Report No. 32167-MG Madagascar Development Policy Review Sustaining Growth for Enhanced Poverty Reduction (In Two Volumes) Volume II: Technical Annex May 16, 2005 PREM I Southern Africa Africa Region Document of the World Bank ICC InternationalChamber o f Commerce ICT Information Communication Technology IF Integrated framework IFC International Finance Corporation IGM Gemological Institute o fMadagascar IMF InternationalMonetary Fund INSTAT Institut National de la Statistique IT InformationTEchnology ITC InternationalTrade Center ITQ Information Technology Quality IUCN World Conservation Union JIRAMA Jiro sy Ran0 Malagasy WC Letter o f Credit LDI Landscape Development Interventions LPDR Lettre de Politique de DCveloppement Rural M&E Monitoring and Evaluation MDG MilleniumDevelopment Goals MFA MultiFiber Agreement MFI Microfinance Institution MIGA Multilateral InvestmentGuarantee Agency MSC Marine Stewardship Council MTM Maison duTourisme de Madagascar NEAP National Environmental Action Plan NERICA New Rice for Africa NGO NonGovernmental Organization NRB Natural Resource Base- ODA Official Development Assistance ONE National Environment Office ONTM Office National duTourisme Malgache ORT Office RCgional duTourisme P A ProtectedArea PADR Programme d'Action pour le DBveloppement Rural PER Public Expenditure Review PES Payment for Environmental Services PNRA Projet Nationalpour la Recherche Agricole PNVA Projet Natrionalde Vulgarisation Agricole PPP Public-Private-Partnership PRSP Poverty Reduction Strategy Paper RFT Rkserve Foncitre Touristique RTA Regional Trade Agreement RIAs Regional Integration Arrangements SAC Structural Adjustment Credit SIRAMA Siramamy Malagasy SLA Structural Adjustment Loan SSA Sub-Saharan Africa T&A Textiles and Apparel TEV Total Economic Value TELMA Telecom Malagasy UNIDO UnitedNations Industrial Development Organization USAJSA United States o f America USAID UnitedStatesAgency for InternationalDevelopment VAT Value Added Tax wco World Customs Organization WTO World Trade Organization WWF World Wildlife Fund - 11- .. Major contributors to the technical annex were Jacob. H Bregman, Maria Eugenia Bonilla- Chacin, Patrick RamanantoaninaandAlessandro Nicita (Labor Market andEducation); Jaime de I Melo (Infrastructure); Paul0 de Sa (Mining). Administrative support was provided by Cecile Wodon, Risseme Gabdibe andEmma FabienneRaharinjanahary. -... 111- TABLE OF CONTENTS 1. TRENDS OF POVERTY PROFILE ............................................................................... 1 Poverty by Sector ................................................................................................................ Spatial Poverty..................................................................................................................... 1 Nonmonetary Poverty......................................................................................................... 2 3 2. INFRASTRUCTURE ........................................................................................................ 8 Impact of infrastructure ontrade volume............................................................................. 9 3. LABORMARKETSAND EDUCATION POLICIES ................................................. 11 Labor Markets.................................................................................................................... 11 11 Labor Demand: Labor Force by Economic Sector................................................ Labor Supply.......................................................................................................... 13 Labor Markets: Economic Growth and Poverty.................................................... 14 Simulation-BasedEstimates of Growth on Labor Demand and Poverty .............-16 Education Policy inMadagascar........................................................................................ 19 Skilled-Labor Supply Constraints on the Horizon................................................. . . 17 Madagascar's Education Policy............................................................................ 21 Measures to Improve the Eficiency of the Educational System............................ 22 Conclusions........................................................................................................................ 26 Methodology for Estimating the Effects of Growthon Employment ................................ 27 Labor Supplyand Labor Demand...................................................................................... 31 4. MINING ............................................................................................................................ 35 .iv- FIGURES Figure 1.1:Average Annual Change inIncidenceo f Poverty........................................................ 1 Figure 3.1:Educational attainment profiles across income andplace o fresidence ..................... 23 BOXES 9 Box 2.2: Impact o fpoor infrastructure on trade volume .............................................................. Box 2.1: Status of transport infrastructure ...................................................................................... 10 Box 3.1: Lack o f skilledworkers as a limitation to growth intextile, handicraft, and IT firms in Madagascar ........................................................................................................................... 19 Box 3.2: Growth, Poverty and the Labor Market: Two Extremes ................................................ 32 Box'4.1: The Institute de Gemology de Madagascar (IGM) ........................................................ 40 TABLES Table 1.1: Selected Indicators of non-monetary poverty................................................................ Table 1.2: Poverty according to economic sector ........................................................................... 3 6 Table 1.3:Spatial poverty : Headcount ratio 19893-2002............................................................... Table 1.4: Poverty gap 1993-2002.................................................................................................. 7 7 Table 2.1: The road network and its conditions in2002................................................................. 8 Table 3.1:Labor Force Characteristics across urban and rural areas ........................................... Table 3.2: Employment by Selected Sector in2001..................................................................... 12 13 Table 3.3 : Adjusted Expenditureandpoverty levels across household head's employment sector Table 3.4: Urban labor demand simulations................................................................................. ............................................................................................................................................... 15 17 Table 3.5: Average years of schooling o f adults across place o f residence, income, and gender in Table 3.6: Net enrollment rates across levels of schooling and income levels in2001 in(%)....20 Madagascar ........................................................................................................................... Table 3.7: Percentage increase inthe number of students enrolled by level o f education ...........21 22 Table 3.8: Proportion o f students enrolled ineach level o f education living inpoor households 26 Table 3.9. Multinomial logit estimation (dependent variable: logit sector) ................................. 33 Table 3.10: Wage differential regression (dependent variable: log earnings) .............................. 34 Table 4.1: Madagascar's mineralproduction................................................................................ 35 Table 4.3: Sharingofminingrevenues......................................................................................... Table 4.2: Fiscal Regime under the Law on Large Scale MiningInvestments ............................ 37 42 .v- MADAGASCAR DEVELOPMENT POLICY REVIEW VOLUME I1-TECHNICAL ANNEX The Volume I1o f the Development Policy Review provides selected background and analytical elements underpinningthe storyline developed inthe mainreport -Volume I. Annex Idescribe the poverty trends and profile with an emphasis on the recent growth period (1997-2001) which witnessed an increasing gap between the rural and urban areas on the one side, and the agricultural and the industrial and service sectors on the other side. Annex I describe the significant regional variation among provinces and revisits non-monetary poverty relating to the access to the basic social services and infrastructure. Annex I1provide a diagnosis o f the current status of infrastructure - roads and other industrial infrastructure such as ports, airports, telecommunications- and highlights the impact o f the lack of adequate infrastructureonpoverty and the competitiveness o fthe economy Annex I11deal with the labour markets and education policy. The purpose o f annex I11i s three fold: (i) to describe the labor market and to identify sectors with the highest earnings and first employment growth opportunities; (ii) to assess the labor needs of the sectors identified second, as the most promising for growth (EPZs, mining, tourism and shrimp industry); (iii) to third, outline the reforms needed to match the education policy to labor demand. It provides the technical methodologies used to estimate the impact o f the projected growth on urban employment. Annex IV deal with sector policy. It revisits the mainreforms undertaken inthe miningsector for which no report has been published so far, and outlines the on-going and planned reform programs addressing the identified issues inthe sector. - vi- 1. TRENDS OFPOVERTY PROFILE 1.1 Between 1960 and 2001, Malagasy per capita income has declined from US$430 to US230 and provided the population with few opportunities to improve their living standards. This is reflected in a dramatic increase in poverty. Poverty is widespread with a national headcount average o f 70 percent during the 1990s and reached 73,6 percent in 2003 in the aftermath o f the 2002 crisis. As in most countries, growth and poverty are closely linked, increasing somewhat inperiods o f economic and political turmoil and falling slightly inperiods of recovery, The 1980switnessed the most significant increase inpoverty and similarly, poverty gap has significantly increased during the 1980s, both indicators showing modest swings during the 1990s. Figure 1.1: Average Annual Change in Incidence Madagascar's recent growth had a comparatively of Poverty low overall impact o f growth on poverty; much Averageannualchange In Incidenceof poverty lower than successful countries in South-East I " a m Asia '. More than half and 22.5 percent o f the 1 5'/l population mobilized only a quarter o f the e ' e---- Madagascar incomes respectively in 1999 and in 2001. Mongolia Fplipplnos I t Evidences suggest that growth between 1999 and ao PDR *Vhtnam 8Chlna 2001 was beneficial mainly to the people at the ? i -10 i Malaysia + top o f the distribution or individuals above the Indonesia poverty threshold. -1 5 Poverty i s multidimensional. Poverty persists in, and benefits only, some regions and sectors as Thalland * -25 1 described below. -5 5 10 AwrsfleannualgrowthInpercapitaODP Percent SPATIAL POVERTY 1.2 Poverty is rural in nature with 80,l percent o f the rural population being poor in 2003 compared to a poverty rate o f 51,8 percent in urban areas (Table 1.3 at the end o f 11.). Headcounts inrural areas have been consistently very high, averaging 76 percent between 1993 and 2001. On average, rural population represent 85 percent per cent o f the population, which explains the high rate o f poverty in Madagascar. In contrast with urban poverty which has a strong correlation with economic performance, the trends show that rural poverty has steadily increased, irrespective o f the evolution o f the economic situation. Similarly, poverty remains deep over the same period with poverty gap in rural areas increasing somewhat in the period o f economic recovery as opposed to a decrease o f poverty gap inurban areas (Table 1.4 at the end 'Source :MadagascarCountry Assistance Strategy - 2003. -1- o f 11.). The breakdown according to the socio-economic group reflects the rurallurban structural difference inpoverty distribution inthe country, with consistently higher poverty among people working in the rural sector. All socio economic groups working in urban settings (servicedtrading, workers, salaried workers) are considerably better off. 1.3 Changes in overall poverty also hide significant regional variation. Except Antananarivo -the capital-,allthe fiveprovincesbecomepoorer(at the endofAnnex 2). Theprovinceof Antananarivo, and Antsiranana to a lesser extent, has had the lowest rates o f poverty headcount index and extreme poverty during the last decade, evolving below the national rates. Other provinces* have consistently displayed highpoverty rates. Except inAntananarivo and to a lesser extent Toliary, in general, poverty has been worsening. The trends o f poverty within region follow that o f national headcount index with regional poverty gap improving only for the provinces o f Antananarivo and Toliary between 1993 and 2001. Similarly, regional poverty gap i s becoming deeper in rural areas for all provinces, except in Antananarivo. Although, the significant number o f rural population in some provinces might have accounted for increasing poverty, regional poverty deserves closer scrutiny and further analyses would be usehl to determine the pattern and causes o f differentiated levels ofpoverty among provinces. POVERTY BY SECTOR 1.4 Poverty is inherently an agricultural phenomenon. Households in primary activities, particularly in agriculture, account for the bulk o f national poverty with increasing poverty rates and consistently above that o f the other sectors. The sector contributed to 84 percent o f national poverty in 1999, with poverty rates consistently above 75 percent throughout 1993-1999. The striking feature o f agriculture i s that irrespective o f the size o f farms and location as well as economic performance, poverty has increased for all categories o f farmers in 2001. The size o f population and 75 percent o f the total population - explains the magnitude o f poverty in rural agricultural households in rural areas - agricultural households make up 83 percent o f the rural areas and therefore at the national level. 1.5 Urban agricultural households fared better than rural agricultural households, with the gap widening, particularly during periods o f recovery. Rural agricultural households fared the worst, with increasing poverty rates and accounting for 85 percent o f the rural poor between 1993 and 2001. However, the urban agricultural households, accounting for about one-third o f the urban poor are still the poorest group in urban areas, contributing to 47,l percent o f urban poverty. If small agricultural households are those that fared the worst between 1993 and 2001 with rates consistently above 80 percent, poverty rates for medium- and large scale farming also increased significantly in 2001 with headcount index o f 86,9 percent and 90,8 percent respectively for these two years. 1.6 The evolution o fpoverty inthe primary sector contrasts strikingly with the secondary and tertiary sectors which did very well over the 1997-2001 period, with the poverty headcount and poverty gap decreasing sharply. At more disaggregated levels, households engaged inthe driving In2003, onaverage, 85percent ofthe population inFianarantsoa andMahajangaand 78percent ofthe population inToamasina andToliary were poor. (See Table 1.2 at the end of 11.). -2- sectors during the recent growth period did very well with poverty rates decreasing significantly for sectors such as manufacturing, construction. Similarly households engaged in government services did very well over the period. The decrease in national poverty rates was therefore driven by these sectors. For instance, the improvement in poverty in Antananarivo was almost entirely driven by the very successful EPZ zone around the capital with urban poverty falling from 43 percent in 1997 to 29 percent in2001 although EPZ employment is only 1percent o fthe labor force. The manufacturing sector seems to have played a pivotal role in allocating growth benefits to the household inthe bottom section o f the distribution. The newly created jobs were in majority benefiting unskilled labor while salaries for entry jobs increased significantly between 1999 and 2001.Complementarities working through linkage effects are particularly evident inthe Antananarivo area, as rural poverty also fell from 69 percent to 57 percent. NONMONETARY POVERTY 1.7 The long economic decline impacted adversely on the quality o f public spending and people's quality of life, and the Millenium Development Goals are a far reaching objective, As shown intable 2.1 many social indicators, including some o f MDGs, and access to basic utilities, reflect the low quality o f public service delivery. Ingeneral, public spending is inequitablewith a strong bias against rural areas. Table 1.1: SelectedIndicators of non-monetarypoverty Poverty Indicators 1993 1997 1999 1(simulated after crisis) ~ ServiceDelivery Indicators Net primary school enrolments 48.3 63.1 64.3 64.9 72.1 Repetitionrate inruralprimary n.a n.a 27.9 36.6 38.4 school, 11" grade Pre-natal consultation, percent 78 76.8 68.8 Infant mortality (out o f 1000) 97 99 90 Immunizationrate 43 36.2 38.1 Electricity connections 9.1 11.9 13.7 Water connections 17 19.3 22.5 21.1 26.8 Sanitationconnections 35 33.1 45 52.1 54.7 Source: CAS -Madagascar Social Services 1.8 Education is a key determinant o f poverty. Individuals in households with few or no members who have completed primary education are more likely to be poor (50 percent o f household heads among the poor have not completed primary education) as confirmed by multivariate analysis in Mattia 2001. Not only i s the headcount for households headed by individuals with a primary school education more than 10 percent lower, but poverty i s also considerably less deep, with more people closer to the poverty line. More than 76 percent of the householdheadedby non-educated individuals cannot afford the minimumfood basket. 1.9 Although falling income undoubtedly was key, the inefficiency and inequity in the quality o f public education spending in the past has, in all likelihood, played a critical role in -3- poverty4. With an enrollment rate o f 72 percent inprimary education in2001, only 54.2 percent o f the poorest quintile had access to schools as opposed to a percentage o f 88.1 percent for those at the top o f the distribution. Nearly half o f the population was illiterate (48 percent) in 2001 with a rural contribution o f 61 percent; 60 percent o f the pupils in urban areas complete the primary cycle schooling against only 12 percent o f those living in rural areas. Secondary and higher education display the same pattern. Only 1.6 percent o f the population has a secondary or higher level but access to education is more difficult in rural areas where the percentage o f population not having access to that level o f education is relatively high 53.4 percent. If, for - instance, in urban areas the percentage o f population having reached higher education i s as low as 5.2 percent, the percentage represents only 0.8 percent for rural population. Even if the difference according to the gender does not seem to be so stressed, it appears that the female gender is nevertheless underprivileged. Infact, about half o f women (49.7 percent) have no level o f education against 46.2 percent for the male sex5. The small allocation to the sector inthe past is one o f the reasons accounting for the outcomes ineducation. This trend has been reversed for the last few years. Inthis respect, between 1995 and 2003, the allocation to education increased from 1.5 percent to 3.2 o fthe GDP. 1.10 As regards health service6, public service delivery is also associated with a bias against rural areas. Whereas infant mortality has slightly improved, trends o f the immunization rate and pre-natal consultations suggest a deterioration o f the quality o f health services. Income level i s one o f the main causes o f low recourse to health services: food expenses represent 70.2 percent o f total expenses and households earmark only 2.4 percent o f their expenses to health. As a result, less than half (45.9 percent) o f the patients frequent health centers and the cost of a single consultation (price o f the act, medicines, transport and meals...) represents 5 percent o f the yearly average revenues per poorest household, against 2 percent for the richest households. Income effect i s compounded by the shortage of medical personnel as well as medicines and the distance between health centers and villages or even the absence o f health centers in some localities. Inrural areas, 37.2 percent o fmost health centers are located more than six miles away from the villages. Inaddition, only 21percent o f the population, mainly inurban areas has access to 41 percent o f the health personnel and on average, a physician takes care o f about 10,000 inhabitants. Between 1995 and 2001, the share o f the expenditures increased in relation to the total public expenditures increasingfrom 3.4 percent to 6.7 percent andthe percentage inrelation to the GDP 0.6 percent to 1.5 percent. Yet, the expenditures allocated to basic health services, frequented by the poor, represented only 4 percent o f the expenditures, which i s equivalent to an amount per year per capita o f MGF 10,010 in2000 a long way from MGF 220,000 (USD 34) to cover the costs o fbasic healthservices7. This section onEducationsummarizes the PRSP (2003) based onthe outcomes ofthe HHS 2001. Source : Table 12-13 of the Madagascar PRSP - p.61. 'Recent This section on healthsummarizes the PRSP (2003) based on the outcomes ofthe HHS 2001, economic development doc No4 UNDP, May 2002. -4- BasicUtilitiesandInfrastructure 1.11 Table 1.2 shows that access basic infrastructure, although improving in some aspects is low, If access to electricity and water connections has improved somewhat inthe last five years, new connections to water and electricity have rather favored the rich, unlike sanitation which favored the poorest quintile'. Regarding the source o f energy, apart from petroleum, the poor use of firewood increases the likelihood o f environment degradation. Poverty incidence i s higher in households who have no or little access to infrastructure and basic utilities'. Patemostro showed that poverty rates generally rise monotonically with the degree o f remoteness". Furthermore, distance from the city and road quality affects negatively the production in rural area". Production is lower in regions with high insecurity (zone rouge) which are in general far from urban centers. The need to travel long distances for health services and schools reduces the access to basic services and enhances the risk o fbeingpoor. 'Source: Madagascar - PRSP 2003. Patemostro and a1(2001), Changes inPoverty inMadagascar 1993-1999. loThe remotenessindex i s the weighted sumo f indicators of the existence inthe community of (a) a road, (b) a bus stop, (c) access to agricultural extension services, (d) access to modem fertilizer, as well as the distances to the nearest (e) school and (f) health center. Stifel et a1(2003) :Transactions costs and agricultural productivity : implications of isolation for rural poverty in Madagascar. -5- Table 1.2: Poverty according to economic sector Source: HH-2001 -6- Table 1.3:Spatial poverty :Headcount ratio 19393-2002 1993 1997 1999 2001 2002 OverallPovertyTrend Total Urban Rural Total Urban Rural Total Urban Rural Total Urban Rural Total Urban Rural National 70.0 50.1 74.5 73.3 63.2 76.0 71.3 52.1 76.7 70.1 48.1 76.5 80,7 61,6 86,4 Slight Improvement(Urban) Province Huge Improvement Antananarivo 68.0 42.4 76.2 66.4 52.0 72.1 61.7 43.3 69.3 49.3 29.2 57.5 66 51,2 72,3 (Especially Urban) Worsening(Urban and Fianarantsoa 74.2 64.9 75.3 75.1 83.1 73.6 81.1 55.8 85.9 83.3 59.4 87.8 91 78,5 93,5 Rural) Large Worsening(Urban and Toamasina 77.9 55.8 81.1 79.8 76.3 80.8 71.3 52.6 76.4 83.1 61.1 89.1 86,3 66,9 91,9 Rural) Mahajanga 53.2 37.3 56.7 73.8 68.2 75.1 76.0 65.2 78.8 72.6 50.2 78.5 89,l 71 93,8 Improvement(Urban) Improvement(Urban) Toliara 81.1 66.9 84.2 82.0 69.1 84.9 71.6 66.5 73.1 75.9 50.2 83.4 81,2 58,3 87,s Worsening(Rura1) IAntsiranana 160.2 49.5 63.7 162.3 27.0 69.5 72.6 31.3 80.6 69.7 27.9 79.3 83,7 62,5 88,6 Slight Improvement (Urban) Source: Madagascar PRSP 2003. Table 1.4: Poverty gap 1993-2002 1993 1997 1999 2001 2002 Total Urban Rural Total Urban Rural Total Urban Rural Total Urban Rural Total Urban Rural National 30.3 17.5 33.1 33.6 29.6 34.7 32.8 21.4 36.1 34.8 18.3 39.7 47.6 29.3 53.0 Province Antananarivo 27.8 15.9 31.6 29.1 23.0 31.5 26.0 17.5 29.5 21.0 10.4 25.5 33.7 23.1 38.1 Fianarantsoa 33.7 22.4 35.3 32 42.0 30.1 40.2 25.2 43.1 45.4 25.5 49.4 57.6 42.4 60.7 Toamasina 33.7 18.5 36.0 39.3 39.9 39.2 32.6 21.1 35.7 44.1 28.4 48.2 51.7 33.2 57.1 Mahajanga 18.6 11.6 20.2 29.1 23.2 30.6 36.5 25.3 39.4 35.1 17.3 39.9 57.5 33.6 63.8 Toliara 42.8 25.0 46.5 46.4 37.3 48.5 33.7 29.8 34.8 39.2 25.4 43.5 47.4 25.1 53.5 Antsiranana 22 14.3 24.5 23.9 6.2 27.5 32.0 7.8 36.7 28.9 8.7 34.0 51.4 28.1 56.7 -7- 2. INFRASTRUCTURE 2.1 At independence, Madagascar had good infrastructure. However, it has deteriorated relative to other developing countries. Road infrastructure 2.2 Duringthe last 20 years, Madagascarhas lost an average of 1,000 kmon an annualbasis. From a total network of 50,000 km, 30,000 km remain, of which only 18.5 percent are now in good or fair conditions Table 2.1. Road transport is by far the main transport means used in Madagascar. It accounts for about 90 percent of passengers and freight domestic transport. The densityof road traffic is low, with less than 10percent of the network (2,900km) carrying traffic above 500 vehicles per day which makes it more difficult and costly to maintain.. Type of road Length(km) Goodou fair condition Of whichbituminous Badcondition(km) (km) surface (km) National Roads (RN) 11 862 3 935 3 654 7 927 ProvincialRoads (RP) 12 250 1 070 70 11 180 Local Community 7 500 850 350 6 650 Roads (RC) Total 31 612 5 855 4 074 25 757 Percent 100 18.5 12.9 81.5 IndustrialInfrastructure 2.3 Industrial infrastructure (railways, maritimeand fluvial ports), are also inpoor conditions and/ or have decayed. Infkastructures for concession, such as the twelve main airports, are in more or less acceptable physical conditions, though they also need upgrading, especially regarding safety and security equipments. -8- Box 2.1: Status of transportinfrastructure Ports: Toamasina is the primary port o f call for ocean carriers that link Madagascar to global markets. It serves as the primary cargo switching station for intra coastal traffic. With its existing capacity, the port should have little difficulty coping with its existing level o f traffic or even a substantial increase intraffic. However, as figure below shows, the port i s one o f the least efficient inthe IndianOcean with vessel turnaround times averaging more than 60 hours. Giventhe level o f compensation paid to port workers, port fees are extremely high (comparable to those at Rkunion where wages are ten times higher). The port currently employs 2,700 full-time employees and another 2,000 daily workers. It is estimated that only 800 workers would be needed to operate the port if it were runby a private operator. Airport: Apart from the lack o f standard safety and security equipments, the major problem at Ivato, is its limited capacity. For cargo storage space, the available 1,350m2 o f available storage space (1,000m2 for inbound, and 350m2 for outbound cargo traffic) was already too small to handle efficiently last year's volume. During the two peak air freight seasons, the freight terminal is highly congested and barely serviceable. Additional cargo storage space is urgently needed to handle the expected increase involume. Moreover, neither at Toamisana nor at Ivato are there any cold storage facilities. IMPACT OFINFRASTRUCTUREONTRADE VOLUME 2.4 Box 4.2 shows that if infrastructure had not been deteriorating, Madagascar could have achieved a better performance intrade. -9- Box 2.2: Impact of poor infrastructureontrade volume Carrkre and de Melo (2004) estimate a gravity trade model to quantify the likely reduction in trade volume that could have been accounted for by a deteriorating infrastructure. They estimate the following model over the period 1962-96 using panel data techniques (not to lose time-invariant variables like distance) for Madagascar and its 43 most important trading partners: aI(2) >O,a, Mijt:Totalbilateral imports, incurrent US$, bycountry ifromcountryj at datet, UN-COMTRADE.The original database does not contain any zero. NiGIt:Totalpopulationofcountry i0') at datet, CD-ROM WDI, WorldBank 1999. Yi(i)t:GDP o fcountry i0') at date t, inconstant US$ 1995, CD-ROM WDI, World Bank 1999. Dij: Distancemeasured inkilometersbetweenthe maincity incountry iandthe maincity incountryj. Kia)t:Infrastructure index describedinfigure 1. The criterion for selection is that a trading partner is included inthe sample only iftrade takes place over the entire sample. This gives a sample o f 43 countries and 43*35= 1505 observations that cover over 80 percent o f Madagascar's trade. Results (see below) show that all coefficients have the expected sign and expected magnitudes (is.the income variable coefficient is close to unity as expected by theory), and the distance variable is highly significant, though its magnitude is rather on the high side (usually the Coefficient estimate is in the (-08-1.4) range). All variables o f interest are thus statistically significant and the overall fit is good for cross-section. This is usual in gravity models. O f particular interest here is the value for the Coefficient on the infrastructure coefficient (the index is defined so that a higher value o f the index means a better infrastructure. Taking both imports and exports, on average, they estimate that a 1 percent improvement inthe value o f the infrastructure index would have increased imports or exports by 0.8 to 1percent. They also simulate the volume o f trade Madagascar would have had with an infrastructure index equal to that o f its partners and estimate that even with an index equal to that o f the poorest third among its partners, Madagascar would have had an increase intrade volume o f about 20 percent by the end o fthe period." pGravity modelover l!2-1996 fromMdgto partner! /I ofMdgfrompartneri coeff t-student coeff t-student y mdg 1,09** 9.2 1,35** 2,9 Y partner 1,14** 10,s 8,5 Dmdg-partner -1,84** -13,7 -1,01** 1,99** -15,O Nmdg -1,06** -7,l -1,10** -8,l Npartner -0,12 -1,3 -0,03 -0,3 Kmdg 0,78** 493 0,98** 393 Kpartner 0,25* 2 3 0,25* 2,o Constant 20,16* 2,2 -11,23* -1,7 Number o f obs 1505 1505 R-squared 0,65 0,66 l2Evenifstandard tests on the exogeneity o fvariables were carried out, these results are only suggestive as omitted variable bias and errors inmeasurement must still be present and important, Moreover, some components o f the infrastructure index likeroads and railways are certainly dependent on the volume o ftrade even ifthis was not revealed bythe endogeneity tests. -10- 3. LABOR MARKETSAND EDUCATIONPOLICIES 3.1 Annex 5 is divided in two sections. Section 1 analyzes the labor market. It revisits the current situation in the labor market by (i) offering a description o f the labor force inboth rural and urban areas, identifying the size o f the informal sector and the magnitude o f underemployment; (ii) analyzing the allocation o f the labor force and its characteristics across different sectors and economic activities with an emphasis on workers' educational attainment; (iii)identifying the economic sectors with the highest earnings and employment growth opportunities; (iv) simulating labor demand in urban areas and presenting evidence on labor supply constraints in the economic sectors with high growth potential. Then based on the assessment o f the skills demanded in these sectors, section 2 examines the government's education policy across levels to support the creation o f skills that match the qualifications needed to enter the formal labor market for poverty reductionpurpose. LABOR MARKETS Labor Supply 3.2 The economy in Madagascar is sharply divided between urban and rural areas. While rural areas are overwhelmingly agricultural, the manufacturing sector, as well as a large part o f the service sector is almost exclusively localized inurban areas. There are few contacts between urban and rural labor markets as there i s little temporary or permanent migration from rural to urban areas. Labor markets are poorly developed in rural areas where most o f the labor i s involved in subsistence agriculture. Incontrast, in urban areas, the presence o f larger industrial and service sectors make labor markets more dynamic. 3.3 Table 3.1 reports some characteristics o f the labor force in Madagascar. The unemployment rate i s about 2 percent inurban areas while unemployment is almost not existent in rural areas. Low rates of unemployment are common in developing countries where the subsistence level o f living makes unemployment not an option. In those cases a meaningful measure o f the status o f the labor markets is the rate o f underemployment. Underemployment i s characterized by low hours worked and/or low productivity which translate in low salaries and high poverty rates. In Madagascar almost 30 percent of the labor force in urban areas is ~nderemployed'~,and in rural areas the percentage is 35. At first sight, the large incidence o f underemployment might suggest a vast reserve labor from which economic growth could draw employment. l3Underemployment here is defined as working either less than 7 hours a day, or less than 5 days a week, or less than 20 days per month. -11- Table 3.1: Labor Force Characteristics across urban and rural areas I I Urban Rural Total labor force (Thousand) 20,880 66,120 Unemployment rate ("h) 2 0 Underemployment rate ("h) 29 35 Womenparticipationpate ("h) 60 90 Women underemployment rate (`?A 42 63 Percent o f skilled labor (%) 46 15 11 3.4 The availability o f a substantial pool o f skilled labor in Madagascar signaled good prospects for economic growth but confirmation needs further analysis. Tight markets for skilled labor can slow growth and further increase wage differentials between skilled and unskilled workers. In this regard, the Malagasy labor force in urban areas appears fairly skilled14 (46 percent o f workers are skilled), while in rural areas skilled individuals are scarce (15 percent). However, to measure the capacity o f the labor markets to supply skilled labor it i s important to analyze how much o f the skilled labor force is actually underemployed or employed in low productivity sectors. 3.5 Another indicator ofpossible constraints on the supply o f labor i s the size o f the informal sector, Informal employment i s usually related to uncertainty, underemployment, and high level of poverty. Employment growth in the formal sector often draws workers out o f the informal sector. In doing so it contributes to decreased underemployment, increased wages and reduced poverty. Given the rather large size o f the informal sector inMadagascar (about one third o f the employed in urban areas are working inthe informal sector), economic growth might be fueled by employment drawn from the informal sector without putting substantial pressure on wages andon other sectors o femployment. 3.6 Insum, the pool ofworkers could act as a reserve labor supply from which new demand could draw. Madagascar is still largely a rural economy and because o f low rates o f internal migrati~n'~,migrants are not a major source o f dynamism in the urban labor market. Nevertheless, the structure o f the urbanlabor force suggests that there might be a large potential for growth inboth employment and productivity given the highunderemployment rate, the large informal sector, the possibility o f increasing women's labor participation, and the presence o f a relatively skilled labor force. 3.7 However, this will only be the case if the skills o f the workers in the "reserve" labor are those indemand. Ifthat is the case, it i s unlikely that economic growth will be hindered by tight labor markets, at least inthe short period. However, ifthese workers are underemployedbecause their skills do not match the skills demanded inthe formal labor market, then they will not act as l4Skilled is definedas individuals with nine years o f schooling or individuals with some technical training. l5Migration to Tana i s estimated around 10 thousand individuals per year (or less than one percent of the resident population). Half o f which originating from rural areas o f the Antananarivo region, while the other half from other regions. -12- reserve labor. Scarcity in the skills demanded in growing sectors would put an upward pressure on wages. The following sections will try to elucidate this problem. Labor Demand: Labor Force by Economic Sector 3.8 This section examines the allocation o f the labor force and its characteristics across different economic sectors and identifies the most important sectors o f the economy in terms o f employment andthe type o f labor required. 3.9 The vast majority o f workers are employed in sectors related to land use. As shown in (Table 5.2) about 6.5 million individuals are employed in agricultural activities, much less inthe fishing sector (67 thousand workers), breeding (142 thousand individuals), and other land based activities (72 thousand). The manufacturing sector is still very small, employing about half a million individuals. Textiles and the food industry are the largest employers, absorbing about 36 and 13 percent of the manufacturing workers respectively. Services employ about 1.3 million individuals, a large percentage o f which are employed inprivate services (35 percent) and inthe public administration(16 percent). Table 3.2: Employment by Selected Sector in 2001 ~ ndustry Total Labor Employment percent percent Age Years of percent of force (percent of Female Temporary (mean) Education Employers (Thousand) Total Labor workers Labor force (mean) with Tech force) Education illFarmActivities: 6,819 76 54 18 31.6 6.9 7 Agriculture 6,532 96 55 18 31.4 4.2 6 Fishing 67 1 24 34 34.4 6.8 16 Animal farm 142 2 28 14 37.3 6.9 19 Other 79 1 59 30 36.9 6.6 6 illManufacturing 548 8 44 21 35.8 8.6 28 FoodIndustry 72 13 61 24 38.0 6.9 17 Mining 19 3 18 36 36.8 7.8 20 Textiles 195 36 79 21 33.6 8.8 28 Wood and Handicrafts 64 12 13 26 36.1 8.1 39 Other Industries 151 28 7 38 37.5 9.6 28 ConstructionMat. 47 9 47 22 35.3 7.8 30 illServices 1,343 16 47 17 36.9 8.8 20 Energy 200 15 70 12 36.1 8.5 10 Commerce 108 8 61 16 35.8 8.1 11 Small Commerce 138 10 69 25 34.7 7.2 10 Transport 131 10 1 13 36.9 8.1 25 Tourism 83 6 62 12 32.3 8.4 17 Private Services 471 35 44 24 36.0 8.5 23 Public Administration 211 16 32 8 43.4 12.0 36 Source: HH- 2001 3.10 Table 5.2 also reports the percentage o f female workforce and o f temporary employment ineach economic sector. Among manufacturing, women's employment is concentrated intextile, apparel industry and food industry. Among services, women's employment is concentrated in -13- commerce and tourism sectors while women are underrepresented in the transport sector and in the public administration. The manufacturing sector relies on temporary workers for about 27 percent o f employment. Among manufacturing, the textile and apparel sector employs about 21 percent o f temporary workforce. Inthe service sector about 17 percent o f workers are temporary most o fwhich employed insmall commerce andprivate services. 3.11 Public administration is the sector with the highest average level o f education (12 years) followed by other industrial sector (9.6 years) and textile and apparel (8.8 years). Besides farm activities, the sectors with lower levels of education are small commerce (7.2 years) and the food industry (6.9 years). The tradeoff between age and education is present in almost all economic sectors (since younger individuals are likely to be less skilled). Interestingly the exception i s given by the textile and apparel sector where the workforce i s both younger and more skilled than most o f other sectors. Young labor force is also observed in tourism. Technical education specific to the sector o f employment appears to be important in the wood industry, textile and apparel industry, construction materials, andthe inpublic administration 3.12 Insummary, inthe short to mediumterm, given a fairly vast underemployedlabor force, it appears that the reserve labor sector could supply enough unskilled labor to sustain economic growth without rising labor costs. 3.13 However, there is evidence that the fast growing industries employ a type o f workforce that i s both young and skilled. Therefore economic growth i s likely to be characterized by an increasing demand for young skilled workers which must be matched by adequate education policies, Likely pressures on the labor market will be estimated below. Labor Markets: EconomicGrowth and Poverty 3.14 The labor market serves as the main transmission between economic growth andpoverty, Byincreasingparticipation rates, reducing underemployment, and raisingrealwagesI6, growth is likely to reduce poverty on average. Therefore, it i s important to examine trends inemployment, wages, and eamings across economic sectors. This analysis will help assess whether economic growthandproductivity increases benefit the poor. 3.15 To better understand the potential for growth and poverty reduction on each economic sector it is important to analyze how well workers fared across sectors. For example, if employment growth is limited to low paying sectors and sectors with high poverty rates, this growth will likely do little to reduce overall poverty. Table 3.3 reports the adjusted average income o f the workers and the percentage o f poor in each economic sector.17 Table 3.3 also reports the growth rates o f each sector in terms o f employment and real earnings observed between 1997 and2001. l6 The available evidence suggests that employment status is a strong correlate of poverty (Canagarajah and Mazumdar, 2001; Yemstove, 2001; Berck et al., 2000, O'Connell, 1999 and UNU/WIDER, 2000). l7Data on eamings ineach sector i s fragmentary and o fpoor quality. Following standard practice, earnings are approximated by households expenditure adjusted by numberofpeople working ineach households. Sector o f employment is givenby the one ofthe householdhead. -14- Table 3.3: Adjusted Expenditureand poverty levels across household head's employmentsector Growthin Change in real Industry HHExpenditure Percentageof (USDper month) Poor Employment earnings (yearly percent) (yearly percent) All Farm,Activities: 230 84 3 Agriculture 231 84 3 Fishing 311 50 11 Animal farm 207 70 2 Other landbased act. 306 77 4 0 All Manufacturing 440 42 20 8 FoodIndustry 403 39 4 8 Mining 313 61 15 8 Textiles 511 40 35 8 Wood and Handicrafts 397 38 25 8 Other Industries 451 44 18 8 ConstructionMaterials 411 43 12 8 All Services 374 37 5 7 Energy 460 41 4 7 Commerce 540 23 12 7 Small Commerce 521 49 3 7 Transport 619 26 5 7 Tourism 602 32 7 7 Private Services 436 43 3 7 Public Administration 566 36 5 7 Source: HH-2001 3.16 Among all economic sectors, the service sector is the one with the lowest poverty rates while the agricultural sector fared worst. Only 37 percent o f workers in service are poor. Commerce and transport are the services with the lowest poverty rates of 23 and 26 percent respectively. Petty commerce18fares the worst with almost half of its workers living inpoverty. Finally about 36 percent the public administrationworkers live below the poverty line. 3.17 In manufacturing, about 42 percent of the households reported per capita expenditure below the poverty line. The average expenditure o f households where the head is involved in manufacturing is almost twice as high as those where the head i s involved in agriculture (440 USDper month). Textiles and apparel, the food, and the wood and handicraft industries are the ones with the least percentage o f poor. Incontrast, the miningsector fared the worse with about 60 percent o f its employees inpoverty. 3.18 Compared with other sectors, households obtaining their means o f living from agriculture are among the poorest with a value o f about 230 USD per month." For the vast majority (about 84 percent) o f agricultural workers their level o f expenditure i s insufficient lift them out o f poverty, Among farm activities, fishing and fisheries fare the best in terms o f average Petty commerce was a category defined by the HHand consist o f small trade. This includes not only food consumptionbutall expenditures including imputedrent. Consideringthat the average households has about 4.5 members this corresponds to about 1.6 USD a day, per person. -15- expenditure per month and interm of percentage of poor. This partly reflects the activities o f the shrimp industry which has experienced rapid growth inthe last years. 3.19 To better understand the evolution o f employment in the various sectors and to help analyze the dynamics o f the labor market, Table 3.3 also reports the observed employment growth rate across the different economic sectors. 3.20 The employment annual growth rate in manufacturing between 1997 and 2001 has been extremely high at about 20 percent per year. This was mostly due to the growth in employment inthe textile and apparel sectors (35 percent) and inthe wood andhandicraft sector (25 percent). The service sector has been growing at a much lower rate (about 5 percent) with very low growth rate inthe private service sector and insmall commerce. Among the service sectors, the ones that grew the most were commerce (12 percent) and tourism (7 percent). Public administration employment has been growing at about 5 percent per year. The number o f workers in land based activities, with the exception o f fisheries, has been growing at the rate similar to the one o f the population growth (about 3 percent) 3.21 On earnings, the industrial sector fares best. Real income has been on average constant for agricultural workers. For manufacturing, the average growth inreal income has been about 8 percent per year between 1997 and 2001. Finally, earnings in the service sector have grown about 7 percent2' 3.22 Over the 1997-2001 period when EPZ-led growth was resuming, new jobs created by booming manufacturing sector were accessed by poorer people. Poor households benefited from growth by the large wage differential between the salaries inmanufacturing and those of former sectors o f employment. It appears that the best opportunities for a pro-poor growth reside inthe manufacturing sector, particularly intextiles, where employment is growing at a highrate, entry job salaries are higher, and are increasing at a faster pace. Nevertheless, given the fact that the manufacturing industry is almost entirely inurban areas, and given the low migration flow from rural to urban areas, it i s unlikely that growth in the manufacturing sector in urban areas will have large spillovers to rural areas. Nevertheless, there is evidence (Romani, 2003) that during the growth years, 1997-2001, there were changes inthe structure o f rural households' economic activities away from agriculture. The evidence also shows that the households that were able to access manufacturing activities were also the ones with more schooling. Given that most o f the changes in rural areas described above occurred in the Antananarivo province it will be important to know if this experience can be replicated in other provinces. Although there i s limited information on labor inrural areas, the large differences ineducational attainment across provinces and across urban and rural areas, as will be explained later, highlights the challenges faced inreplicating this experience. Simulation-BasedEstimatesof GrowthonLaborDemandand Poverty 3.23 This section simulates the impact of economic growth on labor demand. The methodology, explained in more detail in annex 7, assumes that the new jobs in the expanding sectors are filled by the individuals that fit best the characteristics o f the workers already 2o Giventhe paucity o f data it was not possible to calculate growthinrealearnings at a moredisaggregated level. -16- employed in those expanding sectors. With this method, it is possible to identify the workers characteristics sought by expanding economic sectors, so to forecast labor demand by skills or by any other relevant characteristics. In the simulation reported in Table 5.4 shows the rates o f employment growth are assumed to be the same as those observed over the period from 1997- 2001, The growth performance used in the simulations is based on the projections from the RMSM-Xmodeling with an overall growth rate o f 5,3 percent o f GDP in2004 and 7 percent o f GDP for the next four years and the corresponding sectoral growth rates. Table 5.4 reports the estimated urban labor demand in the four year period, as well as the overall urban labor force composition by skills, andthe urbanpoor labor force composition by skills. Table 3.4: Urban labor demand simulations Labor Demand Labor Supply Deficit inthe supply of Education level (# workers) (# workers) Labor (# workers) No Education 20,300 51,900 Primary 49,300 66,800 Low Sec. 111,600 107,300 4,300 High Sec. 118,600 61,100 57,500 University 18,200 6,400 11,800 ITotal 318,000 293,500 Staff calculations based on the HH2001. Key assumptions: Labor productivity equal to average labor productivity during the period 1997-2001and sector employment growth as during the period 1997-2001. 3.24 A minimum o f secondary education will be required to enter the driving sectors in the coming years. At the present rate o f productivity, in a four-year scenario economic, growth will create a demand for about 318 OOOjobs. Only a minimal part o f those jobs will be filled by individuals with primary education and university. The bulk o f the labor demand is for individuals with secondary education. 3.25 Demand for skill labor i s likely to exceed the availability o f skilled workers and create a tight labor market for skilled workers, which will possibly results inupwardpressure on skilled wages. This is even more relevant in the case o f tertiary education, as demand for college graduates will likely exceed the supply. On a poverty perspective, the fact that labor demand will likely be filled by individuals with higher level o f education leaves the poor marginalized from the process o f development. The composition o f the labor supply provided by poor people i s overwhelmingly unskilled labor (80 percent). Although, employment growth would likely greatly reduce poverty in urban areas, better pro-poor outcomes could be achieved through complementary policies aimed to increase the skills o f poor people via formal or technical education, so to make poorest more sought by expanding industries. Skilled-Labor Supply Constraints on the Horizon 3.26 Notwithstanding the above simulations, other evidence consistently points towards emerging skilled-labor supply constraints in the medium-term. In urban areas, the number o f fairly skilled underemployed and informal sector workers could potentially offer a reserve labor for future employment growth. However, they will only serve as a reserve o f labor if they have the skills demanded in the fastest growing sectors o f the economy. If they do not, and the education system does not provide these skills, the increase inlabor demand will put pressure on -17- wages, diminishing one o f the main sources o f Malagasy productivity, low labor costs. The data available does not provide information on whether theses workers have the skills demanded by the sectors where employment growth i s expected to be the highest: export processing zones (textiles and apparel, and wood and handicrafts), tourism, the shrimp industry, and mining. Thus, based on these data, it i s not possible to obtain information on whether there are constraints in the supply o f workers needed in these sectors. However, evidence from case studies and data from a urban labor survey, shows that the fastest growing sectors o f the economy are already experiencing major growth constraints due to the lack o f skilled workers. 3.27 A study of the urban labor market in Madagascar (Glick et al., 2003) gives some evidence o f the constraints in the supply o f skilled workers in the Export Processing Zones (EPZs). This study provides evidences that the average year o f schooling o f Antananarivo's EPZ workers decreased from 9.4 in 1995 to 7.8 in 2001, while the self-reported occupational or skill status increased inthe same period. This suggests a possible demand shortage o f the semi-skilled laborers needed for the enterprises working on the EPZs. Indeed, in 1995 not only most o f these enterprises were just starting, but the country was still facing the effects o f the depressed economy that had characterized it inthe early 90s. At that stage, the EPZs were able to select the most skilled workers among the unemployed or underemployed, but once these firms started to grow and needed more workers, the limited number o f semi-skilled workers available forced them to employ workers with less skill and train them. This made the average schooling o f these workers decrease while their occupational or skill status increased. Similarly, anecdotal evidence based on firm interviews from the EPZs (Nasir, 2003) shows that the lack o f trained workers has been a critical challenge for the clothing and textile industry and it is starting to become an issue inthe handicraft industry. -18- Box 3.1: Lack of skilled workers as a limitation to growth intextile, handicraft, and IT firms inMadagascar The results interviews with manufacturing firms inMadagascar showed that many considered the lack o f trained and skilled workers as a limitation for their growth. Many also thought the formal providers o f training were ineffective or too small to offer all the needed training. One o f the main concerns o f the textile and clothing firms interviewed was precisely the lack o f skilledand trained workers: "...Currently most managers, quality control experts, supervisors and technicians are expatriates and they mustdo most worker training inhouse. There are few skilled garment or textile workers inMadagascar andthe sole training institution, FORMACO, is ineffective. I t is too small -one fmsaidthat it only has capacity for 60 workers every three weeks -and it does not teach skills that are appropriate for Madagascar. The shortage of skilled workers will drive up costs and become an ever-bigger problemas the industryexpands. It will also severely limit the industry's ability to moveupthe value chain." A similar but lesser concern was found infirms involved inhandicrafts. Although many o f the workers involved in this sector are not highly skilled, some, especially those involved inembroidery, are. So far the embroidery sector had enjoyed skilled labor trained by religious groups, but inthe near future the lack o f newly trained workers might present a limitation. Most o f the training in this sector is currently done by the firms themselves. At the moment there i s only one institution offering training, this institution is perceived as providing a good but expensive service. Incontrast, another similar training institutioninvolved inthe sewing ofrabanne, is perceived as being completely ineffective. Firms in the Information Technology sector also face labor constraints, as the supply o f relatively high skilled workers in this area i s very limited. This sector is very small in Madagascar, but could offer potential export services (accounting data entry, tele-translation services and call centers) if some infrastructure constraints are lifted. This sector so far has been involved in two different group o f activities. The first group is the design o f web-sites and Internet applications, the second group is data entry and the provisiono f Internet content. The first group needs relatively high skilled workers, which are in quite limited supply at the moment. For instance, in 2001 only 40 computer engineers graduated inthe country. This lack o f skilled workers is putting pressure on wages which might eliminate the main source o f competitiveness o f the country inthis area, low labor costs. Source: John Nasir. 2003 "Manufacturing Case Studies" inJaime de Melo. 2003. Volume I1Diagnostic Trade Integration Study. 3.28 Evidence o f constraints inthe supply o f skilled labor could be also found for other sectors identified as growth poles, such as tourism. A tourism sector study (WB, 2002) identified as one o f the major impediments for growth, deficiencies in the education o f the labor force. Specifically, the study mentioned difficulties in finding workers with basic numeracy and literacy skills to work inthe sector. It also mentioned lack ofproficiency inEnglishas one o f the major challenges faced, as it limits access to European, US, and Japanese tourist supplier markets. EDUCATION POLICYINMADAGASCAR 3.29 The discussion above showed that there is evidence o f constraints in the supply of the skilled labor needed in the sectors that evidence the greatest potential for growth and poverty reduction. This section will assess if the education policy inMadagascar supports the creation of the skills demanded by these sectors. This section will first describe the education sector in Madagascar. It will then assess the government policy inthe sector. -19- 3.30 This section will argue that although the government education policy of attaining universal primary education for all i s a laudable and needed policy, it will not be enough to decrease poverty and to prepare the skilled labor needed in the sectors with highest potential growth: EPZs, tourism, the shrimp industry, and mining.To better understand the challenges that authorities face to provide the needed skills to the entire population, it is first necessary to examine the issues affecting the education sector inthe country. 3.31 Malagasy children have had only limited access to education, very few ever enroll in school, and even fewer finish the first five years o f primary school. The average years o f schooling o f adults in Madagascar, as in other low income countries, is less than the five years needed to complete primary education (Table 3.5). In contrast, in Mauritius, a country that followed a successful growth strategy based on EPZs, an example that Madagascar wants to emulate, adults' average years o f schooling i s about 9. Table 3.5: Average years of schoolingof adults across place of residence, income, and gender inMadagascar Madagascar Low income countries Mauritius Rural 3.6 Urban 6.8 Poor 3.1 Non-poor 6.7 Male 4.8 Female 4.0 All A A .., I A A* 4 I. Note1: Adults here is defined as all people older than 15 years old. Note2: Poor here i s definedas a person living in households were the per capita expenditure is under the poverty line of 988.600 FMGper personper year. 3.32 Not only is the average schooling in Madagascar low, but it also varies largely across income levels and between rural and urban areas. In urban areas, on average, adults complete about three more years o f schooling than adults living inrural areas. Similarly, non-poor adults finish almost four more years o f schooling than their poor counterparts. There is also a small gender difference in adult years o f schooling. This difference however has disappeared in younger generations as both young men and women have similar access to education services in the country. 3.33 The low educational attainment i s in part due to low enrollment rates o f children at any level o f schooling. For instance, in the year 2001, net enrollment rates in primary school were still relatively low (62 percent). The enrollment rate o f children living in households belonging to the 20 percent richest end o f the income distribution is twice as high as the enrollment rate of children in the poorest 20 percent o f households. Similarly, enrollment rates in urban areas are higher than those inrural areas. Incontrast to other low income countries, there is no difference in enrollment across gender; both boys and girls have similar chances of being enrolled in school. Due to high repetition and drop out rates, the primary completion rate in the country is -20- very low. Only 37 percent21o f a cohort that start the first year o f primary school completes the entire cycle. Level Quintile MadagascarTotal - I I1 I11 IV v Mauritius Pre-primary 4 4 12 19 50 14 Primary 45 56 62 76 85 62 95 Lower Secondary 1 3 8 15 44 12 Upper Secondary 1 0 2 3 1 4 4 64 Madagascar's Education Policy 3.34 In light of these facts, the main objective of the Malagasy government policy, as expressed inits "Strategic Plan for the Reform and Development o f the Education Sector" and in its "Education For All (EFA) Plan", i s to ensure that for the year 2015 all children o f school age will be enrolled andwillbe able to complete the five years o fprimary school. 3.35 Many measures have already been taken to address the low coverage of primary education and the large inequalities in access. Some o f these measures have already started to show positive results. For instance, to diminish the negative effects o f the political crisis o f 2002, the government decided to eliminate school fees inprimary school, including private schools in rural areas. This policy, as well as efforts to distribute textbooks and school kits to students and teachers22, resulted in a dramatic increase in enrollments. In the school year 2002-2003 the number of new entrants in primary school increased by 35 percent, the following year the increase was 14 percent.23 School enrollment grew much faster than the education authorities expected and many o f the planned policies will have to be modified to deal with the new situation. When designing the EFA plan, the authorities assumed that by the year 2010 the total number o f students inprimary school will be close to 2.6 million, but already inthe school year 2003-2004 the total number o f students enrolled was close to 3.4 million24. 3.36 Although there are still children out of school, the challenge at the moment is not so much to increase enrollments, but to make sure that those children that have enrolled stay in school untilthey graduate from primary. Many measures will be needed to attain this goal2' two of the most important ones are the following: (i) the unequal distribution o f teachers improve across schools, and lower the student-teacher ratio; and (ii) reduce the high repetition and drop out rates. The first measure will require an effort to train and hire more teachers, and continue the salary bonuses to teachers in remote rural areas to increase the incentives to stay. Another measure to assure the presence of teachers inthese remote areas would be to give support to the teachers already being hired by parental associations. There is a need to evaluate the 21Source: Ministry of Education 2004. 22Effortthat receivedthe financial support ofthe World Bankproject CRESED 11. 23Source : Ministryo f Education2004. 24EFAPlanandMINESEB statistical yearbooks 200112002 and 200212003. 25A detail analysis of possible bottlenecks inthe system and measures to diminish them will be analyze inthe up- coming WB Public Expenditure Review 2004. -21- performance o f these teachers, but as they are originally from the areas where the schools are located they are more likely to remain inthejob. Measures to Improve the Efficiency of the Educational System Secondary GeneralEducation 3.37 In order to provide the skills needed in the sectors with the highest potential growth, (EPZs, tourism, mining, and the shrimp industry), the education sector will have to ensure access to education above the primary level. As shown inTable 5.2, with the exception o f fishing26, the average years o f schooling o f the workers involved in all these sectors is at least 8 years. This average is much higher than what i s needed to graduate from primary school (5 years), and close to the 9 years needed to graduate from lower secondary school or collbge. 3.38 To sustain growth and achieve progressive redistribution, the education policy would need to assure access not only to primary education but also to lower and upper secondary education. Therefore, increasing enrollments and completion rates in"collbge" with the proposed nine years of basic education, is keg7. Limited evidence from Antananarivo (Glick et al. 2003) suggests that the private retums to primary education for both men and women are low and that they also increase sharply with the level o f schooling. Inthe case o f men, an additional year o f post-secondary school raises eamings by 15 percent, an additional year o f secondary school (including collbge) raises earnings by 10 percent, and an additional year of primaryschool raises earnings by only 4 percent. 3.39 The government education plan anticipates that by easing access to primary school and thus increasing enrollments at that level will also increase enrollments in lower secondary school. However, there are reasons to believe that the increase in enrollments in primary education alone will not produce an equal increase in enrollments in lower secondary. Many o f the bottlenecks might still remain in the transition between primary and lower secondary. For instance, in the year after the primary school fees were waived, the number o f primary school students markedly increased, while the number o f collbge students grew less than in previous years (See Table 5.7) Table 3.7: Percentage increase inthe number of students enrolled by level of education Year Primary 1st cycle 2nd cycle As shown inFigure 5.1 (see below), less than 2 percent of young Malagasy inpoor rural areas finish 9 years o f schooling. Incontrast almost 36 percent of the non-poor urban youth do. Inthe 26The data do not allow to differentiate between the shrimp industryand other type o f fishing activities. 27The government's education plan envisions the creation o f a compulsory "Basic Education", composed o f the five years o f primary education andthe four years o f lower secondary. -22- year 2001, only 10 percent o f 15 to 19 year olds had completed 9 years o f schooling2*. At this level of schooling there are also large income and regional variations. Figure 3.1: Educationalattainment profiles across income and place of residence Educational Attainment Profiles o f 15 and Educational Attainment Profile for 20-29 19 year olds in 2001 years old, 2001 1 1 2 0 , 120 100 100 80 80 60 60 40 40 20 20 0 0 0 1 2 3 4 5 6 7 8 9 1 0 1 1 1 2 +poor urban +non-poor urban +poor-urban *non-poor urban poor rural non-poor rural Staff estimates based on the HH2001. 3.40 Finally, incontrast to the many efforts inplace to ensure that all have access to five years o f primary education, and to some increase enrollments in college, not much has been done to reform upper secondary education inMadagascar. For instance, in a cohort o f Malagasies 20-29 years old, old enough to have finished secondary education, only 3 percent had. Again there are also large income and regional differences. Virtually none o f the Malagasies inthis cohort living inpoor households finished upper secondary school (less than2 percent inurbanand rural areas attained at least that level). The situation i s different among the non-poor, in urban areas more than 11percent o f this cohort finished at least 12 years o f schooling, which includes the three years o fupper secondary; inrural areas only 5 percent did. 3.41 Much effort would be needed if all Malagasy in the corresponding school age were to complete nine years o f basic education by the year 2015 as announced in the government's policy. The data needed for a complete study o f the determinants o f demand for secondary schooling are beyond the scope o f this study. A more in-depth study to examine the effect o f not only individual, and household variables affecting the demand at this level, but also the effect o f school fees, distance to school, and quality o f education is needed to determine the best policy options to increase enrollments inpost-primary levels. Technical and Vocational Secondary Education 3.42 International experience2' shows that basic general education offers the essential skills needed to perform in the labor market. These skills are mostly literacy, numeracy, and learning skills, skills that will make training and communications on the job easier. These skills are offered both inprimary and general secondary school and it is here where government priorities should be set. Vocational training and technical education if often seen as a tool to offer the 28 Some o f them might still be inschool, but even if we look at a slightly older generation, only 12percent o f the 20 to 29 years old have. 29Johanson, R. andArvil van Adams. 2004. Vocational Skills Development inSub-Saharan Africa. -23- population specific skills, beyondthe basic ones offered ingeneral education, needed to enter the labor market. In Madagascar the government has traditionally offered this type o f vocational training and technical education, but so far these public institutions has not being successful in providing the skills demanded bythe private sector ina cost-effective manner. 3.43 There is a needto re-think the government's policy with respect to vocational education and training. This sector3' is characterized by an increasing and unregulated private provision for which very little information is available3'. The public sector, representing more than 40 percent of the total number o f students enrolled have the following inefficiencies: (i) the public provision o f this type o f education i s highly fragmented, concentrated invery few provinces, and offering very few options, most o f them only offer two or three subjects; (ii) there i s a low degree of specialization with nearby schools offering the same options; (iii) studendteacher and studendnon-teaching staff ratios are very low, as consequence, the unit costs o f these schools are high, almost twice as large as similar schools in other developing countries. Administrative data indicates that inthe school year 2002-2003, the studendteacher ratio in public CFPs, Centres de Formation Professionnelle, was about 8.6 while in LTPs, , Lyce`es TechniqueProfessionnelles, was about 17.6 (MINESEB 2004) ; (iv) very few students graduate from these institutions (only 40percent of those enrolled in the CFP, and 30percent of those enrolled inLTP; (v) there i s very little articulation between LTPs and tertiary education, graduates from LTP that go to universityusually do so ina subject different from their LTP diploma. 3.44 The private sector enrolls 60 percent o fthe students invocational andtechnical education programs32. Many o f these institutions are unregulated and the public needs to be assured that they are offering a quality service. 3.45 To determine what skills and quantity are needed, it is necessary to carry-out an in-depth study o f these new job-profiles the private sector is in need of, which will require a better connection to the private sector. This i s a priority in the short term as knowledge on the skills needed i s necessary to reform this type o f education. Similarly, a study comparing the private provision o f vocational and technical education with the public one, is required to determine the most efficient manner o f government intervention at this level: (i)should the government continue to directly provide this service; (ii) should it just finance it through contracts with private providers; (iii) it just control it and subsidize needy students to assure equity in should access to these service; or (iv) should the government create incentives for the firms to offer training to its workers. Currently, it is difficult to answer these questions. Evidences suggest that the public sector i s inefficient and poorly linked to tertiary education and the labor market. Whether the private sector is faring any better is not known. 30For the characteristics o fthe Professionaland Technical schools inMadagascar this sectionborrows heavily from World Bank, 2002, "Education and Training inMadagascar". 3`This year for the first time, the educationauthorities implementeda survey to collect data onthese institutions. The data has recently beenavailable, but it still needs to be analyzed. 32World Bank. 2002. Education and Training inMadagascar. -24- 3.46 The education reform plan already envisions the creation o f contracts with the private sector to ensure quality. This has already started, although to a very limited extent. These contracts could serve as a learning experience to determine if this is the most efficient way for the government to intervene. A review of vocational and technical education and training inthe region (Johanson and van Adams, 2004) concluded that the public sector should "complement" the private sector and not duplicate its efforts. More specifically the role o f the government should be directed towards the following areas: (i) regulating the private sector; (ii)setting standards and assuring quality; (iii) developing curriculums and instructional materials; (iv) providing information on the private sector services and on labor market requirements; (v) financing training only for disadvantaged groups that otherwise would not be able to afford it; (vi) anddirectly providing services only when the private sector cannot or would not offer these services, such as insome geographical areas or inskills that are two costly to provide. 3.47 Finally, there is also a need to rethink the structure o f professional and technical education at the secondary level. At the moment two different types o f schools are functioning inthe country. The first one offers training to students with five years ofprimaryschool or with one year o fjunior secondary; the second one offers training to students that have finished junior secondary. Children graduating from primary school are still too young to decide on a "career- employment path" and probably have not mastered the necessary literacy and numeracy skills needed in the labor market. Now that the government is planning to create nine years o f basic general education, it is also time to re-think whether the first type o f training should still function as it does today. Tertiary Education 3.48 Although the priority at the moment i s to increase access to general secondary school and thus assure that a large proportion o f the population masters the basic skills needed to perform in the labor market, higher education should not be disregarded. For instance, the creation o fhighly skilled labor can on itself create comparative advantages inthe export o f certain services; such i s the case o f the IT industry. 3.49 The policy towards tertiary education in Madagascar also needs revision. Tertiary education is provided mostly by the public sector, although some private providers have started to offer programs to a very limited number o f students. There are also new institutes offering tertiary "professional" education, and finally there is a public institution offering distance education. The public provision o f tertiary education is, as in the case o f professional and technical education, inefficient. Inmore detail this level o f education is characterized by33: (i) very fragmented provision, most of the students are concentrated inAntananarivo; (ii) low level o f specialization, most of the universities offer the same options; (iii) poor management o f professors teaching hours, the majority of these teaching hours are "complementary hours" which are better paidthan normal hours of teaching; (iv) large number of administrative personnel; 33 This section ontertiary education inMadagascar borrowsheavily fromWorld Bank. 2002. "Education and Training inMadagascar". -25- (v) large expenditures on students' aid, despite the fact that most o f the students in this level belong to households at the upper end o f the income distribution; (vi) large internal inefficiencies, large number o f repeaters and drop-outs; and as consequence, unit cost ofprovision are very high; (vii) there are large arrears inthe system; (viii) the majority o f the professors are close to retirement age and there i s not a policy set to replace them. 3.50 At the moment, the major source o f financing is the public sector, with very little private participation. The financing o f the sector at the moment is not sustainable due to the large arrears in the system. Under these circumstances financing reforms as well as efforts to eliminate the inefficiencies of the sector are needed. As can be observed in Table 3.8 more than 90 percent o f all students enrolled are not poor. The Strategic Plan for the Reform of the Education Sector listed actions intended to solve many o f these issues, but so far these actions have not progressed. Table 3.8: Proportion of students enrolled in each level of education livinginpoor households Primary Coll&ge Secondary Tertiary Total Non-Poor 0.20 0.52 0.80 0.92 0.27 Poor 0.80 0.48 0.20 0.08 0.73 Total 1.oo 1.oo 1.oo 1.oo 1.00 Source: HH2001. 3.51 At the moment there is not enough informationto determine whether or not the education offered by tertiary institutions in Madagascar is attuned with the demands o f the labor market. The fastest growing sectors o f the economy are labor intensive and will be demanding more semi-skilled and skilled laborers in the future. The majority of workers demanded by these sectors, on average, will not be higher skilled workers, H o w many high skill workers will be needed, and what type o f skills will be demanded i s not known. Evidences suggest that there are constraints in the supply of higher skilled workers (as mentioned in Box 3.1). Anecdotal evidence from the textile industry shows that most o f the managerial, quality control experts, supervisors, and technicians working on them are expatriates. Similarly, evidence from the IT sector shows that there i s also a limited supply o f computer engineers and technicians. This limited supply i s puttingpressure on wages and is therefore diminishing the competitiveness o f the country. Although the IT sector i s still very small, and faces many infrastructure constraints, ifdeveloped,itcouldbeapotentialsourceofserviceexports(accountingdataentry,callcenters, etc.). Any policy trying to improve the links between higher education and the labor market will need to be based on more detailed information that could be offered by an enterprise survey that collects data on the skill profile o f high skill workers, and on the skill profile o f the new positions offered. CONCLUSIONS 3.52 High underemployment rate and a vast informal sector suggest a strong potential for Madagascar's economy to increase unskilled employment without puttinga large strain on labor markets. As economic growth increases employment, women's participation and wages, it also affects poverty by making available better forms o f employment, by increasing migration o f -26- workers from the informal to the formal labor sector and by increasing labor productivity, Policymakers should facilitate growth in sectors which pay higher wages and which are most likely to have larger impacts in terms o f poverty reduction. At present productivity levels, a pattem o f growth favoring the industry and services sector will likely produce the most poverty reduction inurban areas; inparticular, inthe manufacturing sector where employment is growing at a very high rate, entry salaries are higher and are increasing at a faster pace. Among manufacturing, the textile and apparel sector probably represents the most promising sector in term o fpoverty reduction. 3.53 However, the supply o f skilled workers has already reached a critical point. It is therefore important for the education sector to provide workers with the necessary skills to ensure the supply o f qualified labor to the growing industrial and service sectors. Equally important is the emphasize on the participation o f the poor in the benefits o f economic growth and whether the poor have the capacity, the necessary skills and access to assets, resources and services enabling them to benefit from growth inemployment. Without appropriate educationpolicies, the increase inlabor demand will not be able to meet the required skilled workers, which mightresult inan upward pressure on wages and increases inthe wage gap between skilled and unskilledworkers. Furthermore, low internal migration and rapid employment growth in labor intensive industries will likely create further pressure on labor (especially skilled). Inparticular, some o f the fastest growing sectors (textiles and apparel, and tourism) employ a type o f labor force that i s both young and skilled. In this regard, education policies, both with respect to formal and technical education are to play an fundamental role to achieve pro-poor sustained economic growth. 3.54 The education policy in Madagascar with its priority o f attaining universal primary education in the year 2015 i s a laudable policy, but it will not be enough to provide the skills demanded by the fastest growing sectors o f the economy. Five years o f education do not provide enough o f the basic skills needed to perform in the labor market: literacy, numeracy, and communication skills. Government priorities in the short term should be set on assuring the provision o f these skills. Inthe medium term facilitating the provision and acquisition o f other more specific skills will also be needed. Finally, given the segmentation o f urban and rural labor markets and low rate o f intemal migration it i s unlikely that employment growth inurban areas will have a substantial impact on wages inrural areas and poverty in general. One o f the effects o f economic growth will then be likely to increase levels o f inequality between urban and rural areas. Increasing agricultural productivity and enhancing the linkages between urban and rural markets remains a prerequisiteto reduce poverty as more than 80 percent o f the poor are engaged inagriculturalactivities. METHODOLOGY FORESTIMATINGTHE EFFECTS GROWTH EMPLOYMENT OF ON 3.55 The main channel through which economic growth affects social welfare and poverty is by creating employment, raising real wages, and increasing participation rates. However, employment growth per se i s not sufficient to guarantee poverty reduction. For example, if employment growth i s limited to low paying sectors, or if the poor do not own skills sought by expanding sectors, economic growth will likely do little to reduce overall poverty. 3.56 To simulate the extent to which economic growth can help Madagascar rapidly reduce poverty, this report focuses on the effect that economic growth has on employment. The data -27- utilized in this exercise i s from the 2001 household survey (Enqziete Prioritaire Aupr2.s des Me'nages 2001). 3.57 Inthe simplest case, the functioning of labor markets can be summarized through two extremes (see box 7.1). Inone extreme, all growth is bottled up inthe sector, so that households inthe segment reap the benefitsvia higherwages. Inthis scenario, labor markets function poorly as they are assumed to be completely segmented. Inthe other extreme scenario, all growth comes from hiring labor from a 'unlimited pool o f workers' available from marginal activities at a subsistence wage. In this scenario, there i s a perfectly elastic supply o f labor at a fixed wage, reflecting a unified labor market, as there are no obstacles to finding a job in the expanding sectors. The functioning o f the labor market i s somewhere between these two extremes, with each segment o f the labor market according to skills (engineers, technicians, laborers, women seamstresses, etc.) beingcloser to one extreme or the other. 3.58 In developing countries, labor markets are usually segmented in two broad categories: skilled and unskilled. Inthe case o f unskilled labor, the presence o f a large reserve labor sector and widespread underemployment guarantee a large supply o f labor at the prevailing wage. On the contrary, skilled labor is usually more scarce, and increase inthe demand for skilled laborers will likely result inupwardpressure on skilledwages. 3.59 Given data availability, the labor market i s divided into two broad categories (skilled and unskilled workers) which allows to assume (rather than produce a precisely estimate) predeterminedwage elasticities with respect to economic growth. Inthe simulation exercise it is assumed that economic growth does not put upwardpressure on unskilled wages (because a quite elastic labor supply), while it does affect skilled wages (because shortage o f skilled workers). For simplicity, wages for skilled workers are assumed to increase at the same pace that was observed between 1997 and 2001. 3.60 Given those assumptions, the study concentrates on the latter part o f the exercise (described below) which draws on the wealth o f information just available from the extensive 2001 survey. 3.61 Insummary, the simulations involve two steps. First the individuals more likely to fit the skills sought by the expanding industry are moved from the informal to the formal labor sector. Then, for those individuals, the wage increases according to the wage differential between the former employment and the new sector o f e m p l ~ y m e n t .For ~ ~ this exercise, labor market i s differentiated into two main sectors: a formal sector (containing individuals employed in commercial agriculture, manufacturing and formal services sectors) and a reserve labor sector. The reserve labor sector consists o f unemployed, underemployed and low-income individuals employed in marginal activities such as small commerce, temporary and occasional jobs, and low-income primary sectors. The formal sector is further subdivided into several sub-sectors: commercial farming, fisheries, mining, textiles, handicrafts, other industries, tourism, public administration and other services. In the simulations workers are allowed to move from the 34Nicita and Razzaz (2003) offer an application o f the methodology to estimate the effect on welfare o f the growth inthetextile andapparel industry. -28- reserve labor sector to any o f the industrial and business sectors, but they are not allowed to move within sub- sector^.^^. 3.62 To estimate which people will actually move, the study uses a matching methodology (Heckman et al., 1997). Inbrief, to single out the individuals that are more likely to move, the study estimates their propensity scores, that is, the predicted probability that each individual has o f working ina given sector based on his observed characteristics. 3.63 The probability o f finding employment i s sector A can be estimated with a simple multinomial logit model andtakes the form: 3.64 Inthismodelthe log-likelihood function to bemaximizedis: where y, equals 1 if sectorj i s chosen and 0 for all the other alternative sectors and xi is a vector of demographic characteristics. Estimation results o f the multinomial logit model are given in Table 7.1. 3.65 The probability o f moving into any o f the A sectors is a hnction o f a series o f demographic characteristics. These include gender, age, level o f education, technical education, urbdrural dummy, and regional location. The individuals are then ranked according to the estimated probabilities and, for those with highest rank; the sector o f employment is changed to the one with highest probability. Having estimated which individuals are more likely to fill the jobs created by the expanding economy, the impact that the new employment has on the income o f each worker needs to be estimated. To do so, the wage differential between the former and the new sector o f employment needsto be estimated. 3.66 The wage differentials between the reserve labor sector and the sub-sectors are estimated using the household data following the wage premium literature (Kruger and Summers, 1988). The estimation regresses the log of worker i's wages ( InW, ) on a vector ofworker i's characteristics ( x, and a set of industryindicators( ) and takes the Mincerianform: I.. 35 The assumption is that individuals develop sector-specific skills in their formal sectors of employment, making adjustment and switching more costly for them. -29- 3.67 The Mincerian wage equation i s a popular model showing how an individual's characteristics affect his or her wage. The specification follows Krueger and Lindahl, 2001. Results o f the estimation are presented intable 7.2. Workers' characteristics include gender, age, education, marital status, urbdrural and regional dummies. The coefficient on the industry dummy captures the wage premium paid to each industry in comparison to the informal sector (which is omitted).36The results from the econometric estimates are used in the simulation exercise. Having identified the individuals that are more likely to move into the any o f the expanding sector o f the economy and having estimated their specific wage differential, the impact on household income and social welfare and onpoverty ingeneral is e~timated.~' 3.68 Inthe simplest model, the indirectutilityfunction ofthe households canbewritten as: household utility uh i s expressed as a function o f a vector o fprices P faced by the household and the household's incomey, ,which inturn is a function o f the amount o f labor the household sells on the market lhat the prevailing wage w. Prices P are assumed to be fixed, while income varies with the change insector o f employment andthe increase inrealwages. 3.69 Further, assuming that households choose optimally the amount of labor to sell in the labor market, the effect o f prices and wages on profits can be obtained by differentiating (4) and dividingby income o fhouseholdh to obtain the percentage change inwelfare: where 0: = we / y h i s the share o f income obtained inthe labor market by household h. Finally, the new income dwho f the household i s calculated as follows: dWh = Wi [(1+ZWiA)(l+skwi)`]R, (7) i where wi is the old wage o f individual i in households h, zwiA is the wage differential for individual i for sector A estimated by equation (3), skw, is the (calculated) increase in skilled wages3*, and RiAi s a dichotomous variable taking the value one for the individuals which best match the characteristics o f those employed in sector A estimated according to equation (2), and zero otherwise. Finally, the change in welfare is distributed across households members, expenditures are equated to income, and new poverty levels are calculated at the new level o f income. 36Inother words the coefficient ofthe industrydummyrepresents the partofthe variation inwages that cannot be explainedby worker characteristics, but canbe explained by the workers' industryaffiliation. 37Household income is simply the sumof the income o f each o f its members. 38This termtakes the value 0 ifthe individual is unskilled. -30- LABOR SUPPLY AND LABOR DEMAND 3.70 Inthe simulations, total labor supply is approximatedby the sum of individuals that are unem loyed or underemployed3' augmented with the number o f graduated forecasted for each year4! Labor demand is estimated in equation (2) and consists in the sum o f news jobs (assuming fix rate o f productivity) demanded from each o f the expanding economic sectors. In this setup, labor shortage by level o f education is calculated as the difference between the individuals chosen to fill the new jobs fiom the pool o f total labor force (labor demand) and those chosen among those individuals inthe available labor force (labor supply). For example, if labor demand required 80,000 individuals with secondary education, but the pool o f labor supply i s able to provide only 50,000 individuals that have secondary education and have matching characteristics (as estimated in equation 2) to fill the new jobs, then one can argue that the expansion in employment is lacking o f about 30,000 individuals with secondary education. Those jobs will be taken by individuals with lower education (and lower productivity) therefore slowing economic growth. 39 Underemployment here is defined as working either less than 7 hours a day, or less than 5 days a week, or less than 20 days per month. 40 For example, the forecasted labor supply with secondary educationfor the second year includes the number o f individuals expected to attain secondary educationinthe second year. -31- Box 3.2: Growth, Poverty and the Labor Market: Two Extremes The link between international trade and poverty operates via the labor market. The figure below considers two extreme assumptions: a perfectly inelastic and a perfectly elastic labor supply. Inthe caseofaninelastic labor supply,whenthe demandfor labor shifts out fromDoto D1, employment cannot increase and the market must be brought back to equilibriumby an increase inwages from woto wl. Ifsome o f the workers inthis market were poor -or belongedto poor families- the resulting increase inwages would have a direct andbeneficial impact onpoverty. Inthe other extreme case (a perfectly elastic labor supply), anincreaseinlabor demandresults inan increase in employment to L1 with no changeinwages. The effect onpoverty dependsheavily onwhat the additional workers were doing before taking these newjobs. Ifthey were poorly employed or engaged insubsistence activities and earning a wage lower than Wo then the impact on welfare will depend on the wage differential betweenthe old and new employment. Real wage _'I WO Lo -L, employment -32- Table 3.9. Multinomial Iogit estimation (dependent variable: Iogit sector) Variable Sector Coeff. S.E. Sector Coeff. S.E. Sector Coeff. S.E. Land Fisheries Mining Age 0.011*** (0.002) 0.020** (0.010) 0.012 (0.020) Education -0.040*** (0.008) 0.085*** (0.032) 0.210** (0.083) SkilledAJnskilled -0.448*** (0.054) -0.219 (0.274) -0.444 (0.708) Region 1 0.147 (0.096) -1.161*** (0.444) 20.605*** (1.045) Region 2 0.017 (0.097) -2.292*** (0.778) 19.577*** (1.171) Region 3 0.011 (0.100) -1.073** (0.521) 20.573*** (1.026) Region 4 -0.390*** (0.109) 0.265 (0.394) -11.107 Region 5 -0.015 (0.105) 0.120 (0.418) -10.807 Urban Rural -0.852*** (0.057) 1.265*** (0.335) 0.139 (0.552) Gender 0.693*** (0.052) 1.566*** (0.299) 1.147** (0.537) Technical Education 1.066*** (0.323) -31.303 -31.783 Constant -1.039*** (0.115) -6.929*** (0.610) -28.253** Other Manufacturing Textiles Services Age 0.029*** (0.004) -0.004 (0.005) 0.024*** (0.002) Education 0.123*** (0.014) 0.099*** (0.016) 0.139*** (0.008) SkilledAJnskilled 0.482*** (0.132) 0.940*** (0.155) 0.206* ** (0.071) Region 1 0.310** (0,174) 1.601*** (0.252) 0.343*** (0.105) Region 2 -0.783*** (0.226) -2.073*** (0.557) -0.5 lo*** (0.125) Region 3 -0.825 (0.232) -1.107*** (0.400) -0.5 75*** (0.130) Region 4 -0.600** (0.228) -1.181*** (0.430) -0.597*** (0.135) Region 5 -0.820*** (0.252) -2.048*** (0.628) -0.290** (0.132) UrbanRural 1.424*** (0.161) 0.815*** (0.169) 1.058*** (0.079) Gender 1.505*** (0,120) -0.578*** (0.132) 0.577*** (0.061) Technical Education -1.723 (1.056) -30.774 -1.997*** (0.772) Constant -6.51 1*** (0.283) -4.978*** (0.335) -4.159*** (0.150) PublicAdministration Wood Industry Tourism Age 0.083*** (0.005) 0.028*** (0.008) 0.008 (0.007) Education 0.288*** (0.016) 0.064** (0.026) 0.135*** (0.023) SkilledAJnskilled 1.047*** (0.202) 0.217 (0.239) 0.312 (0.204) Region 1 0.229 (0.214) -0.224 (0.304) 0.532* (0.289) Region 2 -0.162 (0.247) -1.036** (0.400) -0.742* (0.384) Region 3 -0.42 (0.264) -1.741*** (0.519) -0.724* (0.392) Region4 -0.023 (0.256) -0.670* (0.383) -0.857** (0.427) Region 5 0.405 (0.249) -1.338*** (0.488) -0.049 (0.362) Urban Rural 1.009*** (0.169) 1.294*** (0.293) 0.466** (0.207) Gender 0.561 *** (0.117) 3.089*** (0.424) -0.261 (0.177) Technical Education -1.66 1** (0.842) -32.003 -31.535 Constant -10.130*** (0.374) -8.196*** (0.621) -5.224*** (0.400) Observations 11513 Pseudo&? 0.167 Note: Standard errors are shown inbrackets. Significance level o f 1%, 5% and 10% are indicated by***, ** and * respectively. -33- Table 3.10: Wage differential regression (dependent variable: log earnings) Variables Coeff. S.E. Age -0.007* (0.005) Age squared o.ooo** (0.000) Education 0.088*** (0.010) Education Squared -0.002*** (0,001) Land 0.104*** (0.022) Manufacturing 0.670*** (0.046) Mining 0.443*** (0.076) Textile 0.629*** (0.055) Fisheries 0.281** (0.110) Services 0.708*** (0.029) Tourism 0.741*** (0.086) Pub Adm. 0.710*** (0.047) Wood 0.524*** (0.107) region 1 0.056** (0.028) region2 -0.277*** (0.030) region3 -0.290*** (0.030) region4 -0.088*** (0.033) region5 -0.086*** (0.036) Marital Status -0.051*** (0.021) UrbadRural 0.134*** (0.016) Constant 14.686*** (0.083) Observations 9009 R-Squared 0.403 Note: Standard errors are shown inbrackets. Significance level o f 1%, 5% and 10% are indicated by***, ** and* respectively. -34- 4. MINING 4.1 This section on miningdescribes the existingpotential inminingand stressesthe need for institutional reforms to consolidate Madagascar' position o f leader in the gemstones sector. It revisits the recently set up legal framework and highlightsthe objectives o f the planned reforms to regulate activities inmining. Madagascar's endowmentand Potential 4.2 Madagascar is known for the variety o f its mineral resources. Main minerals produced are chromite, graphite, and mica, all o f which are exported, as well as gemstones such as the rubies, sapphires, and emeralds (see Table 4.1). Table 4.1: Madagascar's mineral production 1998 1999 2001 2002 2003 Chromite, Mica, 12,350,529 11,104,068 14,759,192 8,360,076 7,821,590 Graphite Precious Stones 2,631,325 12,202,836 13,828,591 5,943,702 7,628,621 Semi-precious 566,566 798,868 1,496,606 808,779 1,40 1,011 Stones OrnamentalStones 3,258,226 3,163,841 6,594,543 4,062,442 9,475,712 Metallic Ores 118,999 88,485 2,622,349 106,953 Gold 82,142 71,396 162,128 86,900 81,802 TOTAL 19,007,787 27,429,494 39,463,409 19,368,852 26,408,736 -35- 4.4 Other resources potential. Three projects are moving into the feasibility stage with planned investment o fmore than US$ 1billion : Ilmenite. QIT MADAGASCAR MINERALS S.A. (QMM) i s a joint venture company held at 80 percent by QIT-FER, a subsidiary of RIO TINT0 (Great Britain and Australia) and at 20 percent by the Malagasy State, aiming to develop ilmenite deposits in Fort- Dauphin, Toliary. The deposit contains 67 million tons o f high quality ilmenite, a world class deposit. Total investment i s put at US$ 350 million, for a total annual production of 750,000 tons at full capacity. The investment decision i s expected in2005. Madagascar Resourcesis a Malagasy company where Group Boule i s the shareholder of reference. The company has entered an agreement with Ticor for the development of ilmenite deposits close to Tulear. Nickel. DYNATEC / PHELPS DODGE i s ajoint venture company heldby Dynatec at 53 percent and Phelps Dodge at 47 percent, for the mining o f nickel in the region of Moramanga. The deposit contains 210 million tons of minerals with nickel content of 1.10 percent and cobalt of 0.10 percent. Expected production is 5,000 tons o f nickel and 4,000 tons of cobalt per year for more than 20 years. US$ 20 millions are being invested inthe preparation of a feasibility study, expected to be concluded by June 2004. Total investment i s put at more than US$ 800 million, including a metallurgical plant near Toamasina for the processing o f the ore. 4.5 Gemstones. There i s a general consensus that Madagascar's rich endowment o f mineral resources coupled with its relatively cheap labor costs provide genuine opportunities for adding value to mineral resources and further increase mineral exports. Madagascar's success incertain gemstones, such as sapphire and rubies, where it is consolidating a position o f world leadership, are clear examples o f this potential. T h e recent reforms 4.6 The Government is convinced that its central objective o f reducing poverty through accelerated growth can only be accomplished through institutional and administrative reforms that refocus the role o f the State, remove inefficient and discretionary regulations, and build a strong partnership between the State, the private sector, and the civil society at large. The new minerals policy shifted the role o f the State from producer to promoter and regulator o f the sector. TheLegal Frameworkfor mining in Madagascar 6.10 Inthe late 199Os, Madagascar undertook major institutional reforms inthe miningsector. The reforms carried out included : (i)a new mining code and its regulations, that have established an adequate legal and regulatory framework to attract private investment into mining, including joint environmental regulations for mining, published jointly by the Ministry o f Environment and the Ministry o f Energy and Mines; (ii)a special law for large-scale mining investments, defining an attractive special investment regime for FDI in mining in Madagascar, and providing for a fair share o f revenues between the Government and the private sector, an adequate cut for the Provincial Governments; and (iii)improved governance through the -36- establishment o f a non discretionary and transparent system to grant, manage and cancel mining permits, the MiningCadastre. The MiningCadastre Registry was created inMay 2000 to enforce a simple, transparent and accessible management o f mining licenses. Based on the principle "first come, first served", the B C M M grants mining licenses based on simple conditions, the payment o f a fixed mining administration fee based on a standard surface units o f 6.25 square kilometers, and subject to submit an environmental assessment or an environmental commitment plan. (i) The MiningCode. In 1999, the Malagasy Government has approved a new MiningCode (Law n099- 022 of August 30th, 1999), with a view to simplify the mining system and make it more transparent, as well as to eradicate conflicts and improve the management o f mininglicenses. The new code puts all investors on the same base, irrespective of their origin or their capital ownership. It takes into account the new constitutional provisions with regards to the decentralization of administrative services, and is in conformity with the concern to preserve the environment and conduct the activity in a better socio- economical climate. The Mining Cadastre establishes and maintains the updated public registry of mining leases accessible for consultation by the public, an innovation in the Malagasy legal framework. The fundamental principle for the granting i s based on the "first come, first served" principle. Discretionary procedures and discrimination have been abolished in the granting o f mining leases. Reasonable and progressive fees are established to discourage speculation. All mininglicenses provide exclusive rights for all commodities inside the mining lease area, with guaranteed security of tenure duringthe transition from exploration to mining. The free commercialization of the products is guaranteed, as well as the reduction of custom duties for imported equipment and goods for the exploration and mining.An amelioration inthe "liquidity" of mining investments has been set up by the liberalization of mining right transfers through leasing, mortgage and other transactions. (ii) The Law on Large Scale Mining Investments (LGIM). Inan economic study realized in 1999, Professor James from Colorado School o f Mines (Denver, USA)had compared the Internal ReturnRate (TRI) o f a copper mine type project andthe effective tax rate o f such project in24 countries having an interesting geology. Table 4.2: Fiscal Regime under the Law on Large Scale Mining Investments tax return rate onhistoric cash-flows of more than 20 percent, the IBS is applied at 35 -37- Professional Tax consideration after the start o fthe commercial production Yearl: 11150; year 2: 11120; year 3: 1/90; year 4: 1160; year 5 and so on: 1/30 EIS contribution FMG410 million+(0,l percent* investment); deductible IFT lpercent o fthe sale value ofthe land IFPB lpercent o f rentalvalue (cumulated constructioncosts) upto FMG 1.000.000.000.5 year holiday for new construction TAFPB lpercent o f rentalvalue (cumulated constructioncosts) up to FMG 1.000.000.000.5 year holiday for new construction Loss carry forward 5 year limit 50 percent of the tax owned on the investment at the IBS rate o f the year when the Investment credit investment has been made. The investments providing by this credit are listedin article 01.01.07 o fthe GeneralTax Code (CGI). -exploration: 33 113 percent straight-line depreciation Treatment o f costs -feasibility Study: 33 113 percent straight-line depreciation -development expenses: 10percent straight-line depreciation -equipment:25 percent straight-line depreciation -loan interest: expensed as incurred 4.7 This survey reveals the lack o f competitiveness o f the Malagasy Common Tax Code as far as mining i s concerned. This situation has urged the Malagasy Government to pass a new law called "the Law on Large Scale Mining Investments". This Law establishes a balance between large mining investors who are subject to international competition and the interests o f the country concerning taxes, economical and social consequences produced through the development o f the mining industry. The LGIM brings great advantages to mining investments, namely stability, better conditions to repatriate export revenues, andinternational arbitration. The stability concerns the legal framework, change andtax system. Thelegalframework for environmental management in mining. 4.8 The biodiversity conservation in Madagascar i s a world priority, considering its variety, its endemic character and its high potential o f degradation. Madagascar i s considered by NGOs such as International Conservation as one o f the three places in the world that grants the most priority for the biodiversity conservation. Four factors contribute for that : 0 The number of endemic plants (9 704) 0 The number of endemic vertebrates (77 1) 0 The number of endemic plants per 100 kmsquare (16,4), The number of endemic vertebrates per 100 kmsquare( 1,3). 4.9 These factors are used to make investments as priority inthe conservation and protection field. Since Madagascar is among the five first zones for each factor, its global classification i s among the first three zones. Madagascar is then situated on the same level as the Philippines and Sundaland (Indonesia Island). This wealth i s then threatened by strong degradation to such an extent that several species risk disappearing definitely, without having beendiscovered. 4.10 The mining sector has taken into account in its management the integration o f the environmental dimension. The harmonization o f the environmental protection constraints with -38- sector operations has been introduced through a detailed regulations based on the environmental framework law and sector specific legislation. The key diplomas o fthe legislations are : Environment Charter no90-033 of December 21, 1990. Public or private investment projects that are liable to affect the environmentshould be the subject of an impact study, considering the technical nature and the extent of the aforementioned projects and the sensibility ofthe establishmentenvironment. Decree on the complianceof Investmentswith Environmental Management (MECIE) No 99 -954 of December 15. 1999. This decree governs the Environmental Impact Study (EIE) assessment procedures, as well as the Environmental Commitment programs (PREE), for exploration activities andsmall-scalemining. The MiningCode (no99 022 of August 1999) andits rewlations (n"2000-170 ofMay 15 2ooo). The preparation of an environmental impact study, and an environmental managementplan, including the preparationfor mine closure andthe rehabilitation of the site are prior conditions before the beginning of all operations. No mining activities can start (and, eventually, also detailed exploration) without prior approval by the relevant environmental authorities, as per of the regulations on environmental protection, of the commitments containedinthe environmentalimpact study. All prospecting, researchand exploitation works are bannedwithin natural reserves andprotectedareas. Mines-Environment ioint Interministerial order. The inter ministerial order no 12032/2000 of November 06, 2000 setting the regulation of the mining sector as far as environment protection i s concerned, defines and specifies central and provincial proceduresandmodes on the P E E file examination. 4.11 The State does not operate anymore directly inthe miningindustry, which means that its role is now to establish an attractive socio-economic environment favorable to the development of the national private sector and to attract foreign investment. The role o f the administration is limited to the management and follow-up o f license holders obligations, the enforcement o f the legal and regulatory framework in mining and environmental matters, and last but not least, to guarantee the security of investments. The current status inthe miningsector 4.12 Weak enforcement o f the mining code, in gemstones in particular, diverts the artisanal mining outside the formal channels, with insufficient linkages to the rest o f the economy, generating a substantial loss of fiscal revenues and causing damage to the environment. The legal and regulatory framework, andparticularly the procedures for the grantingo fminingtitles, needs to be revamped. For instance, rushes in gemstones do not fall systematically under the legislation. As a result, value added from mining escaped the country due to smuggling o f precious stones out of Madagascar. Illegal exports are estimated at US$200 million per year, or 5 percent o f GDP. 4.13 The weakness in the administration is also reflected in the collection o f the mining royalty. The contribution o f Madagascar's mining to Government's revenues i s small even when measured against the modest size o f the sector. The mining royalty due by the mining operators in 2002 totaled US$ 1.9 million. Because of cumbersome procedures and lack of cooperation -39- between the mining and the tax administrations, only about 10 percent o f this amount was effectively collected. The collection o f the key sector specific tax - the mining royalty ("redevance minibre") - is slow, complex and inefficient. Based on a yearly and voluntary declaration by the mining companies to the provincial mining authorities, it opens the way to discretionary behavior by the mining administration. For instance, there is little control on the declared production and no follow-up on those that fail to declare. In the beginning o f 2004 many miningcompanies, and not the smallest ones, were still negotiatingthe amount o froyalties due for 2000. Because o f the failure to enforce efficient taxation mechanisms on mining activities, the government introduced in the 2001 Fiscal Law a special tax ("droit d'accise") o f 75 percent o f the value o f gemstones. The tax, that would represent a strong incitation to the continued smuggling o f gemstones, was never enforced. Government wants now to trade it o f f against a reform o f the tax collection mechanism for the mining royalty, as well as other taxes due by miningcompanies. Annual declarations would be replacedby retention at the source, and no gemstones would be exportedwithout proofo fprevious payment o f all taxes due. 4.14 Furthermore, cumbersome procedures reduce exports and increase the risk o f corruption andsmuggling.The threshold o feligibility to the special investmentregime is too high(US$200 million) and needs to be lowered to reflect investment conditions inmining in Madagascar. The current export licenses mechanisms provide substantial leeway for red tape and corruption with the foreign exchange restrictions (obligation o f repatriation) on exports by Malagasy companies and the discretion o fthe authorities on fees related to gemstone purchases by tourists. The second generationof reformsin mining 4.15 A key element o f the institutional reform o f Madagascar's mining administration are measures planned to improve transparency and governancein the gemstone sector, the most important being the creation o f the Gemological Institute o f Madagascar (IGM - see Box 6.1). The IGM will enable mine operators to improve their knowledge o f precious stones; moreover, the lapidary training by the IGM will reinforce technical and economical capacity of small lapidaries, and will also allow having reference o f certification and expert for gemstones export norms. The IGMi s expected to begin operations inJuly 2004. Box 4.1: The Institutede Gemology de Madagascar (IGM) The Institutede Gemology de Madagascar (IGM) is a project of the Ministryo f Energy and Mines and financed by the World Bank with assistanceofUSAID. The vision ofthe IGMis inthree parts. 1.Instructionin Gemology. With the active participation on one of the world's foremost schools of gemology the IGM will become the first gemological institution in Africa. The goal is to train Malagasy students to an intemational standard. The students will then go on to work inprivate enterprise (stone buying and selling, banking, appraisal, and consulting), education (working as instructors for future generations o f students at the IGM or at satellite offices invarious regional cities o fMadagascar)or for the administration as experts inthe Service de Mines, Guichet Unique,Douane etc. 2. Instruction in Lapidary Arts. National and intemational experts will teach IGM students to cut colored gemstones to intemational norms with the two-fold goal o f raising the quality and the perception o f quality o f Malagasy cutting. This will enable the lapidaries in Madagascar to compete with cutters from any country in the world thereby improving local profits and the value added by the mineral sector. Well trained lapidaries will increase the quantity o f well cut stones available in Madagascar and will make Madagascar a more desirable destination for foreign buyers to visit. -40- 3. Establishment of a Gemological Laboratory. A gemstone laboratory capable of issuing internationally recognized certificates will increase buyer confidence incut stones available inMadagascar. This should lead to an increase in the volume and value of gemstones sold locally and will position Madagascar among the world's gemstone research laboratories. The Gemological Institute of America (GIA) has completed its first extension course in Gemstone Identification taught inFrench inAntananarivo. Dialogue with several international gemological institutions is going on. The need for a decentralized structure 4.16 The immense variation o f Madagascar's geological and environmental conditions implies that over the medium term, Government's strategy for the sector i s based on the recognition that the sustainable development o f mining cannot be achieved without a decentralized administration and a deep involvement o fthe communities. 4.17 The Government plans to create the Agence de Promotion du Secteur Minier (APSM), funded through a contribution of 5 percent of the royalties collected in the sector, to: (i) ensure appropriate enforcement o f the Law on large scale mining investments in the provinces, (ii) decentralize mineral resources management. The activities include the delivery o f extension services to small-scale and artisanal miners to improve their efficiency and help to access to commercialization channels. In concrete terms, this calls for : (i)capacity building o f the Provincial mining administration, with the establishment o f a better alignment between core central public sector functions ; (ii) definition o f effective ways and means for community empowerment and participation, including the establishment o f public/private partnerships with responsible miningcompanies willing to invest part of their profits in human resources capacity building, social, and physical infrastructure and; (iii)increase o f fiscal revenues to the communes, including the decentralizationo f tax collection. 4.18 The APSM's actions would therefore need to be accompanied by decentralization to bringthe institutional support at all levels. A key element o fthis strategy would be the provision of technical assistance to community associations and municipal governments for the integration of mineral resources management in their participatory development plans. This implies that local institutions and structures should be empowered to directly deal with the problems they face and propose solutions to which public policy and investments should respond accordingly. Empowering local communities to local management o f mineral resources and the preparation o f Local Economic Development plans will be an effective tool in the process o f managing economic and social development, and reducing the State's role while creating the conditions for increased private sector participation. The on-going decentralization process provides a unique opportunity to effectively put rural communities in the driver's seat and tailor public policy and investments to local reality, thereby increasing their relevance. 4.19 At the municipal level, the strategic choice made by Government is to increase the tax revenues of the communes through the decentralization of tax collection mechanisms. A major challenge associated with this strategy i s improving good governance and transparency o f revenue expenses by the communities. The mining code defines basic revenue sharing arrangements for the minerals royalty and the surface rental fees collected by the Mining Cadastre, according to the table below. -41- Table 4.3: Sharing of mining revenues Mining Cadastre fees Royalties (2 percent o f first transaction) Center 10percent 10percent Province 30 percent 70 percent Commune [1/3 ofprovincial revenue] [113 ofprovincialrevenue] Cadaster Office 60 percent 15 percent GoldAgency 5 percent 4.20 In 2001, the Mining Cadastre collected the equivalent to about US$ 450,000 in fees, Because o f the lack o f adequate mechanisms, it has not been able to transfer the 30 percent o f revenues due to the Provinces (and communes), and the equivalent o f 40 percent o f the revenues has been transferred to the Central Government. A reform o f the mining fiscal regime (in coordination with the Ministry o f the Economy, Finance and Budget) is expected to substantially increase fiscal revenues from mining in resource rich communes, and to channel resources directly to them. c -42-