands Nigena rten)a j 1 Maiah Samba Tne Sao Tome ana P m c w C3ie a do re Cameroon Ma Benin hca'aaaa Tajikislan i I Sudan Togo Pakistan Comoros Mozambique Burkina Faso Uganda Ethiopia Guinea Tanzania -ao PDR i Rwanda Maoagascar i 1 1 S erra -eole I Yemenheca j-t Rep haer ,-I Bhutan I Central Afncan Republic Haiti Bangladesh Guinea-Bissau Cambodia Chad Congo Dem Rep 0 5 10 15 20 25 30 35 40 Source: Staff estimates based on World Bank live database Efficiency gains are possible at either present or higher levels of spending If the first challenge in making public spending effective as a tool for promoting equitable development (or indeed any other policy goal) is to improve revenue generation, fiscal space can also be expanded by improving the efficiency of spending (i.e. seeking to achieve the same level of output with less input, or greater output with the same input). Weaknesses in PFM systems not only create unacceptable levels of fiduciary risk but also impose high costs in terms of allocative and operational inefficiency. Examples o f wasteful spending range from moonlighting civil servants to relatively expensive administrative overhead costs to misdirection of financial resources Institutions and equig in the developmentprocess 159 to non-priority uses. Cash-based payments have been a major constraint: budget execution has suffered from delays and unpredictable release of funds, undermining operational planning and resulting in the build-up of arrears. The system is plagued by gate-keeping and deficient accounting and reporting systems, leading to a weak control environmentand increasingopportunities for corruption. Cambodia's system ranks below average for low incomecountries, indicatingthe need for substantialupgrading. The Government's Public Financial Management Reform Program (2005-20 15) is focusing first on making the budget credible and reducing fiduciary risk (by improving budget planning and executionand strengtheningthe control environment). The Program is increasing use of the banking sector, improving planning and prioritization in cash management, and developing a financial management information system (FMIS). Strengthening the internal and external control environment will require not only an appropriate legal framework but also an enhanced public accounting system, a culture of legal compliance, ethical and professional standards, and appropriate oversight arrangements. In this context the National Assembly is strengthening its oversight capacity so that it can promote greater accountabilityregardingthe use o f public funds. Moreover, in order to reach RGC goals of poverty reduction and equity, spending needs to be linked more closely to priority outcomes. Increased effectiveness can be attained by improving the pro-poor targeting of resources through more tightly linked sector plans and budgets. In education and health, planning processes have improved and greater linkages between planning and budgeting have been developed, resulting in improved prioritization of spending. Agriculture and road transport have yet to make progress on this front. Weaknesses in formal and informal accountability mechanisms have a negative impact on the efficiency and effectivenessof public sector spending, and thus on poverty reduction. To the extent that citizens do not have effective means to express their preferences both to politicians and frontline service providers the efficiency of spending may be reduced. To the extent that citizens do not have effective means to monitor the outputs and outcomes produced by the public sector, the effectiveness o f spending may be reduced. Scalingup participatorymechanismsthat promote inclusionand transparency would likely have significant positive effects on service delivery outcomes, thereby increasingthe value for money o f each riel spent. Equity and patterns of public expenditure Having concluded that increasingthe level of spending is possible and desirable, it is then necessary to examine the composition of public spending, in terms of its allocation to sectors and the balance between capital and recurrent spending. 160 Sharinggrowth: egtlig and developmentin Cambodia Spending has shifted from defense to development; but more money is needed for public spending that promotes growth Table 8.1 summarizes trends in sectoral allocation, expressed in terms of spending as a percentage of GDP in that year4. These figures show that the share allocated to "core Government" have declined over the last five years. This reflectsthe reductionby a third in spending on defense and security, but also the extremely low share allocated to the judiciary. Spending on social services is now considerably higher than ten years ago (even if it has dipped slightly in relative terms inthe last two years); but allocations to the economic sectors-agriculture, transport and others-has remain virtually unchanged in percent-of-GDPterms, and about halfofthe level of social sector spending. Table 8.1 As core Government costs have been contained, developmental spending has increased\ a Government-executedexpenditure by summary function, 1996-2005 (percentage of GDP) 1996 1997 1998 1999 2000 2001 2002 2003 2004' 200512 Core Government 6.1 5.8 6.3 5.4 6.4 5.6 6 5.5 4.5 3.7 Economic Services 1.1 . 1.3 0.8- .la!!. 1,4.- - 1.8" --I*?- _!L- I2.1" --1:6-- I Agriculture 0 2 0 2 0 2 0 2 0 3 0 4 0 4 0 4 0 3 0 4 Transport 0 3 0 4 0 2 0 3 0 6 0.7 0 9 0 6 0 5 0 5 Other Economic 0 5 0 7 0 5 0.9 0 5 0.7 0 7 0 5 0 3 0 4 Environmental protection 0 0 0 0 0 0 0 0 0 0 Rural Development 0.1 0 0.1 0.1 0.1 0.2 0.4 0.5 0.4 0.3 Social Services 1.9 2 1.7 2.6 2.7 3.1 3.5 3.4 3.2 2.9 Health 0 5 0 5 0 4 0 9 0 7 0 9 1 1 0 9 1 Education 0 9 0 8 0 9 1 1 1 2 1.4 1 7 1 6 1 5 1 4 Community&SocialServices 0 5 0 6 0 5 0 5 0 8 0 8 0 8 0 8 0 7 0 5 Humanitarian I Relief 0.6 0.1 0.2 0.4 0.3 0.2 1 1.8 0.8 1.9 Total 9.8 9.2 9 9.9 10.9 11 12.9 12.7 10.1 10.1 ShareofTotal(RGCplus0DA) 043 0 4 5 039 0.47 046 048 051 052 nla n/a Giventhat ODA has compriseda significant percentageof expenditureover the last decade, it is worth looking at the picture for total expenditure (Government-executed plus externally- financed) as well as, within this, specifically Government-executedexpenditure (as presented in the table here). The former has been fairly stable, averaging around 25 percent of GDP in any given year; the latter has been increasing since 2000, and more erratic as a percentage of GDP, ranging between a low of 9 percent in 1998 to a highof 12.9percent in 2002. Institutions and eguio in the developmentprocess 161 b. Treasury-executedspending by function, 1996-2005 (percentage of total) Function 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005' Core Gover!!!!?!!! __-- _ 63.2 69.4 54.7 58.9 61 46.6 43 44.6 36.7 I I^ -62.6______I_ ~ _ l _ l l General Administration 17 157 26 19 291 257 222 246 239 1 8 4 Defense 325 327 295 252 202 1 6 5 123 116 127 114 Security 128 146 137 1 0 1 9 3 8 5 6 6 6 3 7 5 6 4 Judiciary 0 3 0 3 0 2 0 3 0 4 0 4 5 4 0 5 0 5 0 5 Economic Services _ _ _ _ _ _ 14.1 14.2 12.6 16.9 16 12.9 12.2 13 I . 10.8 9.3 Agriculture 2 1 2 6 1 7 1 9 2 3 3 9 3 3 3 4 3 2 4 1 Transport 3 1 4 1 1 8 2 6 5 1 6 4 7 2 5 5 4 4 7 Public Works 2 6 2 6 1 2 2 2 4 7 5 4 7 4 6 5 1 4 3 Other Transport 0 5 1 5 0 5 0 4 0 4 1 0 2 0 3 0 3 0 4 Other Economic Services 5 5 7 3 5 7 9 5 4 8 6 1 5 1 4 2 3 2 3 7 Environmental Protection 0 2 0 2 0 1 0 2 0 3 0 4 0 4 0 4 0 4 0 4 Rural Development 0.5 0.4 0.6 1.2 0.8 1.8 3 3.6 3.8 2.7 Social Services 19.5 21.2 18.8 26.4 24.8 28.1 27 26.8 31.7 28.7 Health 4 8 5 9 4 2 9 4 6 7 7 9 7 9 7 6 9 1 9 6 Recreation, Culture & Religion 0 6 0 8 0 6 0 8 1 8 1 9 1 4 1 4 1 2 1 3 Education 9 1 9 1 9 4 1 1 4 1 0 7 129 131 127 152 137 Social Protection 5 5 5 4 6 4 8 5 7 5 4 4 5 5 6 1 4 1 - __ 6.5 I 1.9 3.5 2.9 2.2 7.5 13.8 7.7 18.9 I I_I__ Other _I Debt 6 5 1 1 9 2 5 2 1 8 1 7 6 5 3 9 5 5 Capital Expenditure3 - 4 9 5 8 3 3 7 7 Other not classified - 1 1 0 9 0 5 0 9 1 5 0 5 5 7 Total 100 100 100 100 100 100 100 100 100 100 Source !ased on World Bank estimates and MEF TOFE, presented in Taliercio 2006 pp 12-14 Notes Only government expenditures are presented for 2004 and 2005 (i e ODA figures are not available) The reported figures are budget plans for year 2005 The way government budget expenditures are reported changed in 2001 Up until then, all capital expenditures were reported under the corresponding ministry Thereafter, some of the capital expenditures that were reported by the government were not allocated across the ministries and the remainder is presented separately The improvements in service delivery and human development outcomes (and likely positive impact upon productivity and growth) describedin chapter 6 attest to the positive results achieved through rising social sector spending. (Rising household spending on schooling and healthcare, combined with improved sector-level expenditure efficiency, have o f course also contributed.) By contrast, at $88.5 million in 2004 (constant 2005 prices), spending on economic sectors (agriculture, transport and rural development) is inadequate to meet the pressing need for spending on public goods to stimulate rural- based, labor-intensive, poverty-reducing growth of the kind needed to meet the Government's goals for growth and poverty reduction. Under-spendingon the economic sectors is an underlying weakness that needs to be addressed, both by the Government and its partners. However, to derive more detailed prescriptions for the sectoral allocation of public expenditure on the basis of presumed effects on growth and distribution (and the interaction betweenthese two) is extremely hard (see Box 8.7). Thus the broad approach of the Government at present this time-namely, to stabilize social sector spending at current levels (relative to GDP) and to focus increases on the economic sectors (agriculture and infrastructure)-makes good sense at this point. It should be noted however that increases on current spending levels in either the more advanced social sectors or the economic sectors is likely to a decline inthe overall efficiency o f spending, 162 Sharinggrowth: eqtli& and developmentin Cambodia unless some o f the increase is dedicated Box 8.7 Difficulties in identifying to increasing capacity, particularly optimal allocation for goals through civil service reform. It is clear that the ministries of agriculture, rural Analytically, using expenditure policy to meet development, and public works and outcome goals (e.g. growth and human transport are in need o f considerable development targets in the NSDP) is not so straightforward. The principal problem can be capacity development before across-the- framed in terms of the complementarity of board increases could confidently be expenditure, both in functional and economic made in their budgets (see World Bank, terms. Thus, a rural road in district X might do 2003)5. more to improve enrolment rates than building Another way to formulate the another school. The impact of infrastructure spending (roads, water, energy, etc.) on social problem of where best to allocate public outcomes is increasingly understood as highly spending so as to promote growth and complementaryto social sector spending. This poverty reduction (including through makes it difficult to provide robust high level measures that promote equity) is to policy recommendations about the allocation phrase the issue in terms o f a choice o f spendingacross major functional categories between investing in institutions (core The second major analytical complexity- functions of the state such as the complementarity of economic types of judiciary, the civil service, and the expenditure-relates to the fact that absorptive overall regulatory regime) versus capacity is endogenous. That is, greater investing in physical assets (such as spending on wages, management, and facilities roads, schools, and irrigation systems). could increase absorptive capacity, thus This has implications for the balance to justifying greater investment in sector projects be struck between capital and recurrent and programs. Writ large, the trade-off spending: building public sector between public spending and the capacity of institutions, defined either narrowly or the public sector to spend is a difficult one to broadly, is likely to imply increased assess, especially in post-conflictsettings with current tools. Despite these difficulties in prioritization of recurrent spending assessing the relative merits of different relative. The mapping is obviously not functional categories and direct as opposed to perfect-infrastructure requires capacity-building spending, this is what the recurrent spending on maintenance to Government and National Assembly in retain its value; correspondingly, practicemustdo every year. building core public sector capacities may require some one-off investment in, for example, training facilities and curriculum development-but the point remains: a serious effort to build core state capacity strongly implies increasingspending on recurrent uses, and most notably, committing significantly more funds to public sector salaries (see Box 8.8). 5 Tanzi (2004) provides an overview note on measuring efficiency in public spending. The quantitative comparative assessment of sectoral efficiency is still at an early stage of analytical development. The recent application of techniques such as Free Disposal Hull (FDH), however, is promising, and could be considered as follow on work to this report. See Herreraand Pang(2005) for a paneldata analysis of public spendingefficiencyin developing countries. Herrera and Pang (2005) also find that higher spending is correlated with lower efficiency, and inparticular with respectto higher shares ofthe wage bill to total spending. Institations and eguig in the developmentprocess I63 Box 8.8 Civil service salaries and development outcomes Arguably, Cambodia might generate higher returns from recurrent `investments' in the civil service as compared to a given roadproject. This issue-the returns to investing in institutional software rather than economic hardware-is a topic of considerable debate: the IMF notes that "In low income countries, it is not obvious that public investment should take precedenceover current spending", while the World Bank observes that "Governance is a key determinant of the efficiency and effectiveness of public spending" (see World Bank 2003 for a discussion of this issue in the Cambodian context). A better paid and managed civil service will reduce the incentive and opportunities for corruption, which a number of analysts have found directly reducesthe productivityofpublicinvestment(seeTanzi andDavoodi(2002)cited inIMF, 2006). There is not a clear consensus on this, however. Gupta et al. (2004) present a different perspective,concluding that achieving fiscal consolidation by reducingthe wage bill, as opposed ~ to capital spending, results in higher growthrates. On this analysis, then, capital investments are still to be preferred to public sector salaries. However, analysis strongly suggests that Cambodia ~ is not yet at a stage where it is desirableto reducethe wage bill. The conclusion that there is now also ample literatureon the linkbetweengovernanceandgrowth similarly provides no clear practicalguidance. The importanceo fgovernance to the efficiency of public spendingwould suggest, on the one hand, caution in urgingthe creation of fiscal space in countries where governance is weak: but, on the other, it also implies the need to improve governance in order to enhance the overall impact of public spending, most especially in countries where fiscal space will be rapidly expanding(e.g., as a result of the discovery of natural resources, as in Cambodia,which have been found to have a strong, negative effect on growth by underminingthe quality of institutions:Sala-i-Martinand Subramanian, 2003). , Sources:IMF2004 p. 23; WorldBank2006. Given the serious problems afflicting the civil service-low pay, low skills, and thus low capacity-strategic civil service reform will have to be accelerated in the short term and carried out over the medium term if the RGC's vision of poverty reduction is to become reality (World Bank, 2003). Cambodia's reform program must look at every possible avenue for improving civil service pay, to attract and retain skilled staff, especially for high level management and priority sector staff; rationalizing civil service employment, to ensure that human resources are deployed well in high priority sectors; and strengthening civil service management, to guarantee that human resource expenditures are subject to controls and linked more closely to the budget formulation process. Indeed, one o f the principal risks to the NSDP is the capacity of the civil service to deliver. Though the details of Cambodia's pay reform have yet to be developed, it is clear that increasing civil service pay will require a large amount of additional resources. This `investment,' which is recurrent in nature, is undoubtedlyone of the highest priority uses for Cambodia's future petrochemical revenues (assuming efforts are also made to improve the management o f the civil service). Yet there is no analytical tool that allows for a cost-benefit assessment of the impact o f building the institutions of state through higher pay versus investing in roads and school buildings. Nonetheless, a civil service pay reform policy clearly needs to be prepared to lay claim to, at the very least, the minimum amount of resources neededto establisha functional civil service in the priority sectors, subject to the recommended wage bill in the medium term macroeconomic 164 Sharinggrowth: eqtlzg and developmentin Cambodia framework. A sound strategy for boosting wage spending needs to be a major feature of Cambodia's medium term expenditure policy. Building vs. sustaining improvements The other way to examine the composition of public spending and its likely contribution to promoting an equitable growth path is to examine the balance between investmentor capital spending (that is, spendingon new buildings, equipment, and so on) and recurrent spending (regular expenditures for public sector salaries, maintenance costs, and so on). There is a pressing need to increase spending on the maintenance o f public sector infrastructure investments. As a post-conflict country, Cambodia has an extremely long list of urgent investments needed to rebuild the country's physical and human capital. At the same time, it is highly inefficient to use limited resources nearly exclusively on new investments (e.g. a new or rehabilitated road), which then rapidly degrade for lack o f recurrent maintenance spending, necessitating a krther round of extremely expensive investmentspending. Overall spending has risen over three-fold in real terms Figure 8.2 The economic classification of public spending, 1996-2005 between 1996 and 2005 (Figure 8.2, Table 8.2): this rate of 600 1 increase has been roughly similar for recurrent and capital spending, although capital spending has been more erratic. Averaging 2 percent of GDP over decade 1995-2004, Cambodia's capital spending has been less than halfthe low- income country (LIC) average (4.9 percent), placing it eighth from bottom out of 30 LICs for , I , , , I > I I , which data is available. This 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 helps explain Cambodia's extremely low rates of Source: MEF TOFE reports electrification; poor penetration o f telecommunications;and road and irrigation networkswhich are amongst the lowest in the region (World Bank 2004, 2006). Chronic and severe public-sector underinvestment in infrastructure are also likely to have serious effects on inequalityas well as aggregate growth (Box 8.9). Throughoutthis report, numerous statistics point to the existence o f very pronounced geographical inequalities. Although other factors (e.g. soil quality, length of time elapsed since armed conflict ended in the locality) are also at work in explaining differences in spatial averages, a key explanatory factor is the difference between core and well-connected parts of the country on the one hand and remote, isolated parts o f the countryside on the other. In lowland areas close to towns and national roads, inhabitants have access to more varied and extensive markets, and enjoy proximity to schools and healthcare Institutions and equip in the developmentprocess 1G5 facilities; in poorly-connected, typically upland areas, opportunities and service Box 8.9 infrastructure and equality: cross-country and locality coverage are much more limited. findings Improving total spending on basic infrastructure and, where appropriate, Calderon and Serven (2004) argue that targeting investments to poor areas infrastructure spending can reduce income inequalityby having a disproportionate impact could play a powerful role in equalizing on the poor by connecting them to economic opportunitiesand, eventually, outcomes. activities (e.g., roadsthat link poor agricultural If it is accepted that public spending areas to markets increasethe value of the poor on infrastructure can generate high farmer's assets) and aiding in human capital returns and positive externalities (World development (e.g., roads have been shown to Bank 1996, 2005; Aschauer 1989), it is correlatepositivelywith school attendance, and clean water can have an enormous impact on important that Cambodia develop an child mortality). Calderon and Serven (2004) effective investment appraisal system to found that both infrastructure stocks and the maximize the rate of return on public quality of their services have a negative and investments. This should also serve to significantimpact on the Gini coefficientusing identify the budget allocations that need panel data for over 100 countries spanning the to be set aside for maintenance of new years 1960-2000. investments. At present, maintenance At the other end o f the scale spectrum, the expenditures are extremely low, MOPS research "suggests that investment in covering probably only two-thirds of making communities as accessible as possible what is needed to sustain the value of i s critical and should continue, preferably as a existing investments. national and coordinated initiative rather than on the current, often ad hoc basis, which sees Finally, while the public sector wage roads constructed, often by NGOs or donors, bill has increased from $39m to $loom, but then not maintained; or where road access this is below the rate of overall or is controlled by private interests, as is the case recurrent expansion, with the result that inBabaong. Accessibilitywill inturn facilitate the salaries and allowances fell from access to and demand for new markets, about one third to one quarter as a share services, and opportunities including for of total recurrent spending. The employment and trade" (CDRI 2007b, inadequacy of pay for public servants, forthcoming). and the perverse, anti-poor outcomes that it gives rise to, are readily apparent to most Cambodians. The somewhat unhelpful conclusionfrom analysis of public expenditurepolicy is that levels of spending on all sectors and categories-including the construction and subsequent maintenance of infrastructure, civil service salaries, and running costs-are far too low to meet the level o f need, achieve efficiency in implementation, or effectively translate policy goals and priorities into economic and social outcomes. There is however two good reasons to be optimistic about the prospects that public expenditure could, in the medium term, be an effective instrument for the Government's goals of equitable development. The first is the progress in the PFM Reform Program. The second is the recent discovery of significant reserves of oil and gas in Cambodian territorial waters, which greatly expands the fiscal space available to the Government. It does however come with its own set of complicationsand risks. 166 Sharinggrowth: equig and developmentin Cambodia Table 8.2 Spending has risen three-fold since 1996 - Treasury-executed civil administration spending by economic category, $m constant 2005 prices (indexed, 1996= 100, in italics) Budgetchapter 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 Sub-Total 123.3 122.3 134.4 172.8 217.2 274.2 318.5 367.2 341.8 369.0 Current 100 99 109 140 176 222 258 298 277 299 10: Salaries and 39.1 41.2 42.8 52.6 60.6 64.0 83.0 89.5 88.8 100.3 Allowances 100 105 I09 135 155 164 212 229 227 257 11: Operating 50.5 50.7 52.9 74.1 101.6 130.7 120.1 132.2 117.1 110.6 costs 100 I00 105 147 201 259 238 262 232 219 12: Subsidies - 5.4 5.2 4.5 4.1 5.7 6.9 15.4 29.2 29.5 31.2 provincial admin. 100 96 83 76 706 128 285 541 546 578 13: Special prog. 0.1 0.3 8.1 1.4 3.2 20.5 37.7 38.0 39.4 43.2 agreement I00 300 8,100 1,400 3,200 20,500 37,700 38,000 39,400 43,200 Other current 28.1 24.9 26.1 40.6 46.1 52.0 62.1 78.3 67.0 83.7 (20, 30-2, 40-1) 100 89 93 144 164 185 221 279 238 298 Sub-Total 34.5 29.3 34.0 63.8 89.0 80.1 164.7 172.1 113.0 144.3 Capital 100 85 99 185 258 232 477 499 328 418 50.1 constrction 16.7 22.5 23.3 53.3 63.2 53.2 126.9 93.1 54.2 84.5 & Equipment 100 135 140 319 378 319 760 557 325 506 50.2 Counterpart 3,1 5.0 9.2 7.5 23.0 14.8 27.9 25.0 27.9 24.4 funds 100 161 297 242 742 477 900 806 900 787 Other capital 14.7 1.8 1.5 2.9 2.8 12.2 9.8 54.0 30.9 35.4 (50.3, 52, 53) 100 12 I O 20 19 83 67 367 210 241 Total 157.8 151.6 168.4 236.6 306.1 354.3 483.2 539.3 454.8 513.3 100 96 107 150 194 225 306 342 288 325 Oil wealth and equity The major opportunity on Cambodia's fiscal front is the Box 8.10 Current estimates of Cambodian discovery of significant amounts of petroleum reserves offshore petrochemical reserves The RGC has demarcated six offshore blocks for (Box 8.10). The magnitude of these licensing. In addition, there may be deposits in the oil and gas assets has the potential to Overlapping Claims Area (OCA), under transform Cambodia's fiscal negotiation with Thailand, and in several inland situation. Indeed, the magnitude of blocks. To date, contractshavelikely beenawarded the discovery is such that it raises for five off-shore blocks (A-E), and E is under questions about to what extent negotiation. According to the Cambodia National Cambodia should continue to pursue Petroleum Agency (CNPA), the block A contract growth of non-oil related revenue area contains an estimated 700 million barrels of oil and 3-5 trillion cubic feet of natural gas. sources. Maintaining the current Estimates for other blocks are not yet available, revenue to GDP share but and it will take another 1-2 years before there is augmenting it considerably with oil- good data on Blocks B-E. In Block A CNPA related revenuecould be sufficient to estimatesthat oil production will start in 2009 and generate the resources needed for the gas productionin2011. RGC's medium to long term Institutions and equip in the developmentprocess 167 program. Estimates show that oil revenue will exceed all current revenue sources: 500 million barrels of recoverable oil would yield an annual peak revenue stream to the government of more than $700 million (the revenue stream peaks four years after the start of production), equal to about 11 percent of GDP in 2005.6Estimatesat this point, however, are highly uncertain due to imprecise estimates of Block A's volume (Chevron is currently conducting additional exploratory tests). The price of oil is also highly volatile over time and, while high now, might decline over the next decade to a somewhat lower level. At projected future oil prices closer to today's price (about US$ 60 per barrel), annual revenue from Block A alone could easily exceed US $ 1 billion on an annual basis (which would represent about 15 percent o f 2006 GDP). If other blocks also turn out to be commercially viable, government revenue would increase further, but it is too early at this stage to speculate on precise amounts. Thus, at this point a forecast o f around 10 percent of GDP per annum from Block A seems reasonable, and would, in effect, double Cambodia's revenue-to-GDP ratio. If one other block (of the other five) contained as much as Block A, revenue to current GDP couldtriple. Even if actual revenue is less than current estimates, oil and gas will completely transform Cambodia's fiscal space. It is possible that the revenue to GDP ratio could double over a 3-4 year period, which would put great pressure on the public expenditure management system to deliver resources more efficiently. Managed well, this has the potential to improve state spending on economic and social development and accelerate progress towards Government development goals. If managed in a strategic and accountable manner, these funds could help Cambodia escape least-developed country status in a much shorter time frame than has previously been thought possible. However, global experience suggests that petrochemical wealth may equally well result in a "resource curse": in more cases than not, the discovery of oil has greatly exacerbated inequalitieswithin countries and actually reduced rates of growth and poverty reduction. This pattern is usually explained in terms o f the effect of mineral wealth in undermining a state's accountability to its citizens: a government that obtains all the money it needs from mineral rents has no need to collect (or account for) tax (and so less reason to see itself as accountable to its citizens), and lacks incentivesto invest in reforming state institutions or fostering broad-based growth. These problems are more pronounced in countries with low state capacity and relatively weak formal institutions, and are at their worst in post-conflict countries in which the levels of transparency and accountability are very low and state-society relations have been shaped by a tradition of authoritarianism and extraction. Nigeria, where an average of $12 billion per annum in oil wealth over a thirty year period has fostered pervasive corruption and done nothing to promote growth or reduce poverty is the most obvious example but far from the only one. The Nigerian experience contrasts markedly with many east and southeast Asian economies (e.g. Japan, Korea, Taiwan or Thailand) that are not richly endowed with mineral wealth, but which have sustained broad-based, poverty-reducinggrowth over several decades. 6 This estimate is based on a constant projectedworld oil price of $43.75, 168 Sharinggrowth: equig and developmentin Cambodia An additional concern is that oil revenue might trigger "Dutch disease" effects. Foreign exchange generated from the oil sector could result in an appreciationof the real $/riel exchange rate, which would weaken the competitiveness of traditional exports, notably the garment sector. This sector has served as the main wage employer of unskilled (and mainly female) labor and as such has been a key factor in enabling growth with static consumption inequality since the late 1990s. It is thus important for sustained poverty reduction and equity in Cambodia that the potential-macroeconomic effects associated with oil wealth are anticipatedand avoided. There is thus good reason to be cautious about predicting the impact of any future stream of petrochemical revenue on the Cambodian economy and the prospects for Cambodia's poor. Carefully designed institutional mechanisms that promote transparency and accountability in the use of this national asset will be critical if oil wealth is to help rather than hinder Cambodia's escape from the ranks of least-developed nations (Vallely et a1 2004; UNDP 2006). The potential for a dramatic increase in revenues adds further urgency to the need for the strengthening of public financial management-and sharpens the case for improving transparency and accountability in order to combat corruption in public life. Recommendations To improve the capability o f the Government to use public expenditure policy to promotegrowth and equity objectives, a numberof clear priorities emerge: Increase jscal space by increasing revenue. Cambodia's fiscal space is quite constrainedby its low revenue effort, the impact of which is notedacross all categories of spending. Given the discovery o f very large off shore reserves, Cambodia's fiscal picture i s likely to change very abruptly. Oil revenue will double or triple the revenue to GDP ratio, thus completely transforming Cambodia's fiscal space and opening up the possibility of increasingspending across several types of spending simultaneously. Increase the budget allocation for economic sectors (agriculture and infrastructure); within the general government sector, reallocate from general administration to the judiciary. Cambodia's economic sectors are still starved o f the resources needed to make the difference between lower and higher growth. Though the RGC has increased spending in both agriculture and infrastructure, the overall increases have been meager. Similarly, sustained increases are needed to transform the budget-starvedjudiciary. In both cases, however, there is a very important caveat: spending increases should proceed only after sectoral strategies have been developed and management has improved. While economic sector management is being developed, spending increases could focus on capital investments (see below). In economic terms, the RGC should increase capital spending, maintenance, and the wage bi11. Increased capital spending is needed to take advantage of the high return projects (principally in infrastructure and agriculture). It should however follow the development of project appraisal capacity (an investment appraisal unit based on suitable methodologies),which is currently sorely lacking. Institutions and equig in the dei elopmentprocess I13 Specifically in terms of ODA, donors should reallocate from donor-driven free- standing technical cooperation (FTC) to capital spending where sector wide approaches (SWAps) do not yet exist; or use FTC to set up SWAps, which seem to be the best mechanism in Cambodia for moving toward sustainable, capacity-enhancing, Government-ledreform programs. FTC in the absence o f SWAps seems to be generally quite wasteful and in most cases these resources could be more productive on the capital side. The RGC also clearly needsto increase public sector wages as the cornerstone of civil service reform. This is likely to require a significant increase in the civil administration wage bill. The RGC is preparing a new pay policy for 2007, which will providethe basis for an increase in the wage allocation. This should be regarded as a priority claim on scarce resources. Building institutions in Cambodia is as much of a priority as building physical assets. Increase maintenance spending. Maintenancespending i s inadequate. Physical assets are deteriorating in many cases, undermining the high potential of capital spending to deliver sustained growth. Maintenance spending should be increased dramatically to about 1.5 percent of GDP. Reduce wasteful spending. The budget is characterized by pockets o f inefficient spending, in some cases due to high administrative overheads, in others due to the perverse effects of low wages on other spending categories (e.g., the operations and maintenance budget in education being used to pay teachers' special allowances). In still others it is reflects the lack of a coherent sector strategy. The RGC should conduct an efficiency review to locatethese pocketsof waste. Legal andjudicial institutions and access tojustice for the poor The system by which justice is Box 8.11 Justice systems play a mediated is an important underpinning crucial role in long-term of economic and social development development (see Box 8.1 1). Capable, honest and autonomous legal and judicial The 2006 World Development Report summarizes international experiences on the institutionshave the potential to limit the importance of legal and judicial systems as emergence o f gross inequalities in power follows: between the state and the citizenry, or between the rich and the poor. Legal institutions play a key role in the Independent courts enforcing the distribution of power and rights. They also principle of equality before the law underpin the forms and functions of other institutions that deliver public services and should serve as a limit upon the exercise regulate market practices. Justice systems o f unrestrained power by the executive can provide a vehicle to mediate conflict, or abuse of position by officials. Equal resolve disputes, and sustain social order. treatment by the courts in the event of a But inequitable justice systems may dispute over ownership or contracts is a perpetuate poverty traps by maintaining or key component of equality of economic reproducing elite interests and discriminatory opportunity, and an efficient justice practices. Equitable justice systems are thus crucial to sustained equitable development. World Bank 2005 p. 156. 170 Sharinggrowth: equip and developmentin Cambodia system bolsterscommitmentsto equity inthe delivery of public services. Unfortunately the legal system in Cambodia has yet to fulfill this potential. Severely lacking in capacity and resources and badly affected by lengthy delays, a shortage of legal aid and difficulty in enforcing judgments, the court system (and other elements of the de facto administrationofjustice, such as the police) are widely perceivedto be either deeply corruptedor underthe direct control of the executive. A public opinion poll (CAS 2003) found that Cambodians perceived only slight improvement in the performance of courts from 1998 to 2003; and had little or no faith in the courts as institutions ofjustice, identifyingjudges and prosecutorsas the public officials they were least likely to trust. A more recent survey of popular experiences of corruption in Cambodia suggests the picture has not changed significantly: the courts were seen as by far the most corrupt of all public institutions, followed by the police (CSD 2005). As a result, many Cambodians (and especially poor Cambodians) seek to resolve disputes and obtainjustice by working around rather than through the formal institutions of the state (CAS and World Bank 2006; UNDP 2005). In the absence of equitable rule-based systems for allocating resources and resolving disputes, collective action is a crucial mechanismfor improving the responsiveness of the state to the needs of the poor. With pressure on land and natural resources increasing, recentresearch suggests that there is increasingwillingness for villagers to pursue claims, even against people who are rich or powerful (CAS and World Bank 2006). This research finds that the state is not imperviousto these claims. Without suggestingthat the disputes studied were resolved satisfactorily, it was clear that when villagers had acted collectively they had been able to achieve more equitable outcomes than might otherwise have been the case. This finding echoes that of Nee and Healy (2003) who suggest that the existence of social capital is a major factor in the achievement of pro-poor results from local dispute resolution systems. While acknowledging that it is ultimately the responsibility of the state to ensure the rule of law they note that "local associations working together are already seeing successes" (p.101) and "it is often possible for village associationsto obtainjustice simply by building a strong network of relationships with both NGOs and government officials at all levels" (p.102). There is, however, an operational dilemma in these sorts o f dispute resolution `successes'. In that these cases show collective disputes being dealt with by powerful individuals within the administration, they reinforcethe picture of a formal justice system that is of only marginal importance. To the extent that parties were successful in extracting increased responsiveness from the state, they did so largely by putting political pressure on key decision makers and not by relying on the law or seeking access to officially mandated dispute resolution process. Written laws, legal processes or rule- based forms of decision makingwere notable by their absence and institutions established outside of the patronage system of power were seen as ineffectual. As such, these studies provide few direct insights as to how the rule of law might be established in the Cambodian context. The findings do, however, place Cambodia in a position that is historically common, given that the model of a modern judiciary emerged in an iterative process, with independence from the executive secured over time in response to a myriad of social, political and economic pressures. This observation supports the case for an interim Institutions and equi9 in the developmentprocess 171 institutional approach when thinking about the development of more equitable institutionsofjustice in Cambodia. Effective equitable institutions require open (and open-ended) processes Research on evolving patterns of disputes and dispute resolution in Cambodia suggest there is value in seeking "interim" institutional arrangements, rather than necessarily looking to transplant "international best practice" organizational forms into Cambodia (see Adler et a1 2007 and Box 8.12 below). This can be illustrated with referenceto two cases of policy and institutional reform. Box 8.12 "Interim institutions" for governance reform The interim institutional approach provides a method for engaging with situations of deeply entrenched legal pluralism. Acknowledging that such situations limit the effectiveness o f state law as a driver of more equitable governance, an interim institutionalapproach argues that: (i) equitable, rule-based systems for allocating resources and resolving disputes-Le., the very content and legitimacy o f modern institutions of government-emergeand continue to develop because they are subject to ongoing social contest; and (ii) because equitable institutional arrangements for social regulation must emerge through social contest, there will be limits to the extent to which they can be designed ex ante on the basis of technical knowledge (Le., by governance `experts'). Rather than from posing de-politicized solutions to complex political and social contests, an interim institutional approach focuses from the outset on the process whereby more equitable relationships of power might be brought about. Indoing this, it asks what spaces exist (or can be created) for the negotiation of developmentconflicts, and considershow such spaces can be filled with institutions that bothrespondto the realitiesof power as it is currently exercisedand provide the potential to transform these in the direction of greater equity. In other words, rather than emphasizing coherence, consistency or harmony, it understands that social contest is an essential partof institutionbuilding. The interiminstitutionalapproachis relativelyagnostic with respectto claiming to know (or even worryingabout) what the final end-stateof any given institutionbuildingor reformprocessmight "look like"; rather, it focuses on the more immediate and pragmatic tasks of (i) creatingpolitical space for reform; and (ii) taking feasible steps on the basis of processesthat are deemed equitable and legitimateinthe eyes o fthe various stakeholdersconcerned.Supportingand facilitatingsuch political processes should be the primary concern of external development professionals concernedwith institutionbuildingand reform in environmentscharacterizedby pervasive legal pluralism. "Getting the contests right" could be the simplest statement of the programmaticgoal andanalytical distinctiveness ofthe interiminstitutionalapproach. Source:Adler et a1(2007) Thegarment industry and core labor standards Cambodia's 1997 labor law incorporatedseveral important rights; but it was unclear how the law, once issued, would be translated into practice. Classically, law on the books is enforced through administrative and judicial mechanisms which monitor compliance and/or respond to grievances: actual enforcement is required only infrequently, as anticipation of enforcement gives the law a more general effect. In the context of the 172 Sharinggrowth: eguzg and developmentin Cambodia Cambodian workplace, however, the credibility of the law was underminedby the fact that the bodies expected to monitor compliance or respond to grievances (namely, the courts and labor inspectorate) were widely seen as lacking capacity and subject to bias and corruption. Without expectation of systematic enforcement, legislation becamejust one among several norms shaping working conditions at the enterprise level: as such, the law was in competition with market forces and interaction between the managerial culture of (often China-based) employers, corporate codes of conduct imposed by global buyers, and the culture of spontaneous (and sometimes violent) collective action which emerged in the dormitorieshousingmigrant workers. The basis for more equitable labor relationsbeganto evolve in the late 1990swith the rapid emergence of Cambodia as a garment exporter. A 1999 US-Cambodia bilateral trade agreement imposed quotas on Cambodia's garment exports; provided that Cambodia would support implementation of "a program to improve working conditions in the textile and apparel sector, including internationally recognized core labor standards"; and, in an interesting innovation, created incentives for improvement by committing to increase quotas as long as working conditions complied with such labor law and standards". The agreement thus created immediate incentives for all stakeholders to implement the labor law: employers would gain business, unions could gain members and the Government could generate both formal and informal revenues from allocation o f quotas. The question remained, however, how the labor standards clause would be implemented, given the lack of credible country systems. The answer was provided by two ILO projects: one to "operate an independent system to monitor working conditions in garment factories" (the monitoring project) and the other to focus on "transparent, fair and expeditious dispute procedures" (the labor dispute resolution project). Box 8.13 summarizes the key issues and features of both projects. The success of monitoring and dispute resolution efforts in the garment industry seems to have rested on the willingness and ability of key players to engage with the inherently contested process by which rights are attained. Rather than investingprimarily in more detailed regulation, administrative or judicial capacity, a number of alternative initiatives were undertaken. In both initiatives, there was an implicit recognition that there may be advantages in supporting the emergence of rights through ongoing processes of contestation and negotiation-where issues of power are at once more overt and more fluid-before trying to fix them in formal legal and enforcement frameworks. In these circumstances, the Labor Law's main (equity-enhancing) effects have been: (i) establishing a set o f (more or less) agreed standards regarding working conditions, which even though not formally enforced, can, in certain circumstances, be drawn upon to legitimate an argument as to what working conditions should be; and (ii)providing a framework for collective action, by legalizing unions, providing for tripartite dispute resolution procedures and legitimizing the right to strike. Institutions and equip in the developmentprocess 173 I Box 8.13 Monitoringand dispute resolution in the garment industry The ILO-supportedmonitoring project (now Better Factories Cambodia) produces regular reports on conditions in garment factories and is generally agreed to have improvedcompliance with the labor law in the garment industry. Several factors appear to have contributed. Firstly, the project promoted labor rights through embeddedsocial dialogue, the provision of public information and trade preferences. Secondly, export licenses were granted conditional on an enterprise subjecting itself to ILO monitoring, andthe Ministries ofCommerceandLabor can deprive employersof their , export licenses ifthey are found to be in continued breach of core labor standards.And, finally, by publishing information on individual factories the system generates pressure for compliance both I from others in the industry and their brand name buyers in the US. The willingness of both employers and the Government to continue monitoring after the expiry of the agreements and the quota system in 2004 suggests that the financial incentives included in the trade agreement, while instrumental in establishing the system, were only part of the story in sustaining it. Whether i monitoringcan generate the incentives neededto drive longterm improvementswithout the carrot I and stick ofaccessto US markets is yet to be seen. Inthe labor dispute resolution project, ILOprovidedtechnical assistanceto the Ministry ofLabor in ~ consultation with the union movement and employers' associations. When it was established in I 2002, each of the three options available for dealing with labor disputes (negotiated outcomes, referral to the Ministry of Labor for conciliationor enforcement proceedings, or the courts) had ~ significant drawbacks. Systems to handle grievances at the factory level were under-developed, ~ reflected in increasinglevels of often violent industrialaction. The labor inspectorate suffered from I problems common to the Cambodian public sector: limited capacity and low wages led officials to rely upon informal payments from industry to support their livelihoods, undermining their credibility. The courts sufferedfrom similar problems. There was a range of possibleresponsesto these circumstances. The project could have focusedon the establishment of a labor court. However, the ILO concluded that it would be difficult to get classical institutions of the rule of law (those which are both independent and have the power to make binding decisions) to work in Cambodia. Observing previous attempts at legal andjudicial reform, it was anticipated that pursuingthe rule of law directly would result in either a) stallingthe ~ 1 I process of setting up a new labor court or b) the capture of the new institution by powerful Governmentandprivate sector interests. As a result, the project focused on creating an Arbitration Council, a tribunal body providedfor in the 1997 law but not previously operationalized. As a tripartite body that generally issues non- binding and practically unenforceable awards, the Council is a hybrid between a rule of law institutionanda forum for bargainingbetweenmanagementandorganized labor. It would appear to have been successful in focusing and legitimatingcollectiveaction for more harmonious, equitable and constructive industrial relations. The Arbitration Council had received 272 cases as of June 2006, of which68 percentwere reportedas resolvedsuccessfully. The Arbitration Council is not without critics. Unions complain that enterprisesdo not give effect to arbitral awards and that it would be better if the Council could issue binding awards. While , generally supportive, employers have on occasions criticized the Council for displaying a pro- ' worker bias. Institutionally there are questions about whether the Arbitration Council is a viable institutionofstate absent ongoing donor support. Source:Adler et a1 2007. Thinking of law in this way allowed the development o f a dynamic institutional setting for political contests, which began to foster equitable rule-based systems of 174 Sharinggrowth: equip and developmentin Cambodia governance. These included: (i) support for the capacity of workers to engage in collective action (through unions); (ii)independent monitoring of working conditions within both a local tripartite structure and a framework of incentives and sanctions negotiated at the international level; and (iii)the establishment of credible (though non- binding) dispute resolution mechanisms through which rights could be contested accordingto agreed `rules of the game.' Ifthe case of labor relations illustrateshow a more equitable regulatory context can emerge, it also shows some clear limitations of this approach. Firstly, the law is applied almost exclusively in enterprises (primarily garment factories and hotels) which are accountable to an internationalpublic through their branding: the extent of any spillover from these sectors into the labor market more generally is unclear. Secondly, these systems operate primarily at the level of the collectivity: in practice, the current regulatory system neither protects workers' rights as individuals, nor does it extend to vulnerable workers, such as those in small and medium sized enterprises producing for local markets. Nevertheless, an evolutionary approach that valued interim institutions would appear to provide a viable first step on a path to hopefully more equitable, formal mechanismsto mediate relationsbetweenlabor, employees and the state. From project law to statute in local governance In the case o f labor relations describedabove, and land titling describedin Chapter 2, innovative interim institutional arrangements have been introduced to facilitate the implementation of legislation that was already enacted but remained unconnected to existing, alternative frameworks for managingeconomic relationships. The case of local governance development provides an interesting contrast. In this case, institutional and policy reform proceeded in advance of legislative change; but with the intention of trialing and demonstrating arrangement that might later be incorporatedin legal, political and administrative developments. In the mid-late 1990s, under the auspices o f the Seila program, Cambodiaintroduced two sets o f institutions of local governance, namely the Commune Development Committees (CDC) and District IntegrationWorkshops (DIW)'. Seila emerged from the CARERE (Cambodia Area Rehabilitation and Regeneration) program which between 1992 and 1996 promoted integrated rural development, with a focus on delivering small scale infrastructure in five north-easternprovinces to aid the return and reintegrationof refugees. CARERE was controlled through parallel systems, according to UN and other donor `project law.' By the time it officially closed in 2006, Seila had become a nationwide program mobilizing funds from both the Government and donors with both governance and direct poverty alleviation goals. From the start, the program's execution and management arrangements were hybrid, involving aspects o f both donor and government regulation. To an unprecedented extent, however, Seila generated an increasingly comprehensive framework of Government laws and regulations. This transformation occurred in part because of a policy decision (expressed in a new set of objectives in 1996) to change the program's purpose and 7 Details on the Seila program can be found in Wong and Guggenheim(2005); and on http:llwww.seila.gov.kh. Institations and equio in the developmentprocess I75 strategy, which gradually moved from a focus on `local development for poverty reduction'to one primarily concerned with `local governance'("strengthening institutions at the commune and provincial levels" and financing "participatory development."), through which sustainable improvements in local development would support poverty reduction. Most importantly, Seila aimed to pilot models of "decentralization and deconcentration". The program thus came to focus expressly on creating broader `demonstrationeffects' for institutional reforin. The CDC and DIW were explicitly concerned with creating new arenas for political interaction and regulated contest both between citizens and the state, and between government institutions at different levelswithin the state (see Box 8.14). Eventually, the norms promoted by these interim institutional arrangements came to feature as key aspects of national policies and laws about local governance. The CDCs, and the systems for planning, budgeting and financing that were integral to their operation, found a place in national law when the commune councils were legislatively reconstitutedinthe LAMC 2001. The CDCs were disbanded immediately following the elections, with their participatory planningfunctions placedunder the Planningand BudgetingCommitteeand their managerialand governance functions falling under the Commune Council. Box 8.14 Commune Development Committees and District Integration Workshops Commune Development Committees(CDCs) were mandatedby sub-decreein 1994 as part of an administrativestructure for rural development repeated from village to province levels. CDCs were seen by CAREREiSeila as providing an opportunity to experiment with decentralized planning, budgeting and management in service provision. This had both technical and political aspects. CDCs were given discretionary powers to allocate donor finds according to needs expressed and negotiatedthrough village and commune planning processes. From a technical viewpoint, it was expected this would result in a more `efficient' conversion of available local and international resources into useful infrastructure and other investments. The larger ambition, however, was political.The planning and budgetingprocesses being piloted through the CDCs were pilots for a new regulatory framework for local governance, one which would structuremore equitable contests amongcitizens andtheir local representativesabout how public spendingshouldbeprioritized.This process was to build `primary accountability'; it was intended to promote the notion that elected leaders-and through them, the administration-should be accountable to citizens for how they madeand implementedplanning, budgetingandrelatedmanagementdecisions. The District Integration Workshop (DIW) likewise hadtechnical andpoliticalambitions.The DIW was introduced as an annual event in the planning and budgeting calendar. Its goal was to secure from provincial agencies the technical and other support the CDCs required to implement their investment plans. The larger political ambition related to `intra-governmental accountability'. In brief, this included the idea that higher level administrative agencies should be accountable to priorities expressed by citizens throughtheir local politicalrepresentatives.It also includedthe idea that local choicesneededto be negotiatedwithinthe policyframework ofthe government expressed at higher levels. The DIW process-part o f sub-decrees and regulations for commune planning-has not been `institutionally embedded' in the same way as CDC institutions. In fact, it has not yet substantially affected what donors or the administration do beyond the Seila program, despite the fact that roughly 60 percent of all agreements made at the DIW are 176 Sharinggrowth: egg@ and development in Cambodia financed by projects outside the Seila program. However, the basic precepts of the DIW-that higher level authorities should respond to local priorities, and that these local priorities need to be negotiated with national policy priorities-have become influential. Thus, it might be argued that the DIW has contributed to norms that now feature in an extensive range of project laws, political debates and policy commitments about how Government should behave in Cambodia. In order to understand how local governance innovations such as the CDCs and DIWs came to be conceived and implemented in Cambodia, it is important to understandthe sources of intellectual and political support on which they were able to draw, both internationally and domestically. Internationally, the CDCs and DIWs (and Seila more generally) were both products of and contributors to contemporary international development policy debates about decentralizationand `community driven development'. In general terms, the case for decentralization was that individuals, households and communities were best placed to determine their needs, and that government functions should be devolved so as to be closely accountable to them. More specifically, there were two key claims put forward regarding the virtues of decentralizationand community-baseddevelopment(Mansuri and Rao 2004): Firstly, it was argued that decentralized and community-baseddelivery enhanced the eficiency of public spending by reducingwaste and corruption. Such approaches were thought to increase the share of public funds (Government and ODA) that was converted into useful services (`productive efficiency'); and to ensure a better match betweenwhat peoplewanted and what Governmentdelivered (`allocative efficiency'). 0 Secondly, and somewhat more contentiously, decentralization was argued to increase inclusionand equity, through more even-handed distribution of public funds across the country (`geographic equity') and/or arrangements to ensure that disadvantaged groups in any area get a fairer share ofthe benefits(`social equity'). Within Cambodia, by 1997 Seila had a reputation with Governmentand donors for an ability to (i)swiftly convert aid money into a very wide range of local investments and (ii) to achieve even-handed working arrangements with the sub-national leadershipo f the various political factions of the northwesternprovinces.These capabilities made rapid aid delivery possible; which in turn meant that Seila had made some notable contributionsto peace and reconciliation. For the Government, Seila had demonstrated that it knew how to behave, technically and politically: it was regarded as a dependable, non-threatening contributor to security. Seila's reputation as a non-political `safe pair of hands' was matched by its close alliance with Provincial governors and, through a wide experience o f sub-contracting arrangements, with the core administrative departments. At the same time, donors were comfortable that this did not put at risk the management systems Seila maintained to safeguard and protect their funds. The program had by the mid-1990s created Province- level systems for coordination and management that were `hybrid'. In other words, the formal and informal systems and procedures which enabled the program to operate were severally managed by official government authorities, headed by the Governor (through Institutions and eqtlip in the developmentprocess 177 what later became PRDC/Ex Comm*); by local networks of contractors and suppliers often affiliated with administrative officials; and by local staff contracted to administer UNsystems.The programwas thus ofthe Governmentbut still separate from it. Through its processes and outcomes, Seila was able to provide legitimacy to the political regime. The governing political party, especially after the 1997 crisis, was able to direct benefits down to the politically and electorally important Commune level and thereby bolster their credibility `from below'; at the same time Seila could also show to the international community that the Governmentwas serious about local democracy, and thus bolster its credibility `from above'. In summary, CDCs and DIWs helped strengthen political equity, but constraints remain. In assessing the successes and limitations of Seila-induced institutional innovations, it would be nai've to assume these were premised simply in terms of their aim to enhance equity in sub-national governance. Further, in a system characterized by legal pluralism it is clear that to be durable, the new institutional arrangements for local governance need to be underpinned by formal law, administrative conventions, informally sanctioned expectationsabout what is `right' and `fair', and social norms that support and provide credibility for the kinds of state-citizen relationships and political contests implied by these innovations. These various forums relate in complicatedways: sometimes they reinforceeach other, as often they are mutually contradictory. With these caveats in mind, it is possible to draw a number of conclusions regarding the durabilitylembeddingofthese interim institutional arrangements. The institutional arrangements for equitable participation in Commune-level governance-and the reasonable expectation that such governance will be more equitable and pro-poor as a consequence-have survived and strengthened with the transition from project law to Government statute. In 2001 two laws were passed, one for commune- level elections and a second governing commune administration and management (LAMC). This cleared the way for a nationwide election, in 2002, of 1621 Commune/Sangkat Councils (C/SCS)~.Proponents of the CDC as a means to more accountable (and thus equitable) local governance felt these laws vindicated Seila's approach. It s possible to identify at least six aspects of legal conventions, conducive to local political equity, that came into being as a result of the 2001 laws: 0 Territorial authority. The law empowered CCs to govern the `local affairs of its territory', establishing the legal convention that CCs are the primary instrument of representative democracy, in and through which elected leaders will be made accountable to all citizens for what occurs in theirjurisdictions. 0 Equality.of political rights and voice. CCs are comprised of members elected through popular majority. Furthermore, administrative systems provide special provisions to promote social and geographic equity in CC decisions (for example, through quotas for female representatives, and the requirement that each village in the Commune be representedon the Planningand BudgetCommitteeCHECK). 8 Provincial Rural DevelopmentCommittee/ ExecutingCommittee. 9 A sungkut i s the urbanequivalent to the rural Commune. 178 Sharinggrowth: equip and developmentin Cambodia Rules of contest. With minor modification, the CDC's Local Development Fund becamethe Commune Council's discretionarybudget, known as the Commune Sangkat Fund (CSF), tied to a local development planning process and budgeting regulations that reflectprocedures introducedandrefinedunder the CDCs. Politics of resource allocation. Although broad and imprecise (see below), the 2001 LAMC assigned to CCs a broad mandate and powers in relation to this mandate. As importantly, these laws expressed the fundamental principle that resources must be made available to CCs commensurate with their mandate. This effectively shifted the focus of local governance from an administrative process of `planning' toward an essentially political process ofcontest around the `budget' and a concept of entitlement. Social acceptance. The relatively smooth electoral process and qualitative research seems to indicate that people believe that CCs have become embedded as a permanent feature of Cambodian governance (Kim and Ojendal 2007). While there are clearly limits to what CCs are able to do (see below), it is nonetheless true that these institutions are in most places at most times seen to be an improvement upon the unelectedCommune authoritiesthat precededthem. While the 2001 legislation has helped consolidate and formalize institutional arrangements conducive to political equity that were originally introduced through Seila project law, such legislation cannot on its own be expected to ensure this embedding occurs. For each of the above indications, there remain strong countervailing pressures that weaken their effect. Thus, the operations of key line departments (e.g., health or forestry) are not aligned with the territorial authority of the CCs; and the C/SCs still have little or no mandate to manage local natural resources and in the process raise revenue). While most donors and NGOs recognize the authority of the CC, most planning and funding decisions run in parallel with the CC's procedures. Finally, public acceptance of CCs is qualified by a common perception that they are seldom able to respond to local demands (especially for the resolution of local disputes) and not themselves immune from corruption. The situation with regard to the DIW is clearly different, but perhaps not fundamentally so: the constraints on the DIW are connected to those that threaten the durability o f CCs. The DIW remains an artifact of project law (although correspondingly supported by sub-decrees and regulations): the consolidationachieved when CDCs were reformulated into the CC structure has not occurred with respect to the DIW. However, the Strategic Framework for Decentralization and Deconcentration Reforms (D&D Framework) approved in June 2005 reiterates a political commitment to local representative democracy as the primary way of making the state accountable to Cambodian citizens: it also commits the Government to creating durable institutional arrangements that encapsulate the strategic principles reflected in the DIW. These include a `two way' accountability process between CCs and District/Province administrations, in which local and national priorities are to be reconciled; and the subsidiarity principle that a `unified' local administrationwill be needed for this purpose. How the government intendsto achieve this is far from clear. But from international experience, institutional arrangements of the sort promoted by the DIW-that is, multi- level planningand budgeting, supportinga process of political negotiation,administrative Institutions and equi& in the developmentprocess 179 performance and public scrutiny-will need to be created. In light of the `demonstration effect' of the CDC, it is plausible that the administrative familiarity with the DIW may result in its incorporation in subsequent mainstreamregulations. On the other hand, there is a frustration with the DIW process. Despite the fact that it facilitates around 40,000 agreements each year, it has proven largely ineffective in extending its remit beyond Seila-funded activities: in particular, it has yet to influence how Province line departments and other significant non-Seila mechanisms for planning and resource allocation operate (IndependentStudy 2006; GTZ 2005). Interim institutional solutions and "good contests" The cases o f labor law and local governance above suggest that technical short-cuts, using aid as an incentiveto reform, often have limited long-termeffect, precisely because the effectivenesso f formal institutions depends critically on the degree to which they are embedded in informal institutions (values, expectations, negotiating styles, ability to draw on support in other groups in society, and so on). An implication of this is that stronger, more sustainable reforms depend heavily upon the degree to which they are perceivedas legitimate in local context. This legitimacy in turn depends heavily upon the processes by which change is achieved, rather than simply the form of the arrangements that are achieved. To a great extent, what then maters is that there are fair rules and appropriate forums in which debate between contesting parties can occur and new organizationalarrangements can emerge. A number of the other cases o f equity-enhancing policies cited elsewhere in this report reinforce these conclusions. The evolution of the systematic land titling program, for example, can be considered a good example of significant improvement through a cautious, evolutionary approach to the improvement of regulatory institutions. And in those MOPS case study villages where testimony suggests domestic violence is beginning to decline, it is because the law on the book is reinforced by local actors disseminating information, facilitating discussion and connecting citizens to institutional actors (notably NGOs) that can support their case. In this case, too, the outcomes are short of the ideal: abusive husbands are jailed locally for a matter of days as a warning, without cases proceeding to court and the far heavier sentences that application of the statute law would mandate. Nonetheless, it is progress: wives, neighbors, police and local authorities no longer regard domestic violence as a purely private issue; victims now know where to seek solidarity and material support; abusers now expect punishment; and outcomes, in terms o f the incidence of violence as reported in focus group discussions, appear to have improved markedly. As an interim step on a road to the ultimate goal o f strict application of the law by state institutions o f justice, the current situation-in which existence of the law backedup by The roles of civil society and external development partners Organized civil society has a potentially valuable role to play in promoting the ideal o f equity in Cambodia's national development process. Community organizations and NGOs can help poor and marginalizedgroups in society to achieve a collective voice in the policy process, channeling concerns to policy-makers at both local and national levels. Together with an independent press, these organizations help improve 180 Sharinggrowth: equip and developmentin Cambodia communication from the grassroots to senior policy-makers, helping officials to identify when problems are occurring and what solutions might be most likely to succeed. Civil society also plays a key role in establishing a culture of transparency and accountability. Institutional arrangements that create a space for citizens, individually or collectively, to feed back to the upper level of Government can help to identify both when more commitment is needed, and when policies are failing in implementationdue to corruption or capacity bottlenecks. For their part, Cambodia's official development partners need to ensure that they effectively support Governmentcommitments to equitable development. To do this, they need to fulfill their commitments under international and national agreements on ownership, harmonizationand alignment, moving away from donor-driven models of aid (that is, models characterized by tied spending, projects and externally-determined technical assistance), and towards genuine partnerships around collectively-agreed development goals. References ADB. 2001. Participatory poverty assessment in Cambodia. ADB. Manila. http://www.adb.org/Documents/Books/Participatorv Povertv/Participatorv P0verty.D -AccessedJune5th2007. df Adler, Daniel, Doug Porter and Michael Woolcock. 2007. Legal pluralism and the role of interim institutions: challenges and opportunitiesfor equity in Cambodia. Mimeo. World Bank. Working Draft-25 April 2007. AHRC. 2007. Cambodia: code of ethics for judges will not be effective without compliance mechanisms. Asian Human Rights Commission. May 28 2007. http://www.ahrchk.net/statements/mainfilements/lO34/ Accessed May 31 2007. Anand, Sudhir and Amartya Sen 1994. Sustainable human development: concepts and priorities. http://hdr.undp.ora/docs/publications/ocational papers/Oc8a.htm Background paper to UNDP 1994. New dimensions of human security (Human Development Report 1994). OUP: Oxford and New York. Accessed 31 May 2007. Angrist, Joshua and Alan Krueger. 1999. "Empirical strategies in labor economics." chapter 23 in Orley Ashenfelter and David Card (eds.) Handbook of labor economics.Amsterdam: Elsevier Annear, Peter, David Wilkinson, Meng Rithy Chean and Maurits van Pelt. 2006a. "Increasing access to health services for the poor: health financing and equity in Cambodia." Mimeo . Annear, Peter, David Wilkinson, Men Rithy Chean and Maurits van Pelt. 2006b Study ofJinancia1 access to health services for the poor in Cambodia. Phase 1: Scope, design, and data analysis. For the MoH, WHO, AusAID and RMIT University. 30th April 2006. Phnom Penh. Annear, Peter, David Wilkinson, Maryam Bigdeli and Lo Veasna Kiri. 2006. "Improving access to health services for the poor: a review of health Equity Funding in Cambodia." Mimeo Aranvind, Vinod. 2005. Social relations and local level power structures in Cambodia: a review of the literature. Unpublished discussion paper: World Bank Cambodia Country Office. Ballard, Brett M. and So Sovannarith. 2004. Cambodia land titling program: baseline surveyproject. Final report: Rural phase I.PhnomPenh: CDRI. Barslund, Mikkel (forthcoming). "Credit" chapter 6 in Katleen van den Broeck and Finn Tarp (eds.). Vietnam access to resources household survey 2006. Descriptive Report. Development EconomicsResearchGroup, University o f Copenhagen. Brereton, H.2005, Final report: evaluation of the SEILA Gender Mainstreaming Strategy 2001-2005. PhnomPenh: Seila. 182 Sharinggrowth: equig and development in Cambodia. Bourgignon, Franqois. 2004. Thepoverty-growth-inequality triangle. Paper presented at the Indian Council for Research on International Economic Relations, New Delhi, February4 2004. Bratton, M.and N.de Walle (1997) Democratic experiments in Afiica. Bray, M. 1999. The private costs of public schooling: household and community financing of primary education in Cambodia. Paris: UNESCO IIEP. http:l/unesdoc.unesco.orrr/imaaes/OO11/001176/117632E.pdf Accessed 31 May 2007. Bray, M. and S. Bunly. 2005. Balancing the books: household financing of basic education in Cambodia. Comparative Education Research Centre The University of Hong Kong and World Bank, Development Unit, East Asia & Pacific Region. Butzler, Rita, Yair Mundlak, and Donald F. Larson. "Intersectoral migration in Southeast Asia: evidence from Indonesia, Thailand and the Philippines." Journal of Agricultural and Applied Economics. Supplemenet 35 (2003): 105-117. Calderdn, Cesar and ServCn, Luis. 2004. The effects of infrastructure development on growth and income distribution. WashingtonDC. CAS and World Bank. 2006a. Justicefor the poor? An exploratory study of collective grievance over land and local governance in Cambodia. Phnom Penh. October 2006. CAS and World Bank. 2006b. Towards institutional justice? A review of the work of Cambodia's Cadastral Commission in relation to land dispute resolution. Phnom Penh: CAS and the World Bank. CDRI. 2007a. "We are living with worry all the time." A participatory poverty assessment of the TonleSap. Editedby Brett Ballard. PhnomPenh. April 2007. CDRI 2007b (forthcoming) Moving out ofpoverty study. PhnomPenh. Chan Sophal, Tep Saravy and Sarthi Acharya. 2001. Land tenure in Cambodia. A data update" Working Paper 19.PhnomPenh: CDRI. Chan Sophal and Sarthi Acharya. 2002. Facing the challenges of rural livelihoods. A perspectivefrom nine villages in Cambodia. Working Paper 25, PhnomPenh: CDRI. Conway, Tim. 1999. Poverty, participation and programsmes: international aid and rural development in Cambodia. Unpublished PhD. Department of Geography, University of Cambridge. Conway, Tim and Andrea Crossland. 2002. Review of mechanisms to improve equity in access to health care, Cambodia. DFID Health Systems Resource Centre. London. www.odi.orrr.uk/PPPG/publications%5C~apers reports%5Cdfid%5Ccountries%5Cca m02.pdf Accessed 3 1 May 2007. CRD. 2004. Garment employees in Cambodia: a socio-economic survey. Cambodian Researchers for Developmentfor ADB. February2004. 183 References Dahlberg, Elisabet. 2005. Insights into migration and spending patterns based on a small-scale study of garment workers in Phnom Penh. Working Paper 221. Stockholm School of Economics. December 2005. Demobynes, G. and B. ozler. 2005. "Crime and local inequality in South Africa." Journal of Development Economics 76. Deininger, Klaus, and Lyn Squire. "New ways of looking at old issues: inequality and growth". Journal of DevelopmentEconomics. Vol. 57 (1998): 259-287. Deininger, Klaus. 2005. Quantifying the poverty impact of land policy interventions: The case of Cambodia. mimeo, the World Bank Delvert, Jean. 1958. 1961. Lepaysan Cambodgien. Paris, The Hague:Mouton and Co. Deutsch, Robert. 2006. Beneficiary assessment of land titles recipients under Land Management and Administrative Project (LMAP). Prepared for MLMUPC Ebihara, May. 1968. Svay, a Kher village in Cambodia. Unpublished PhD dissertation. Columbia University. Available from University Microfilms, Ann Arbor. Economist. 2007. "Invasion o fthe landgrabbers." March loth- 16th,p. 58. Frings, Viviane. 1993. The failure of agricultural collectivization in the People's Republic of Kampuchea (I979-1989). Working Paper No. 80. Centre for Southeast Asian Studies, MonashUniversity. Frings, Viviane. 1994. "Cambodia after decollectivization (1989-1992)" Journal of ContemporaryAsia. 24 (1). Gill, Indermit and Homi Kharas. 2007. An east Asian renaissance. Ideasfor economic growth. World Bank. Washington DC. Global Witness. 2005. Natural resource governance-a test of political will for the Cambodian Government and the international donor community. 29 June 2005. http://www.globalwitness.org/media library detail.~h~/383/en/naturalresource gov ernance a test o f political wi Accessed 31 May 2007. Gottesman, Evan. 2003. Cambodia after the Khmer Rouge: inside thepolitics of nation- building. Yale University Press. Grosh, Margaret and Paul Glewwe. 2000. Designing household survey questionnaires for developing countries: lessonsfiom I 5 years of the Living Standards Measurement Survey. The World Bank. WashingtonDC. Herrera, Santiago and Pang, Gaobo. 2005. Efficiency of public spending in developing countries: an efficiency frontier approach. March 2005I Holsinger, D et al. 2004. Education inequality in the Cambodian labor force: measurementsand implications. August 2, 2004. Hughes, Caroline. 2003. The political economy of Cambodia's transition, 1991-2001. RoutledgeCurzon. IRL.2007. Perceptions of inequality in Cambodia: a survey. PhnomPenh:Indochina Research Limited. 184 Sban'nggrowtb:equip and developmentin Cambodia. Kane, Thomas J, Elena Rouse and Douglas Staiger. 1999. "Estimating returns to schooling when schooling is misreported." NBER Working Paper 7235. Keller, Sheryl and J. Brad Schwartz. 2001. Final evaluation report: contracting for health services pilot project (CHSPP)-a component of the Basic Health Services Project. PhnomPenh. November 2001. Killick, Tony. 2002. Responding to inequality. Inequality Briefing 3. ODI. London. www.odi.org.uk/PPPG/publicationslbriefings/inequalitv briefinad03.pdf Accessed 31 May 2007. Knowles, James C. 2004. The poverty reduction impact of Cambodia's health programme. Refining pro-poor planning, budgeting and monitoring in the health sector. Oxford Policy Managementfor DFID. July 18 2004. Knowles, James C. 2005. Health, vulnerability and poverty in Cambodia. Analysis of the 2004 Cambodia Socio-Economic Survey. Report to EAS Country Units, World Bank, WashingtonDC. October 31 2005. Knowles, James C. 2006. A new set ofpoverty indicatorsfor Cambodia, 1993/4 to 2004. Report to the EAS Country Units of the World Bank, Washington DC. Original report 2005; revisedversion 28 November2006. Labonne, J., D. Biller and R. Chase. 2006. Trust and inequality in the Philippines. (Draft) Ledgerwood, Judy. 1990. Changing Khmer conceptions of gender: women, stories and the social order. UnpublishedPhD dissertation. Cornel1University / Ann Arbor. Mansuri, Ghazala and VijayendraRao. 2004. "Community-basedand driven development: a critical review." WorldBank ResearchObserver Markussen, Thomas. 2005, Property rights and productivity in Cambodian crop agriculture. mimeo, the World Bank. Second draft. March 2007. Markussen, Thomas. 2007. Land titling in Cambodia-findings from a field trip. mimeo, the World Bank. Second draft. March 2007. McKay, Andrew. 2002. Defining and measuring inequality. OD1Inequality Briefing 1 March 2002. Overseas Development Institute. London. McKay, Andrew, 2002. Defining and measuring inequality. Inequality Briefing 1. ODI. London. http://www.odi.org.uk/PPPG/publications/briefin~s/ineaualitvbriefindo1.pdf Accessed 31 May 2007. Mekong Economics. 2005. Vietnam inequality report 2005: assessment and policy choices. Mekong EconomicsLtd. SynthesisPaper ofthe "DFID Drivers of Inequality in Vietnam" Project. Hanoi. Third draft September 18 2005. http://www.1nekon~economics.com/Document/Publications/2005~KE%20Vietnam %20lneauality%20Report%202005.pdf Accessed 3 1 May 2007. MoH. 2006. Health Coverage Plan 2004-5 with atlas. Department of Planning and Health Information, Ministry of Health. March 2006. I85 References www.moh.gov.kh/content/docunients/l/BMC BB To Y020KY020ChamHCP06.pdf Accessed 3 1 May 2007. Moser, Caroline and Andy Norton with Tim Conway, Clare Ferguson and Polly Vizard. 2001. To claim our rights: livelihood security, human rights and sustainable development. ODI. London. www.odi.org.uk/PPPG/publications/books/tcor.pdf Accessed 31 May 2007. MOWA. 2005. Violenceagainst women: a baseline survey. IndochinaResearchLimited for Ministry of Women's Affairs. PhnomPenh. MOWVA and PADV. 1996. Household survey on domestic violence in Cambodia. PhnomPenh: Ministry of Women's and Veterans' Affairs and ProjectAgainst Domestic Violence. Naschold, Felix. 2002. Why inequality mattersfor poverty. OD1 Inequality Briefing 2 (March 2002). Overseas Development Institute. London. http://www.odi.org.uk/PPPG/publications/briefings/ineQuali~briefings/02.pdf Accessed 31 May 2007. NIPH, NIS and ORC Macro. 2006. Cambodia demographic and health survey 2005. Phnom Penh and Calverton, Maryland. December2006. NIS, MoH and ORC Macro. 2001. Cambodia demographic and health survey 2000. Phnom Penh and Calverton, Maryland. June 2001. Oxfam GB. 2007. Report on land holding in Cambodia. Oxfam GB Cambodia: Land and Fishery Program. Phnom Penh. 12 March 2007. Phnom Penh Post. 2006. "Ethnic kids face language barriers." Eva Shum. July 14-27 2006. Ragatz, Andy. 2005. Cambodia Poverty Assessment - Education Section. September 2005. RGC. 2004. TheRectangular Strategyfor growth, employment and equity. PhnomPenh. RGC. 2006.National Strategic Development Plan, 2006-2010 (NSDP). PhnomPenh. Ridao-Cano,Cristobal. 2004. Cambodia-skills and growth. Working Paper Series No. 2004-2. Human Development Sector Unit, East Asia and Pacific Region, World Bank. May 2004. Rosenboom, Jan Willem, Jeremy Ockelford, Andy Robinson. 2007. Water supply and sanitation in Cambodia: poor access for poor people. Background paper commissionedfor World Bank equity report: Water and Sanitation Program/ Oxford Policy Management. Rusten, Caroline, Kim Sedara, EngNetra, and Pak Kimchoeun. 2004. Thechallenges of decentralisation design in Cambodia. CDRI MonographNo. 1. Phnom Penh Sakellariou, Chris. 2007. Schooling, inequality and earnings in Cambodia. Background paper for World Bank equity report. Mimeo, February2007. I86 Sharinggrowth: equip and developmentin Cambodia. Schonberg, Steven. 2007. Land distribution for employment and growth: Cambodias Social Land Concession Program. Background note for Cambodia Equity report. Draft, March 25 2007. Sen, Amartya. 1995.Inequality re-examined. Clarendon Press. Sen, Amartya. 1999. Development asfreedom. OUP. Oxford. Sen, Amartya. 2002. "Why health equity?" Health Economics 11: 659-666 So Sovannarith, et. al. 2001, Social assessment of land in Cambodia. A field study. Working Paper 20, Phnom Penh: CDRI. Taliercio, Rob. 2006. Cambodia: sources of growth. Growth and public expenditure policy. World Bank. Mimeo. November 2006. Tanzi, Vito. 2004. Measuring eficiency in public expenditure. Paper prepared for the World Bank conference on "Public Expenditure Efficiency and Growth:" World Bank. October 27, 2004 Teang Tnaut. 2005. Land not for sale - A brief study of informal settlements, landlessness and housing rights issues in Sihanoukville, Koh Kong and Kampot, Cambodia. http://www.teangtnaut.org/research.html.Accessed 31 May 2007. UNIFEM, WB, ADB, UNDP and DFIDLJK. 2004.A Fair Share For Women: Cambodia Gender Assessment. UNIFEM, WB, ADB, UNDP, DFIDLJK. Phnom Penh, April 2004. Urashima, Cheryl, Ingrid FitzGerald and UNFPA Cambodia Country Office. 2007. Gender inequalities: background paper for the World Bank Cambodia equity report. Phnom Penh. May 2007. US Embassy. 2006. Twofreedoms: freedom from corruption andfreedom to own land. Remarks by Ambassador Joseph A. Mussomeli, Pact Cambodia Anti-Corruption Conference. http://phnompenh.usembassv.gov/sp 092806.html Accessed 31 May 2007. Vallely, T., D. Dapice and B. Wilkinson. 2006. A SWOT analysis of Cambodia: Executive Summary. Harvard University Kennedy School of Government, for UNDP Cambodia. Van den Brink, Thomas, Binswanger, Bruce, Byamugisha. 2006. Consensus, confusion and controversy; selected land reform issues in Sub-Saharan Africa. World Bank Working Paper No. 71, The World Bank Van de Walle, Dominique. 2007. Land inequality in Cambodia. Unpublished manuscript: World Bank. VanLeit, Betsy. 2007. Secondaryprevention of disabilities in the Cambodian Provinces of Siem Reap and Takeo: perceptions of and use of the health system to address health conditions associatedwith disability in children. HandicapInternational Belgium. Wong, Susan and Scott Guggenheim. 2005. "Community-driven development: decentralization's accountability challenge. pp. 253-267 in East Asia decentralizes: making local government work. Washington, DC: World Bank. I87 References World Bank. 2002. Project appraisal document on a proposed credit in the amount of SDR 19.3 million to the Kingdom of Cambodia for a Land Management and Administration Project. WashingtonDC: World Bank World Bank. 2003. Land policies for growth and poverty reduction. World Bank Policy ResearchReport. World Bank and OUP. June 2003. World Bank Group. 2004. Cambodia seizing the global opportunity: investment climate assessment and reform strategy. Report No27925-KH. Prepared for the Royal Cambodian Government by the World Bank Group. World Bank. 2004. Cambodia: quality basic educationfor all. January 2005. World Bank. 2005. WorldDevelopment Report (WDR) 2006: Equity and development. World Bank I Oxford University Press. Washington and New York. htm://econ.worlclbank.org/VC'BSITE/EXTERNAL/EXTDEC/EXTRESEARCH/EX ?1~'DRS/~XT\~DR20(3G/O..contentMDI(:20:64167689-~iPK:641676 73- tlieSitePI<:477642.00.html Accessed 4 June 2007. World Bank. 2006. Halvingpoverty by 2015? Carnbodiapoverty assessment. World Bank. 2007a. Introduction to oil and gas. Petroleum Sector BriefingNote No.1. March 2007. World Bank. 2007b. Oil and gas: a blessing or a curse? PetroleumSector BriefingNote No. 2. April 2007. World Bank and Asian Development Bank. 2003. Enhancing service delivery through improved resources allocation and institutional reform. Integrated Fiduciary Assessment and Public ExpenditureReview (IFAPER). Cambodia. World Bank and RGC. 2007 (forthcoming). Cambodia health PETS-Public expenditure tracking survey. Draft, March 2007. Zimmerman, Cathy. 1995. Plates in a basket will rattle: domestic violence in Cambodia. Asia Foundationand UNDP. Phnom Penh Annex I Inequalitv andgrowth - combarators. I88 Annex 1 Inequality and growth: Cambodia in context of ASEAN and rest of world Table 1.3 Inequality, growth, and per capita GNP: Cambodia cf other ASEAN countries, 1961-2005 comparison time per capita GNP at Gini coefficient (%) cumulative GDP periods the mid-pointof the growth rate for the time period analyzed time period analyzed (in 2000 US$) (W Cambodia 1993-97 226 34.7 29 1998-04 297 40.3 56 Indonesia 1961-65 195 33.3 4 1966-70 212 30.7 32 1971-75 276 34.6 36 1976-80 355 42.2 38 Thailand 1961-65 366 41.3 34 1966-70 473 42.6 39 1971-75 590 41.7 26 Malaysia 1961-65 877 45.6 30 1966-70 1044 50.0 25 Vietnam 1991-95 266 35.7 40 1996-00 364 35.5 28 2001-05 444 37.6 34 Laos 1990-94 241 30.4 27 1995-99 299 37.0 27 2000-04 351 34.6 26 Source: World Development Indicators (http://ddp-ext.worldbank.org/exVDDPQQ/); Global inequality dataset of Deininger and Squire (1996) at http://econ.worldbank.org/wbsite/external/extdec/extresearch/ O,,contentMDK: 20699070-oaaePK:64214825-piPK:64214943-theSitePK:469382,OO.html Figure 1.9 Inequality in Cambodia and the rest of the world 80.0 70.01 t ~8 60.0 - 4 . * ****.**e .A* *, . .-P) f 50.0 - *e **8 . 40.0 - 4 .$6 **3& #.** $ 60) C a r n b o d i k k ; .-c 30.0 - in '94 and '0%" *,+* 5 20.0 - I::: 3 4 5 6 7 8 9 10 11 ~ In(per cap GDP in 2000 US$) Source. World Development Indicators Annex 2 - Empiricalpamework and regression restlltsfor urban inequalig ana4Ji.r. I89 Annex 2: Empirical framework and regression results for analysis of urban inequality To assess the extent of structuraltransformationthat has taken place in urban areas, the following empirical framework was used: There are two wage (or consumption) equations: K L 1°gyi = c a 9 7 , k w k , ~ + c P97,kX/,r + Y97'1 +p97.1 for 997 (1) k=l /=I where y, represents real wages or consumption; W, = { w ~ , ~ are the dummy variables If=, correspondingto the 4 education levels; X, = {x/,~ are the dummy variables correspondingto the 12 industriesand 2, is the vector including all the other independent variables such as gender, age, and marital status. Estimates the equations (1) and (2) for each individual are: 1°gj, = '9 7 y +b 9 7 x ~ +797'1 for 1997 (3) log>, = '04y + +704'1 for 2004 (4) The wage or consumption difference during the period 1997-2004can be decomposed as the follows: l0gTOb,-l o g 7 9 7= ' 9 7<%,-K7)+ (&04 -' 9 7 1 6 4+ b 9 7 (F04 --'9 7 +( b o 4 - b 9 7 1'04 + (5) fJI -f0Ao where (F4K,)denotes - the main effects of educationwhich signify the changing educationstock of the population; (hod -&97) denotes the year effects of education which indicatethe changes in returnsto education; (xo4-x9,)denotes the main effect of industry that signifL a structuraltransformationor the changing industry composition; and (bo4-b9,)denotes the year effects o f industry which suggest the changes to industry-specificpremiums. First, a sub-sample of paid employees with wages in urban areas was analyzed. Then, the entire sample of urbanhouseholds with consumption approximatingincomewas analyzed. The results are similar. 190 Sharing growth: equity and development in Cambodia Table 1- Estimation of wage equations (Paid employees) Dependantvariable: logarithmof real wage: standarderrc in paranteses 1sI I 2c Model 1 Model 2 Model 1 Model 2 Male 0.091 0.139 0 073 . 0.139 (0.050)' (0.050)"' (0.030)" (0.030)"' Age -0.011 0.012 0.011 0.032 (0.010) (0.010) (0.010) (0.010)"' Squared age 0.011 -0.015 -0.022 -0.044 (0.020) (0,010) (0.010)" (0.010)"' Head of household -0.011 0.001 -0.057 -0.018 (0.060) (0.060) (0.040) (0.040) Married 0.062 0.020 0.056 0.087 (0.060) (0.050) (0.040) (0.030)" Immigrant 0.248 0.235 -0.071 -0.073 (0.090)"' (0.080)*** (0.030)" (0.030)" Work hours in the past 7 days 0.014 0.010 0.012 0.009 (...)"' (...)"' (.,.)"' (...)`e' Highest grade level successfully completed (ref.=Class 3 or below) Class 4-6 0.196 0.240 0.168 0.174 (0.080)** (0.080)"' (0.050)*** (0.050)"' Class 7-9 0.009 0.144 0.195 0.251 (0.070) (0.070)" (0.040)"` (0.050)**' Class 10-12 -0.091 0.111 0.341 0.468 (0.070) (0.070) (0.050)"` (0.050)"' Class 12 or above -0.079 0.113 0.473 0.637 (0.080) (0.080) (0.050)"* (0.050).*' Industryduring the past 7 days (ref.= Public administrationand defense; compulsory social security) Agriculture, hunting and forestry; Fishing 0.255 0.341 (0.130)' (0.070)"' Mining and quarrying; Manufacturing: Electricity,gas, and water supply 0.579 0.762 (0.070)"' (0.050y Construction 0.637 0.543 (0.080)- (0.060)"' Wholesale and retailtrade; repair of motor vehicles, motorcycle and personal and household goods 0.688 0.370 (0.110)"` (0.080)"' Hotels and restaurants 0.492 0.557 (0.180)"' (0.080)*** Transport, storage and communications 0.593 0.763 (0.080)"` (0,080)"' Financial intermediation -0.126 0.494 (0.120) (0.160)"` Real estate, renting and business activities 0.209 0,124 (0.180) (0.180) Education 0.371 0.245 (0.080)"' (0.050)"' Health 0.065 0.125 (0.130) (0.100) Other community, social and personal service activities 0.432 0.604 (0,090)"' (0.050)"' Extra-territonal organization and bodies 1.388 1.046 (0.140)"' (0,090)"' Constant 10.798 10,086 10.925 10.053 (0.230)"' (0.220)"' (0.150)"` (0.150)"' RZ 0.075 0.219 0.089 0.189 Number of observations 1299 1299 3089 3089 Notes: " ..." signifiesthat the absolutevalue is inferiorto 0.001 *** indicates coefficient significantat 1% level: ** indicates coefficient significantat 5% level; indicates coefficientsignificant at 10% level. Source: Royal Government of Cambodia Ministryof Planning, National Institute of Statistics: Cambodia Socio-economic Survey 1997 and 2004. Sharinggrowth: equity and developmentin Cambodia 191 Table 2 Decomposition of wage difference between 1997 and 2004 (Phnom Penh and other urban) Contribution of vari IS variables to wage diierence] Education lndustrv Other Wage Main Year variable difference effect effect S Percentage 100.0 15.2 6.9 43.7 Absolute value Total 0.554 0.084 0.038 0.242 Agriculture, hunting and forestry; Fishing 0.007 0.005 Mining and quarrying; Manufacturing; Electricity, gas, and water supply 0.030 0.032 Construction 0.002 -0.009 Wholesale and retail trade; repair of motor vehicles, motorcycle and personal and household goods -0.011 -0.011 Hotels and restaurants 0.014 0.003 Transport, storage and communications -0.024 0.005 Financial intermediation 0.003 0.005 Real estate, renting and business activities -0.001 -0.001 Education 0.007 -0.013 Health ... 0.001 Other community, social and personal service activities 0.046 0.030 Extra-territorial organization and bodies 0.013 -0.011 Notes: Group of reference: Public administration and defense; compulsory social security. " ..." signifies that the absolute value is inferior to 0.001, Source: Royal Government of Cambodia Ministry of Planning, National Institute of Statistics: CSES 1997, CSES 2004. Figure 1 Contribution of various industries to wage growth (Phnom Penh and other urban) Main effect Year effect Agriculture, hunting and forestry; Fishing 7 Financial intermediation Mining and quarrying; Manufacturing; Electricity, 8 Real estate, renting and business activities gas, and water supply Construction 9 Education Wholesale and retail trade; repair of motor vehicles, 10 Health motorcycle and personal and household goods HGotels and restaurants 11 Other community, social and personal service activities Transport, storage and communications 12 Extra-territorial organization and bodies I92 Sharing growth: equity and development in Cambodia Table 3 Summary statistics of urban heads of households (i.e., in Phnom Penh and other urban areas), 1997 and 2004 1997 2004 Gini Coefficient 0.433 0.432 Real per capita consumption (Riel) 3905 4924 Male (%) 72.7 75.6 Married (%) 75.2 78.4 Sectorllndustry during the past 12 months (%) Primarysector Agriculture, hunting and forestry; Fishing 24.7 23.4 Secondary sector 10.1 9.8 Mining and quarrying; Manufacturing;Electricity,gas, and water supply 6.2 5.7 Construction 3.9 4.0 Tertiary sector 50.2 55.2 Wholesale and retail trade; repair of motor vehicles, motorcycleand personaland householdgoods 18.4 18.7 Hotels and restaurants 0.2 0.8 Transport, storage and communications 8.6 9.0 Financialintermediation 1.3 0.5 Real estate, renting and business activities 0.8 0.6 Public administrationand defense; compulsory social security 14.3 13.4 Education 2.2 3.9 Health 1.o 1.3 Other community, social and personalservice activities 2.7 5.7 Extra-territorialorganizationand bodies 0.8 1.4 Missinginformation 15.0 11.6 Highest grade level successfully completed(%) Class 3 or below 35.8 28.7 Class 4-6 22.1 22.6 Class 7-9 24.4 24.8 Class 10-12 10.2 11.6 Class 12 or above 7.5 12.3 Number of observations 2200 3494 Source: CSES 1997; CSES 2003. Table 4 Effects of education and industry on household consumption differences between 1997 and 2004 for all urban households Main effects (changes in the stock of Year effects Total effects schooling attainment or composltlon of industry) (changes in returns to schooling or In industry premia) education 13.5% 35.2% 48.7% industry 2.5% 46.8% 49.3% Source: CSES 1997: CSES 2003 Shan'nggrowtb:egtriz) and development in Cambodia 193 Table 5 Rates of return to various levels of schooling and returns to sectors of employment (Le., sector or industry premium), 1997 and 2004 Highest grade level successfully completed (ref.= Class 3 or below) 1997 2004 Class 4-6 0.154 0.187 (0.040)*** (0.030)*" Class 7-9 0.166 0.263 (0.040)"' (0.030)*** Class 10-12 0.363 0.445 (0.050)**' (0.040)"* Class 12 or above 0.400 0.609 (0.060)"' (0.040)"' Sector of head's main occupation during the past 12 months (ref.= Primary sector) Secondary sector 0.252 0.307 (0.050)*** (0.040)"* Tertiary sector 0.368 0.506 (0.040)"' (0.030)*** Source: CSES 1997; CSES 2004 Note: These coefficients are obtained from running a standard Mincerian regression with dependant variable as logarithm of real consumption, and other explanatory variables include age, gender, marital status etc (see details in ANNEX); standard error in parentheses; *** denotes significant at the 1% level. 194 Sharing wowth: eauitv and develoDment in Cambodia Table 6 Estimation of consumption equation (urban household heads) Dependant variable: logarithm of real per capita consumption Phnom Penh and other urban 1997 2004 Model 1 Model 2 Model 1 Model 2 Male -0.059 -0.017 0.000 0.050 (0.050) (0.050) (0.040) (0.040) Age 0.030 0.029 0.030 0.028 (O.OlO)**' (0.010)*** (O.OlO)*** (0.010)"' Squared age -0.024 -0.024 -0.024 -0.024 (0.010)'** (0.010)*** (O.OlO)*** (0.010)"' Married 0.130 0.135 0.059 0.052 (0.050)** (0.050)'* (0.050) (0.040) Immigrant 0.194 0.202 -0.181 -0.185 (0.060)"* (0.060)*** (0.020)"' (0.020)"' Work hours in the past 7 days 0.003 0.003 0.001 0.001 (0.000)*** (O,OOO)'*" (0.000) (O.OOO)* Household size -0.093 -0.092 -0.101 -0.102 (0.010)"' (0.010)**' (0.010)"* (0.010)- Highest grade level successfully completed (ref.= Class 3 or below) Class 4-6 0.154 0.147 0.187 0.170 (0,040)"' (0.040)"* (0.030)"' (0.030Y Class 7-9 0.166 0.154 0.263 0.245 (0.040)"* (0.040)"" (0.030)*** (0.030)" Class 10-12 0.363 0.333 0.445 0.424 10.050)*" (0.050)'" (0.040)"' (0.040Y Class 12 or above ~I0.400 0.376 0.609 0.58i ~ ' ' (0.060)*** (0.060)*** (0.040)"* (0.040)"' Sector of head's main occupation during the past 12 months (ref.= Primary sector) Secondary sector 0.252 0.307 (0.050)"' (0.040)"* Tertiary sector 0.368 0.506 (0.040)*** (0.030)'** Industry of head's main occupation during the past 12 months (ref.= Agriculture, hunting and forestry; Fishing) Mining and quarrying; Manufacturing- Electricity, gas, 0.386 0.463 and water supply (0.060)*** (0,050)"' Construction 0.022 0.077 (0.070) (0.050) Wholesale and retail trade; repair of motor vehicles, 0.416 0.576 motorcycle and personal and household goods (0.040)'"' (0.030)*" Hotels and restaurants 0.358 0.396 (0.250) (0.1lo)*** Transport, storage and communications 0.227 0.326 (0.050)*** (0.040)"' Financial intermediation 0.550 0.673 (0.110)"' (0.150)"* Real estate, renting and business activities 0.439 0.893 (0.140)*** (0.140)- Public admin & defense; compulsory social security) 0.348 0.547 (0.050)'" (0,040)"' Education 0.388 0.434 (0.090)"' (0,060)"' Health 0.563 0.648 (0.120)"' (0,090)'** Other community, social and personal service activities 6.393 ' 0.464 (0.080)*'* (0.050)"' Extra-territorial organization and bodies 0.543 ' 0.705 (0.140)'** (0.090)"' Constant 7.146 7.128 7.326 7.355 (0.170)"' (0.170)**' (0.140)*** (0.140)''' 0.237 0.254 0.352 0.373 ~ ~~ Number of observations i823 1823 2945 2945 Notes: The standard errors are presented in parentheses. *** indicates coefficient significant at 1% level; ** indicates coefficient significant at 5% level; * indicates coefficient significant at 10% level. Source: Royal Government of Cambodia Ministry of Planning, National Institute of Statistics: CSES 1997, CSES 2004. Sharinggrowth: egtlig and development in Cambodia 195 Table 7 Decomposition of consumption difference between 1997 and 2004 (Phnom Penh and other urban) Contribution of variout ariables to consumption difference Education Industry Other Difference Main effect Year effect Main effect Year effect variables Percentage 100.0 13.5 35.2 2.5 46.8 2.1 Absolute value Total 0.181 0.024 0.064 0.004 0.085 0.004 Mining and quarrying; Manufacturing; Electricity, gas, and water supply -0.002 0.004 Construction 0.000 0.002 Wholesale and retail trade; repair of motor vehicles, motorcycle and personal and householdgoods 0.001 0.030 Hotels and restaurants 0.002 0.000 Transport, storage and communications 0.001 0.009 Financial intermediation -0.005 0.001 Real estate, renting and business activities -0.001 0.003 Public administration and defense; compulsory social security -0.003 0.027 Education 0.007 0.002 Health 0.002 0.001 Other community, social and personal service activities 0.012 0.004 Extra-territorial organization and bodies 0.004 0.002 Notes: industry: industry of head's main occupation during the past 12 months. Group of reference:Agriculture, hunting and forestry; Fishing. "..." signifies that the absolute value is inferior to 0.001, Source: Royal Government of Cambodia Ministry of Planning, National Institute of Statistics CSES 1997, CSES 2004.. Annex 3 - How can titling translate intopoveq reduction? 196 Annex 3: How can titling be translated into poverty reduction? For simplicity, the titling program is assumed to work in the following way: householdsaffected by the program initially have none of their plots titled, and will have all their plots titled after the intervention. The assumption implies that the number of households affected by the program will equal the number of titles issued divided by the average number of plots per household among potential program beneficiaries. We also assume that poor householdsare equally likely as non-poorhouseholdsto havetheir plots titled. The poverty impact of the program is calculated in the following way: The poverty rate among potential beneficiaries (householdsthat own land without titles), is estimated to be 42 percent. Then, we estimate how much agriculture income would increase as a result o f titles. Then, we simulate the increase of total income, which is proportionateto the share of agriculture income in total income. The increase in total income is assumed to translate fully to household consumption. Based on simulated post-titling consumption o f these households, we calculatethe poverty rate among beneficiariesafter the program has been implemented. Note that all this is based on the assumption that beneficiary selection is uncorrelated with pre-program poverty status. The number of people escaping poverty as a result of the titling program is computed as the number of households escaping poverty times the average household size among potential beneficiary householdsthat are poor beforethe program (this number is 5.7). The change in the national poverty rate is computed as the total number of poor divided by the total population. This model of simulation ignoresthe following effects that are potentially important: Multiplier effects. The increased consumption of farmers will spill over into the non- farm sector. This is not taken into account in the model, which is a source of down- ward bias in the impact estimates. Effects on other non-crop income. Land titles might affect the ability to invest in other sectors than crop agriculture, because land can be offered as collateral for loans. Therefore titling might also affect income from non-agricultural enterprises, livestock-raising and so on. This is not accountedfor in the simulation. General equilibrium effects. Imagine that titling works by increasing agricultural investment. Then if massive titling were to take place, aggregate demand for agricultural investment goods would increase, which would affect the price of those goods might and production could increase, which in turn prompts more investment and further multiplier effects in the economy. Network-effects. Titles to some extent have the properties of "network goods" - the more other people have them, the more valuable a title is for an owner. One reason for this is that titles make it easier to transact land. Therefore, the more other people have titles, the more potentialtransactionpartners one has, but this potential can only be realized, if one has titled land. These effects would imply increasing marginal returnsto titling. Table 1. Projected Population year Population Male Female (Million persons) (Million persons) (Million persons) 1962 5.7 2.9 2.8 1982 6.9 3.1 3.8 1992 9.0 4.1 4.9 1993 9.3 4.3 5.0 1994 9.9 4.7 5.2 1995 10.9 5.2 5.6 1996 11.6 5.6 6.0 1997 11.9 5.8 6.1 1998 12.1 5.9 6.3 1999 12.4 6.0 6.4 2000 12.6 6.1 6.5 2001 12.8 6.2 6.6 2002 13.0 6.3 6.7 2003 13.3 6.4 6.9 2004 13.5 6.6 6.9 2005 13.8 6.7 7.1 2006 14.1 6.8 7.2 2007 14.4 7.0 7.4 2008 14.7 7.1 7.5 2009 15.0 7.3 7.7 2010 15.3 7.4 7.8 201I 15.6 7.6 8.0 2012 15.9 7.8 8.2 2013 16.2 7.9 8.3 2014 16.6 8.1 8.5 2015 16.9 8.3 8.7 2016 17.3 8.5 8.8 2017 17.6 8.6 9.0 2018 18.0 8.8 9.2 2019 18.4 9.0 9.4 2020 18.7 9.2 9.5 Sources Statistical Year Book 2005 and CSES 200312004, National Institute o f Statistics Table 2. Pojected population by age and sex, 1998 2005 - Age Group 1998 1999 2000 2001 2002 2003 2004 2005 In niimher ~~ Both Sexes 12,132,172 12,35 1,382 12,573,580 12,802,982 13,040,668 13,287,053 13,542,4 I O 13,806,923 0 303,248 322,557 326,691 333,344 341,299 349,75 1 358,494 367,445 1 - 4 1,617,243 1,501,614 1,392,980 1,290,659 1,226,417 1,265,354 1,294,227 1,327,545 5 -9 1,815,638 1,850,836 1,901,630 1,95 1,379 1,964,089 1,869,04 1 1,776,591 1,675,892 I O -14 1,698,245 1,727,619 1,744,740 1,754,525 1,765,741 1,786,935 1,822,625 1,873,971 I 5 1,376,726 1,469,639 1,540,633 1,593,402 1,634,974 1,670,707 1,700,413 1,7 18,220 20 24 763,696 834,101 949,426 1,090,594 1,230,203 1,348,435 1,440,152 1,5 10,626 25 29 9 10,000 873,480 816,272 758,947 729,146 745,2 13 814,497 927,835 30 34 801,585 824,348 853,667 880,868 893,892 885,741 850,809 795,772 35 39 712,675 733,673 746,18 1 754,239 763,541 777,717 800,4 20 829,6 26 40 44 509,072 540,047 578,372 619,591 657,494 687,863 708,632 721,358 45 49 425,977 437,290 445,490 454,000 467,370 488,421 5 18,496 555,703 50 54 320,009 334,698 352,407 371,355 389,07 1 404,060 415,198 423,484 55 263,135 268,716 273,902 279,768 287,626 298,279 312,343 329,304 60 64 209,945 214,313 219,781 225,953 232,205 238,179 243,586 248,727 65 69 170,959 172,909 174,271 175,656 177,646 180,654 184,839 190,043 70 74 ------------19 59 1 14,924 120,489 125,308 129,390 132,695 135,265 137,224 138,828 75' 119,095 125,053 131,829 139,312 147,259 155,438 163,864 172,544 Female 6,274,554 6,384,294 6,495,047 6,609,280 6,727,537 6,850,016 6,976,854 7,108,140 0 l48,65 1 158,242 160,178 163,407 167,273 171,381 175,631 179,982 1 - 4 793,400 737,4 15 684,961 635,500 604,419 623,354 637,206 653,295 5 -9 889,829 908,076 933,274 957,587 963,813 9 17,708 873,052 824,342 I O -14 826,549 840,085 849,559 856,600 864,488 876,666 895,079 920,529 I S - 19 696,500 737,689 766,677 786,3 13 801,466 815,742 829,444 839,221 20 24 401,044 435,980 493,004 562,338 629,900 685,554 726,4 19 755,389 25 29 472,720 455,019 427,264 399,492 385,228 393,467 428,017 484,347 30 34 422,557 433,749 447,772 460,576 466,539 462,396 445,335 4 18,473 35 ---- 39 379,487 387,817 393,756 398,662 404,365 41 1,975 423, I57 437,184 40 - 44 304,526 317,027 330,678 344,578 357,419 368,326 376,653 382,726 45 49 246,697 256,418 265,278 273,953 283,356 294,052 306,324 3 19,778 50 54 -- 184,399 193,203 203,485 214,625 225,724 236,180 245,660 254,368 55 150,285 154,008 157,692 I 61,860 167,138 173,920 182,398 192,326 60 64 121,251 123,926 127,108 130,675 134,386 138,105 141,704 145,318 65 69 98,593 100,161 101,442 102,73 1 104,288 l06,3l 1 108,882 111,950 70 74 ---- 59 67,025 70,504 73,474 76,017 78,157 79,94 1 81,438 82,775 75+ 71,041 74,975 79,445 84,366 89,578 94,938 100,455 106,137 Male 5,857,618 5,967,088 6,078,533 6,193,702 6,3 l3,13 1 6,437,037 6,565,556 6,698,783 0 154,597 I64,3 15 I66,j I 3 169,937 174,026 178,370 182,863 187,463 1 - 4 823,843 764,199 708,O 19 655,159 621,998 642,000 657,021 674,250 5 -9 925,809 942,760 968,356 993,792 1,000,276 951,333 903,539 851,550 10 -14 871,696 887,534 895,18 1 897,925 901,253 910,269 927,546 953,442 15 680,226 73 1,950 773,956 807,089 833,508 854,965 870,969 878,999 20 24 362,652 398,121 456,422 528,256 600,303 662,88 1 713,733 755,237 25 29 437,280 4 18,46 1 389,008 359,455 343,918 351,746 386,480 443,48 8 30 34 379,028 390,599 405,895 420,292 427,353 423,345 405,474 377,299 35 39 333,188 345,s 56 352,425 355,577 359,176 365,742 377,263 392,442 40 44 204,546 223,020 247,694 275,013 300,075 319,537 33 1,979 338,632 45 49 179,280 180,872 180,212 180,047 184,014 194,369 212,172 235,925 50 54 135,610 141,495 148,922 156,730 163,347 167,880 169,538 169,116 55 112,850 114,708 116,210 117,908 120,488 124,359 129,945 136,978 60 64 88,694 90,387 92,673 95,278 97,819 100,074 I O 1,882 103,409 65 69 72,366 72,748 72,829 72,925 73,358 74,343 75,957 78,093 70 74 ------------19 59 47,899 49,985 51,834 53,373 54,538 55,324 55,786 56,053 75 48 054 .-,-- . 50,078 52,384 54,946 57,681 60,500 63,409 66,407 Sources:First Revision Population Projection for Cambodia 1998-2020 National lnstitue of Statistics (NIS) and Royal University of Phnom Penh (RUPP). Table 3. Projected population by province and sex, 1998 2005- Province / Municiplity 1998 1999 2000 2001 2002 2003 2004 2005 In number Both Sexes 12,132,172 12,351,382 12,573,580 12,802,982 13,040,668 13,287,053 13,542,4 I O 13,806,923 Banteay Meanchey 638,892 657,322 675,363 693.805 712,795 732,3 I O 752,392 773,092 Battambang 887,507 901,565 915,876 930,726 946,332 962,672 979.8 23 997,840 Kampong Cham ,691,798 712,448 ,734,143 ,756.657 1,780,286 804,956 ,830,722 ,857,500 Kampong Chhnang 440,144 449,604 459,206 469,120 479,457 490,220 501,455 513.179 Kampong Speu 630,782 643,865 657,005 670,487 684,476 698,949 713,967 729,552 Kampong Thom 599.695 610,559 621,504 632,752 644,429 656,474 668,895 681,692 Kampot 557.322 562,647 568,501 574,600 58 1,072 587,868 595,036 602,624 Kandal ,129,876 144,426 ,159,181 ,174,438 1 ,190,445 207,096 ,224,433 ,242,506 Koh Kong 138,309 146.288 153,544 160,892 168,366 175,947 183,648 191,468 Kratie 276,706 284,428 292,206 300,139 308,287 3 16,606 325 097 I 333,761 Mondul Kiri 34,406 35,492 36,594 37,708 38,854 40,016 41,201 42,4 12 Phnom Penh ,039,607 1,077,286 I,]14,479 1,152,572 1,191,668 1,231,630 ,272,386 1,313,851 Preah Vihear 126,389 129,844 133,372 136,988 140,711 144,521 148,407 152,378 Prey Veng 995,211 1,000,220 1,006,270 1,012,811 1,019,936 1,027,5 18 ,035,672 1,044,376 Pursat 390,004 394,576 399,290 404,266 409,621 415,354 421,528 428,173 Ratanak Kiri 99,703 102,506 105,344 108,260 111,290 114,426 117,664 121,028 Siem Reap 734,126 750,818 767,768 785,204 803,289 821,964 841,268 861,214 Sihanoukville 163,303 169,395 175,591 181,926 188,447 195,130 201,981 209,005 Stung Treng 85,602 88,068 90,562 93,552 95.72 1 98,396 101,136 103,933 Svay Rieng 503,080 507,402 512,093 5 16,882 521,872 527,053 532,478 538,180 Takeo 833,381 841,642 850,397 859,531 869,157 879,220 889,777 900,880 Oddor Meanchey 82,345 84,413 86,5 1 1 88,647 90,843 93,077 95,410 97,803 KeP 30,123 31,132 32,168 33,221 34,3 1 1 35,434 36,592 37,786 Pailin 23,861 25,436 26,6 I 2 27,798 29,003 30,216 31,442 32,690 Male 5,857,618 5,967,088 6,078,533 6,193,702 6,313,131 6,437,037 6,565,556 6,698,783 Banteay Meanchey 3 13,721 322,854 331331 340,999 350,470 360,204 370,224 380,559 Battambang 435,451 442,582 449,873 457,421 465,389 473,718 482,462 491,652 Kampong Cham 817,293 827,522 838,306 849,490 861,315 873,655 886,569 900.010 Kampong Chhnang 208,865 213,741 218,697 223,800 229,136 234,683 240,468 246,499 Kampong Speu 303,251 309,871 316,550 323,393 330,522 337,892 345,542 353,489 Kampong Thom 288,096 293,593 299, I 61 304,874 310,833 316,972 323,307 329,837 Kampot 267,487 270,280 273,359 276,552 279,954 283,512 287,260 291,225 Kandal 543,278 550,829 558,532 566,474 574,841 583,526 592,566 601,990 Koh Kong 70,85 I 75,070 78,933 82,835 86,805 90,827 94,906 99,045 Kratie 137,075 141,061 145,085 149,180 153,396 157,697 162,084 166,559 Mondul Kiri 17,384 17,954 18,532 19,l 13 19,715 20,325 20,946 21,580 Phnom Penh 501.922 519,384 536,675 554,4 I O 572,689 591,407 610,528 630,O I 6 Preah Vihear 62,951 64,814 66,72 I 68,670 70,678 72,726 74,814 76,943 Prey Veng 469.443 471,749 474,626 477,782 48 1,280 484,957 488,941 493,215 Pursat 187,378 189,75 1 192,188 194,751 l97,5 I 4 200,467 203,640 207,048 Ratanak Kiri 49,124 50,519 5 1,93 1 53,380 54,895 56,465 58,087 59,774 Siem Reap 355,722 364,035 372,509 381.21 7 390,291 399,655 409,345 419,362 Sihanoukville 80.800 83,898 87,052 90,272 93,591 96,992 100,473 104,042 Stung Treng 42,4 I 5 43,647 44,897 46.6 I 4 47,479 48,8 19 50, I92 51,591 Svay Rieng 237,371 239,533 241,897 244,304 246,831 249,449 252,194 255,082 Takeo 398,523 402,677 407,l 14 41 1,743 416,652 42 1,774 427,150 432,811 Oddor Meanchey 41,577 42,700 43,844 45,001 46,194 47,400 48,664 49,952 Kep 14.757 15.296 15.848 16.405 16.983 17.577 18.189 18.820 Pai'lin 12,883 13,728 14,372 15,022 15,678 16,338 17,005 17,682 Table 3. Projected population by province and sex, 1998 2005 - Province / Municiplity 1998 1999 2000 2001 2002 2003 2004 2005 In number Continued Female 6,274,554 6,384,294 6,495,047 6,609,280 6,727,537 6,850,016 6,976,854 7,108,140 Banteay Meanchey 325,171 334,468 343,532 352,806 362,325 372,106 382,168 392,533 Battambang 452,056 458,983 466,003 473,305 480,943 488,954 497,361 506,188 Kampong Cham 874,505 884,926 895,837 907,167 918,971 931,301 944,153 957,490 Kampong Chhnang 23 1,279 235,863 240,509 245,320 250,321 255,537 260,987 266,680 Kampong Speu 327,531 333,994 340,455 347,094 353,954 361,057 368,425 376,063 Kampong Thom 311,599 3 16,966 322,343 327,878 333,596 339,502 345,588 351,855 Kampot 289,835 292,367 295,142 298,048 301,118 304,356 307,776 311,399 Kandal 586,598 593,597 600,649 607,964 615,604 623,570 631,867 640,5 16 Koh Kong 67,458 71,218 74,61 1 78,057 81,561 85,120 88,742 92,423 Kratie 139,631 143,367 147,121 150,959 154,891 158.909 163,013 167,202 Mondul Kiri 17,022 17,538 18,062 18,595 19,139 19,691 20,255 20,832 Phnom Penh 537,685 557,902 577,804 598,162 618,979 640,223 661,858 683,835 Preah Vihear 63,438 65,030 66,651 68,318 70,033 71,795 73,593 75,435 Prey Veng 525,768 528,471 531,644 535,029 538,656 542,561 546,731 551,161 Pursat 202,626 204,825 207,102 209,5 15 212,107 214,887 217,888 221,125 Ratanak Kiri 50,579 51,987 53,413 54,880 56,395 57,961 59,577 61,254 Siem Reap 378,404 386,783 395,259 403,987 412,998 422,309 43 1,923 441,852 Sihanoukville 82,503 85,497 88,539 91,654 94,856 98,138 101,508 104,963 Stung Treng 43,187 44,42 1 45,665 46,938 48,242 49,577 50,944 52,342 Svay Rieng 265,709 267,869 270,196 272,578 275,041 277,604 280,284 283,098 Takeo 434,858 438,965 443,283 447,788 452,505 457,446 462,627 468,069 Oddor Meanchey 40,768 41,713 42,667 43,646 44,649 45,677 46,746 47,851 KeP 15,366 15,836 16,320 16,816 17,328 17,857 18,403 18,966 Pailin 10,978 11,708 12,240 12,776 13,325 13,878 14,437 15,008 - Population Figure 3.6 Projected population by sex, 1998 2005 8,000,000 7 ~ 1998 1999 2000 2001 2002 2003 2004 2005 Year 8Male 0Female Source First Revision Population Projection for Cambodia 1998-2020 National lnstitue of Statistics (NIS)and Royal University of Phnom Penh (RUPP) Table 4. Projectedlife expectancy at birth by provinceand sex, 1998-2005 Province / Municiplity 1998 1999 2000 2001 2002 2003 2004 2005 Male Banteay Meanchey 56 8 57 3 58 0 58 6 59 2 59 8 60 5 61 1 Battambang 51 8 52 5 53 3 54 1 54 9 55 7 56 5 57 3 Kampong Cham 49 5 50 2 50 9 51 6 52 3 53 0 53 7 54 5 Kampong Chhnang 44 7 45 4 46 2 47 0 47 8 48 6 49 4 50 2 Kampong Speu 59 4 59 8 60 3 60 7 61 2 61 6 62 1 62 5 Kampong Thom 60 4 60 8 61 1 61 5 61 9 62 3 62 7 63 1 Kampoi 51 3 51 9 52 7 53 4 54 2 54 9 55 6 56 4 Kandal 54 0 54 6 55 3 55 9 56 6 57 3 57 9 58 6 Koh Kong 58 7 59 2 59 7 60 3 60 8 61 3 61 9 62 4 Kratie 58 6 58 9 59 4 59 8 60 2 60 6 61 0 61 4 Mondul Kiri 36 5 37 3 38 1 38 9 39 7 40 5 41 3 42 I Phnom Penh 68 3 68 5 68 6 68 7 68 9 69 0 69 1 69 2 Preah Vihear 58 6 58 9 59 4 59 8 60 2 60 6 61 0 61 4 Prey Veng 48 8 49 5 50 3 51 0 51 8 52 6 53 3 54 1 Pursat 42 6 43 3 44 1 44 9 45 6 46 4 47 2 48 0 Ratanak Kiri 36 5 37 3 38 1 38 9 39 7 40 5 41 3 42 1 Siem Reap 56 8 57 4 58 0 58 6 59 2 59 7 60 3 60 9 Sihanoukville 51 3 51 9 52 7 53 4 54 2 54 9 55 6 56 4 StungTreng 58 6 58 9 59 4 59 8 60 2 60 6 61 0 61 4 Svay Rieng 50 9 51 4 51 9 51 4 53 0 53 5 54 0 54 5 Takeo 52 3 52 8 53 4 54 0 54 6 55 2 55 7 56 3 Oddor Meanchey 56 8 57 2 57 6 58 0 58 4 58 8 59 2 59 6 KeP 51 3 51 8 52 5 53 1 53 8 54 4 55 0 55 7 Pailin 51 8 52 4 53 I 53 7 54 4 55 0 55 6 56 3 Female Banteay Meanchey 63 I 63 6 64 3 64 9 65 6 66 2 66 8 65 5 Battambang 57 8 58 5 59 3 60 1 60 9 61 7 62 5 63 3 Kampong Cham 55 2 55 9 56 6 57 3 58 1 58 8 59 5 60 3 Kampong Chhnang 49 9 50 6 51 4 52 2 53 0 53 8 54 6 55 5 Kampong Speu 65 8 66 2 66 7 67 1 67 6 68 1 68 5 69 0 Kampong Thom 66 8 67 2 67 6 68 0 68 4 68 8 69 2 69 5 Kampot 57 2 57 9 58 6 59 4 60 1 60 9 61 6 62 4 Kandai 60 1 60 7 61 4 62 1 62 8 63 5 64 1 64 8 Koh Kong 65 1 65 6 66 2 66 7 67 2 68 8 68 3 68 8 Kratie 65 0 65 3 65 8 66 2 66 6 67 0 67 4 67 9 Mondul Kiri 40 3 41 0 41 8 42 6 43 4 44 3 45 1 45 9 Phnom Penh 74 8 74 9 75 1 75 2 75 3 75 5 75 6 75 7 Preah Vihear 65 0 65 3 65 8 66 2 66 6 67 0 67 4 67 9 Prey Veng 54 5 55 2 55 9 56 7 57 5 58 3 59 1 59 9 Pursar 47 4 48 1 48 9 49 7 50 5 51 3 52 1 52 9 Ratanak Kiri 40 3 41 0 41 8 42 6 43 4 44 2 45 1 45 9 Siem Reap 63 2 63 7 64 3 64 9 65 5 66 1 66 7 67 3 Sihanoukville 57 2 57 9 58 6 59 4 60 1 60 9 61 6 62 4 StungTreng 65 0 65 3 65 8 66 2 66 6 67 0 67 4 67 9 Svay Rieng 56 8 57 3 57 8 58 3 58 9 59 4 60 0 60 5 Takeo 58 3 58 9 59 5 60 0 60 6 61 2 61 8 62 4 Oddor Meanchey 63 2 63 5 63 9 64 4 64 8 65 2 65 6 66 0 Kep 57 2 57 8 58 4 59 1 59 7 60 4 61 0 61 7 Pailin 57 8 58 4 59 1 59 7 60 4 61 0 61 7 62 3 Source First Revision Population Projection for Cambodia 1998-2020 National lnstitue of Statistics (NE)and Royal University of Phnom Penh (RUPP) Table 6. Percentage of population by religion, age and sex: 1998 and 2004 GPCC 1998 CPS 2004 Total Religion Age Group Popul- ation Ism 1Christia- 1Christi- Not anity Other Reported Both sexes 100.0 96.4 2.1 0.5 0.8 0.2 100.0 96.4 2.5 0.3 0.8 0.0 0 - 4 1000 963 2 3 0 3 1 0 0.2 100.0 96.5 1.9 0.2 1.3 5 - 9 1000 963 2 3 0 4 0 8 0.1 100.0 96.4 2.4 0.2 0.9 0.0 I O - 14 1000 963 2 3 0 4 0 7 0.2 100.0 96.5 2.6 0.2 0.7 0.0 1 5 - 19 1000 967 1 9 0 4 0 7 0.3 100.0 96.1 2.9 0.3 0.7 0.0 20 - 24 1000 965 1 8 0 6 0 9 0.3 100.0 96.4 2.8 0.3 0.5 0.0 25 100 0 965 1 9 0 5 0 9 0.3 100.0 96.3 2.1 0.5 1.1 0.0 30 34-- 29 100 0 964 2 0 0 5 0 8 0.2 100.0 96.8 2.0 0.5 0.7 0.0 35 - 39 100 0 962 2 2 0 6 0 9 0.2 100.0 96.7 2.0 0.2 1.0 0.1 40 - 44 100 0 961 2 3 0 7 0 8 0.2 100.0 96.8 2.2 0.4 0.6 0.0 45 49 100 0 961 2 2 0 6 0 9 0.1 100.0 96.5 2.5 0.3 0.6 0.0 50 54 100 0 963 2 1 0 6 0 9 0.1 100.0 95.8 3.3 0.4 0.5 55 59 -- ~ 100 0 966 2 0 0 5 0 8 0.2 100.0 95.9 3.1 0.4 0.7 60 64 1000 963 2 1 0 4 0 9 0.3 100.0 95.7 3.5 0.2 0.6 65 69 100 0 965 2 1 0 4 0 8 0.3 100.0 96.7 2.7 0.3 0.3 70 -- 74 ~ 1000 961 2 2 0 4 1 0 0.4 100.0 96.6 2.3 0.3 0.6 0.2 75+ 100 0 954 2 3 0 3 1 1 0.9 100.0 95.2 3.6 0.2 0.9 0.1 Over all male 1000 96.3 2.1 0.5 0.8 0.2 100.0 96.4 2.5 0.3 0.8 0.0 0 - 4 1000 96.3 2.3 0.4 0.9 0.2 100.0 96.6 2.0 0.2 1.2 5 - 9 100.0 96.3 2.3 0.4 0.8 0.2 100.0 96.5 2.3 0.3 0.9 0.0 10- 14 1000 96.2 2.4 0.4 0.8 0.2 100.0 96.6 2.6 0.2 0.6 0.0 15-19 1000 96.7 1.9 0.4 0.7 0.3 100.0 96.3 2.9 0.2 0.5 0.0 20 24 ~ 100.0 96.6 1.7 0 6 0.8 0.4 100.0 96.5 2.7 0 2 0.5 0.0 25 - 29 1000 96.5 1.8 0.5 0.8 0.3 100.0 96.3 2.1 0.5 1.1 30 - 34 100.0 96.4 1.9 0.6 0.8 0.3 100.0 96.8 2.0 0.5 0.7 0.0 35.39 100.0 96.0 2.3 0.6 1.0 0.2 100.0 96.8 1.8 0.3 1.o 0.1 40 44 100.0 95.6 2.4 0.8 1.0 0.2 100.0 96.5 2.3 0.4 0.7 0.0 45 49 -- 1000 95.7 2.3 0.7 1.1 0.1 100.0 96.0 3.0 0.2 0.8 50 - 54 100.0 96.1 2.1 0.7 1.0 0.1 100.0 95.8 3.1 0.5 0.7 55 - 59 100.0 96.3 2.0 0.5 1.0 0.1 100.0 96.1 2.8 0.5 0.7 60 100 0 96.2 2.1 0.5 1.0 0.2 100.0 95.3 3.8 0.2 0.7 65 69-- 64 100.0 96.4 2.1 0.5 0.9 0.2 100.0 95.7 3.2 0.5 0.6 70 - 74 100.0 96.0 2 1 0.4 1.2 0.3 100.0 97.7 1.4 0.6 0.2 75+ 1000 94.9 2.1 0.3 1.4 1.3 100.0 95.2 3.3 0.4 0.9 0.2 Over all female 100 0 96.4 2.1 0.4 0.8 0.2 100.0 96.4 2.5 0.3 0.8 0.0 0 - 4 1000 96.2 2.3 0.3 1.0 0.1 100.0 96.5 1.9 0.2 1.4 5 - 9 IO00 96.3 2.4 0.4 0.8 0.1 100.0 96.4 2.5 0.2 0.9 I O - 14 100.0 96.4 2.3 0.4 0.7 0.1 100.0 96.4 2.6 0.2 0.7 0.0 1 5 - 19 1000 96.6 2.0 0.4 0.7 0.2 100.0 95.9 2.9 0.3 0.9 20 - 24 1000 96.4 1 9 0.6 0.9 0.3 100.0 96.3 2.9 0.3 0.5 25 - 29 100.0 96.5 1.9 0.5 0.9 0.2 100.0 96.3 2.1 0.5 1.1 0.0 30 - 34 100.0 96.5 2.1 0.5 0.8 0.1 100.0 96.9 2.0 0.4 0.7 35 39 * 100.0 96.4 2.1 0.5 0.8 0.1 100.0 96.6 2.2 0.2 1.0 40 44 100.0 96.4 2.2 0.6 0.8 0.1 100.0 97.0 2.1 0.3 0.5 45 - 49 ~ 100.0 96.4 2.1 0.5 0.8 0.1 100.0 96.9 2.2 0.3 0.5 0.0 50 - 54 100.0 96.4 2.1 0.5 0.9 0.2 100.0 95.8 3.4 0.4 0.4 55 - 59 100.0 96.8 1.9 0.4 0.7 0.2 100.0 95.7 3.3 0.4 0.6 60 - 64 100.0 96.3 2.1 0.4 0.9 0.3 100.0 95.9 3.3 0.3 0.4 65 69 ~ 100.0 96.6 2.0 0.3 0.6 0.4 100.0 97.5 2.3 0.2 70 - 74 100.0 96 1 2.2 0.3 0.8 0.4 100.0 95.9 2.8 0.1 0.9 0.3 75+ 1000 957 2 5 0 4 0 9 0 6 1000 952 3 8 0 1 0 9 Source 1998General Population Census of Cambodia (GPCC 1998) and Cambodia Inter-censalPopulation Survey 2004 (CIPS 2004). National Institute of Statistics (NIS) m ~ m r m a=. w N . . . . . r r r r r N m m N t o w m h m