Report No. 32869-MX Mexico Income Generation and Social Protection for the Poor (In Four Volumes) Volume II: Urban Poverty in Mexico August, 2005 Colombia and Mexico Country Management Unit Poverty Reduction and Economic Management Unit Latin America and the Caribbean Region Document of the World Bank ReportNo. 32869 -MX Mexico Income Generation and Social Protection for the Poor (Infour Volumes) Volume 11:URBANPOVERTYINMEXICO August, 2005 ColombiaandMexicoCountry ManagementUnit Finance, Private Sector andInfrastructure Unit LatinAmerica andthe CaribbeanRegion Documentofthe WorldBank AbbreviationsandAcronyms BANSEFI Banco delAhorro Nacionaly Servicios Financieros CBO Community Based Organization CIESAS Centro de Investigaciones y Estudios Superiores enAntropologia Social CNBV ComisionNacional de Bancos y Valores CONAFOVI ComisionNacional de Foment0 a la Vivienda CONAPO Consejo Nacional de Poblacion ENEU EncuestaNacional de Empleo Urban0 FDI ForeignDirect Investment FONAEVI FondoNacional de Apoyo Economico a la Vivienda FUPROVI Fundacion Promotora de Vivienda ENIGH EncuestaNacional de Ingresos y Gastos de 10s Hogares GDP Gross Domestic Product GE Generalized Enthropy IMR Infant Mortality Rate IMSS Instituto Mexican0 del Seguro Social INDESOL InstitutoNacional de Desarrollo Social INEGI InstitutoNacional de Estadistica, Geografia e Informatica INFONAVIT Instituto delFondoNacional de Vivienda para 10s Trabajadores ISSSTE Instituto de Seguridady Servicios Sociales de 10s Trabajadores delEstado L A C Latinamerica and Caribbean MW MinimumWage NAFTA NorthAmerican Free Trade Agreement NGOs Non-governmental organizations PAC Programa de Apoyo a la Capacitacion PAHNAL Patronato delAhorro Nacional PET Programa de Empleo Temporal PROSAVI Programa Especial de Crddito y Subsidios a la Vivienda SAEBE Sistema de Apoyo Economico a Buscadores de Empleo SEDESOL Secretaria de Desarrollo Social SEW1 Self-employment without investment SHF Sociedad Hipotecaria Federal SICAT Sistema de Capacitacion para el Trabajo STDs Sexually transmitted diseases STPS Secretaria de Trabajo y PrevisionSocial TFP Total Factor Productivity .. 11 Acknowledgements The Urban Poverty report was managedby Marianne Fay andAnna Wellenstein. Other team members includeCaridad Araujo, MagdalenaBendini, Sara Johansson, Gabriel Montes, Mary Morrison, CaterinaRuggeri Laderchi, Mauricio Santamaria, andJesus Torres Mendoza. The team i s gratekl to the peer reviewers Emmanuel Jimenez and Christine Kessides, andto Gladys Lbpez-Acevedo for their comments. The team benefitedfrom inputsand suggestions from counterparts inthe mexican government, notably Rodrigo Garcia VerduofBanco de Mkxico; Claudia Acuiia, Craig Davis, Cristian Muiioz and Jose LuisRamos y Fusther ofSubsecretaria deDesarrollo Urbanoy Ordenacibn del Territorio, SEDESOL, and Carlos Maldonado Valera, ofthe Subsecretaria de Prospectiva, Planeacibn y Evaluacibn, also of SEDESOL. We are gratekl to Lic. Concepcibn StetaGandarafor documentationandinsights on the expansion of Oportunidades inurban areas. We thank SEDESOL for sharingthe results ofthe survey ofpoor barrios, andJorge Santibaiiez andAlejandro Garcia ofthe Colegio de la Frontera Norte for their helpinaccessingthe data. The Urban Poverty report is part of an overall analytical program on poverty inMexico managedby Gladys Lopez-Acevedo, MichaelWalton and Jaime Saavedra, that includedan initial Poverty Assessment (Poverg in Mkxico: An Assessment of Conditions, Trends and Government Strategy), and companion reports on ruralpoverty (A study of Rural Poverty in Mkxico) and social safety nets (Mkxico:An Overview of Social Protection), in addition to the presentreport on urban poverty. The team thanks Marcela Rubio Sanchez for her efforts inensuring consistency of data and data sources across the various reports. Vice President: Pamela Cox Country Director: Isabel Guerrero Sector Director: John Henry Stein Sector Leader: Anna Wellenstein Task Manager: Marianne Fay, Anna Wellenstein ... 111 Table Of Contents The ManyFacesofUrbanPoverty 1 Poverty, GrowthAnd Public Programs For The UrbanPoor 23 Places O fThe Poor: The Case O f Mexican Poor Barrios 35 Poverty and Labor Markets inUrban Mexico 52 Buildingup Buffers and SteppingStones: Asset Accumulation andAccess to 79 Financial Services Annex 100 Bibliography 112 iv I. THE MANY FACES OF URBAN POVERTY' Half of the moderately poor and one third of the extremely poor now live in urban areas. Mexico i s now a mature urban society with nearly three quarters o f its population living in urban areas, mostly in large cities o f more than 100,000 people. Although poverty incidence i s substantially lower in urban than in rural areas, the high urbanization rate means that half of the moderately poor and one third o f the extremely poor now live in urban areas. And the demographic trends are likely to urbanize poverty further since urban population i s expected to continue to increase faster than the ruralpopulation (Table 1.1). Table 1.1 Povertyis becomingurbanizedinMexico Year Total Population Moderately poor Extremely poor Rural Urban Rural Urban Rural Urban (million) 2002 39.7 63.4 26.8 26.6 13.8 7.2 2005 40.5 65.9 27.3 27.7 14.1 7.5 2010 41.5 70.2 28.0 29.5 14.4 8.0 2015 42.3 74.1 28.5 31.1 14.7 8.5 2020 43.0 77.6 29.0 32.6 15.0 8.9 Source: Own calculations using population projections from Consejo Nacional de Poblaci6n (CONAPO) and poverty estimates from World Bank (2004a). Note: CONAPO's population numbers differ slightly from those o f the household survey (ENIGH) and so the two will not yield exactly the same number o f poor. Rural settlements are defined as settlementswith less than 15.000 inhabitants. Cities offer a number of opportunities and specific challenges for the poor. Compared to rural areas, urban areas provide deeper labor markets, higher access to services and for some, freedom from oppressive social traditions and discrimination. This i s reflected in lower poverty incidence and higher average incomes. But the urban poor also face specific challenges. They mustgenerate cash for survival, which inturnmeans that they mustgetwage-payingjobs or employ themselves. Though there are more jobs available in cities, the urban poor may face difficulties in securing decently paid employment in the formal sector, for lack o f skills, childcare, transportation, or because o f stigma associated with where they live. Low qualityand highcostsrestricts real access to basicpublic services. Services such as electricity or running water are also more widely available in cities; the incidence o f health providers i s higher. However, the quality and reliability o f those services may be very low inpoor neighborhoods or they may be unaffordable. Also, because o f higher population density, the negative health externalities o f unclean water, patchy sanitation and solid waste management are very serious. Within cities the poor are often confined to living in unsafe (because of crime but also vulnerability to natural disasters) and/or unhealthy locations. And precisely because the incidence o f poverty i s relatively low in cities, urban-wide statistics tell us little about the This chapter was written by Marianne Fay and Sara Johansson and benefited from comments from Anna Wellenstein. 1 situation o f the urban poor. A better understanding o f such location-specific challenges for the poor can help the government o f Mexico design more efficient and cost-effective anti-poverty interventions. Yet, the urban-ruraldistinctions need to be seen as a continuum where depth and characteristics of poverty vary with settlement size. For analytical andpolicy design purposes, the urban-rural dichotomy can be misleading. Inwhat follows, "urban'' i s generally defined as all settlements with more than 15,000 people, which still covers very heterogeneous types o f urban centers. Yet, the difficulties facing poor people living in a town o f 15,000 people may be much more similar to those living inrural areas than those living in Mexico's larger towns - and policy interventions need to reflect this. Where possible, we therefore attempt to disaggregate the concept o f "urban'' into a more nuanced pattern o f settlements by size, in order to better illustrate what i s really a continuum from most rural to most urban. This task i s complicated by the fact that the Income and Expenditures National Survey (Encuesta Nacional de Ingresos y Gastos de 10s Hogares, ENIGH) - the household survey which i s a key instrumentfor determiningpoverty levels and analyzing its correlates - i s not designed to be representative at a disaggregated stratum level. As a result, there i s a risk of large measurement errors which needs to be kept in mindfor disaggregated statistics. The objective of this report is to informthe design of urban povertyinterventions. Thereby responding to the concerns and requests of the social cabinet and Social Development Ministry (Secretaria de Desarrollo Social, SEDESOL on Spanish abbreviation) and specifically, the Sub-secretariat for Urban Affairs, which i s in charge o f developing anti-poverty programs targeted at the urban poor. Formulating the requestedpolicy advice requires answering a number o f questions: what i s specifically urban about poor people living in cities?; are there different determinants o f poverty in urban areas?; i s the type o f deprivation suffered by the poor in cities different from what happens in the countryside?; and, most importantly, are the instruments to helpthe poor differentbetweenrural andurban areas? This document is organized as follows. The first section - comprising three chapters- examines what we know about the urban poor, seeking to understand better the dimensions o f urban poverty. The present chapter sets the stage in reviewing trends in urban poverty, the characteristics o f urban poor and the heterogeneity o f urban poverty. Chapter 2 considers the relationship between urban poverty and macroeconomic trends and examines the poor's coping mechanisms as well as the principal public programs available to them. Chapter 3 relies on a recent survey conducted by SEDESOL to look in more depth at life in Mexico's poor urban barrios. Section two o f the report discusses some o f the key challenges facing the urban poor, namely how to integrate labor markets and access "good" jobs (chapter 4), and how to protect themselves against income shocks by accumulating assets or accessing financial services (chapter 5). Companion reports discuss the country's ruralpoverty (AA Study of Rural Poverty in Mexico) and social safety nets (Mexico: an Overview of Social Protection). TRENDSINURBANPOVERTY Poverty incidencehasyet tofully recoverfrom thepeso crisis In 2002, some 11 percent of Mexico's urban population was estimated to be extremely poor and around 42 percent moderatelypoor. These incidences compare favorably with rural areas, where poverty incidence reached 35 and 68 percent respectively (see Annex 2 Table 1-A for more details). Butwith a large urban population, a relatively low poverty incidence still translates into large numbers o f poor people: 27 million urban poor, o f which 7 million are extremely poor (Figure 1.1). Urban poverty has not yet recoveredfully from the peso crisis of the mid-1990s. The number ofurbanpoor is still some five millionhigher than it was in 1992 although incidence has actually fallen somewhat. Moderate and extreme urban poverty largely display the same development between 1992-2002, whether poverty i s based on income or consumption measures (for a discussion o f the effects o f usingconsumption vs. income to calculate poverty, see Box 1.1) Poverty improved in the early 1990s, but increased sharply with the 1994-95 macro-economic crisis, and has only slowly recovered since. The depth and severity o f poverty follow the same pattern as the poverty headcountindex. Box 1.1.Urban poverty: Some definitions There are two definition o f "urban" in Mexico. The Instituto Nacional de Estadistica, Geografia e Informatica (INEGI) and most Government institutions classify as urban any settlement with population greater than 2,500. This implies an urbanization rate o f 75 percent. Instead, Population National Council (ConsejoNacionaldePoblacidn,CONAPO)the government agency in charge o f demographic analysis and projections, instead relies on an analytically more appealing cut-off o f 15,000. This yield an urbanization rate o f 61.5 percent, which is used throughout this report. Unless otherwise stated, the poverty and inequality indicators reported in this document are based on World Bank's estimates usingSEDESOL's official poverty lines for income in order to be consistent with official poverty lines. However, it is generally preferable to use consumption expenditure instead o f income to measure poverty. Consumption i s a better measure o f actual well-being, and consumption expenditure is also less likely than income to be mis- or under- reported in surveys. For comparison, consumption poverty rates were calculated, fixing these for one year to yield the same poverty rates as the official (income) poverty line in 2000. Fortunately, the levels and trends in poverty are fairly similar whether income or consumption is used (Box Table 1). Table 1.Poverty Trends for Extreme and Moderate Urban Poverty usingIncome and Consumntion 1992 1994 1996 1998 2000 2002 Extreme Poverty Income 13 10 27 21 13 11 Consumption 13 10 25 21 13 11 Moderate Poverty Income 44 44 62 56 44 42 Consumption 39 37 55 50 44 42 Source: World Bank(2004a) In general, these trends are consistent with those for rural areas, with two exceptions. First, while both rural and urban areas suffered from the peso crisis, the number o f urban poor increased by 11 million between 1994 and 1996, tipping the balance so that most o f Mexico's poor were now in urban, not rural areas. Moreover, the poverty increase was largely drivenby a tripling o fthe number o f extremely poor. Second, the pace o frecovery was quicker in 3 urban than in rural areas (Figure 1.1). However, neither has extreme poverty recovered to above 1994 levels (the relationship between urban poverty and economic growth is examined more closely inchapter 2). Figure1.1 - Urbanpovertywas strongly affectedby macro-economicinstabilityinthe mid-1990s 21n ............................................................................... Extremely......poor (left) and moderately......poor (right), inmillions, 1992-2002 .......................... ............................................................................................ ......................... ........ ............................................................................................................................................................ 40 35 34 ORural 31 30 27 0Urban 26 27 25 20 9r) 0 2000 1992 1994 1996 1998 2000 2002 19yz 1994 1996 1998 2002 Source: Own calculations basedon ENIGH, 1992-2002. Poverty incidence increases as settlement size decreases' The incidence of poverty falls along the rural-urbancontinuum. Urban areas form a very heterogeneous group, which is reflected in variations inpoverty between different types of urban areas: it i s lowest in cities o f more than 100,000 inhabitants and highest in small rural communities o f 2,500 inhabitants or less. The prevalence of extreme poverty i s three times higher insmall cities than inbig ones; the share ofpeople inmoderatepoverty is 60 percenthigher.But a third o f the poor - some 16 million people - still live in large cities, because o f the concentration o fpopulation inbiggerurban centers. The impact of macro-economicturbulence inthe mid-l990s,is much more visible in cities, both smaller and larger, than in small rural or semi-urban areas. However, across urban areas, the crisis-driven increase inpoverty incidence was at least as sharp in small towns as in larger ones and recovery was slower (Figure 1.2). Indeed, in 1998, the moderate poverty incidence was as highin small cities as insemi-urban settlements. We remind the reader that ENIGH- the basis for poverty estimates - is only designed to be representative at a national, rural and urban level, not at stratum level, nor at regional level, which in turn may introduce large measurement errors inthe analysis. 4 Figure1.2 Poverty incidenceincreasesas settlement size decreases, 1992-2002 Extreme poverty incidence(left) and moderate poverty incidence(right) 100 100 ,+~ 100,000or + 90 +15,000- 99,999 9 0 1 80 2500 - 14,999 70 60 50 40 : : I , , , , , 30 20 10 0 0 92 94 96 98 00 02 92 94 96 98 00 02 Source: Own calculations basedon ENIGH.Note: ENIGHi s only designed to be representative at a national, rural and urban level andnot at stratum level. These results shouldtherefore be considered indicative. Poverty incidencealso shows large regional variations Mexico's regions display vast differences in poverty rates and the geographical location of urban areas is consequently a strong determinant of poverty levels. In 2002, extreme poverty rates for urban areas ranged from 32 percent in the Southern Pacific region to less than three percent in Mexico City, and the differences are huge also for moderatepoverty. As seen inAnnex Figure 1.A, there i s a very strong correlation betweenurban andrural poverty rates within regions. Looking at trends, moreover, urban areas inthe Southern Pacific andthe Southern Gulf and Caribbean regions saw the sharpest increases in extreme poverty after 1994 and have subsequentlyrecoveredmuchmore slowly than the center region, for example. In all this suggests a regional divergence in urban (as well as rural) poverty rates. Interpretingthese stylized facts should be done carefully. First, part of what drives these regional differences may in fact be their urbanization patterns (Southern cities tend to be quite small as opposed to Central region where there are many larger town^).^ Second, the household survey i s not in fact designed to be representative at regional level so that these results should be seen as indicative. Note however that the fact that region matters more than the rurallurban dichotomy i s one that has been found in all countries where this analysis has been done (Brazil i s a notable example). Especially in the South Pacific region, there are relatively more small urban agglomerations (size 15,000- 30,000) and less very large agglomerations (size lOO,OOO+). The regional differences are not remarkable, however. 5 Table 1.2 Urbanpoverty trends vary enormously across regions4 Region 1992 1994 1996 1998 2000 2002 Extreme Poverty Incidence (% of population) North 8 8 19 14 7 5 Center 18 18 33 26 16 14 South-Pacific 37 23 47 44 29 32 South-Gulf and Caribbean 18 19 34 25 28 24 Mexico City 5 3 13 10 3 3 ~ Moderate Povertv Incidence (% of r>or>ulation) ~ North 35 38 55 45 30 31 Center 52 53 70 63 52 48 South-Pacific 76 59 78 74 61 67 South-Gulf and Caribbean 48 55 71 59 60 58 Mexico City 31 26 47 42 24 26 Source: Own calculations based on ENIGH, using income poverty lines. 1. ENIGH i s only designed to be representativeat a national, rural andurbanlevel andnot at regional level. These results shouldtherefore be considered indicative. CHARACTERISTICSOF URBANPOVERTY Urbanpoor have more children, worse housing conditions and less education than the non- poor Poor urbanhouseholds share many characteristics with poor rural households. Most o f them are located in the Central region o f Mexico, they tend to consist o f larger families with more children - and thus higher dependency rates - and live in more modest housing conditions. Poor urban households are also disproportionately headed by less-educated people. Unlike income poverty, some o f these indicators improved over the 1990s. So, for example, the proportion o f urban extremely poor living in dwellings with dirt floors has fallen from 25 to 18 percent since 1992. The percentage living in households whose head has not completed primary education has fallen from 69 to 57 percent. On the other hand it i s not clear that human capital gaps between - inparticular- extremely poor andnon-poor have narrowedsignificantly over time (see chapter 4). This follows CONAPOs (Consejo Nacional de Poblacion) classifications of Mexico into four regions, plus Mexico City: North (Baja California, Baja California Sur, Chihuahua, Coahuila, Durango, Nuevo Leon, Sinaloa, Sonora, Tamaulipas, and Zacatecas) Center (Aguascalientes, Colima, Guanajuato, Hidalgo, Jalisco, Edo. de Mkxico, Michoach, Morelos, Nayarit, Puebla, Querktaro, San Luis Potosi and Tlaxcala) South- Pacific (Chiapas, Guerrero, and Oaxaca), and South--Gulf and Caribbean, (Campeche, Quintana Roo, Tabasco, Veracruz and Yucatan). 6 Table 1.3 The urbanpoor have more children, worse housingconditions, less education than the non-poor Extremely Moderately Non- poor poor poor Average householdsize 5.8 4.6 3.6 Average number o f children under 12 2.6 1.8 1.5 % ofpopulationlivingindwellings with dirt floor 18.4 8.2 1.o % ofpopulationindwellingsthat are: borrowed 16.8 15.4 11.5 rented 14.0 17.7 16.5 own property 66.2 64.2 70.4 other 3.0 2.7 1.5 % ofpopulationlivinginhouseholds where: The household headis illiterate 22.0 14.1 3.7 The household headhas no education, primary incomplete 56.7 43.4 18.9 primary complete 26.6 28.6 21.7 secondary complete 14.2 21.6 25.1 The householdheadworks inthe informal sector 41.6 34.3 21.6 Source: Own calculations basedon ENIGH2002. Consumption patterns do not differ markedly betweenthe poor in rural and urban areas, however, with two exceptions (Table 1-4). First, food expenses make up a smaller share o f urban poor households' consumption share, but this difference may simply reflect that poverty i s deeper in rural areas. Second, housing expenses account for a larger share o f total consumption in urban areas - indeed urban poor spend a fifth o f their income on housing. This i s a direct implication of higher congestion cost inurban areas which results inmuch higher land prices than in rural areas (low income housing markets are dominated by land markets.). In fact, access to affordable housing i s one o f the key challenges facing the urban poor. The poor, particularly moderatepoor, also tendto spend somewhat more on education. Table 1.4 Urban poor spend less on food and moreon housingthan ruralpoor Urban Rural Extreme Moderate Non- Extreme Moderate Non- Poor Poor Poor Poor Poor Poor Food, Beverage and Tobacco 42% 37% 24% 48% 42% 32% Clothing 4% 5% 5% 6% 6% 5% Housing 19% 22% 23% 14% 16% 19% Health 4% 4% 4% 4% 5% 4% Transport and Communications 10% 13% 16% 7% 13% 16% Education and recreation 8% 9% 14% 7% 7% 10% Other 12% 11% 13% 13% 12% 14% Total 100% 100% 100% 100% 100% 100% Note: Includes household articles andhygiene. Source: Own calculations basedon ENIGH2002. 7 More of urban poor work in agriculture are self-employed or in informal occupations than non-poor Employment characteristics of household head are also a key poverty correlate. Compared with non-poor, a higher share o f poor individuals live in households where the household head i s self-employed or working in the informal sector (Table 1-5). Note that significant share o f rural laborers among urban extreme poor i s similar to what has been found in other countries and i s indicative o f an "urban agriculture" at cities periphery. On the other hand the share working inmanufacturing inurban areas i s fairly similar across income levels. Table 1.5 The labor-market characteristicsof the urbanpoor comparedwith non- poor Extreme Moderate Poor - Poor - Non-Poor - UrbanLRural UrbanLRural Urban-Rural Proportion inthe population 11.4 34.8 30.6 32.7 58.0 32.6 Rural-urban composition 35.0 65.0 60.6 39.4 74.5 25.5 Profileby education N o Education - Primary Incomplete 50.1 73.0 33.6 64.4 17.5 47.3 Primary Complete 26.8 20.4 29.4 21.8 21.0 21.3 LSecondary Complete 18.4 5.4 27.8 12.1 25.1 15.5 USecondary Complete 3.7 1.o 8.0 1.5 19.6 8.0 UniversityComplete 1.o 0.2 1.3 0.3 16.8 8.0 Total 100 100 100 100 100 100 Profileby employment Non-agricultural Laborer 57.0 17.2 70.6 36.0 69.8 43.0 Agricultural Laborer 8.7 32.1 1.8 22.4 0.5 9.6 Employer (under 5 employees) 1.3 2.6 2.6 3.8 7.3 9.1 Employer (5 or + employees) 0.0 0.5 0.1 0.1 2.4 0.8 Self-Employed 33.0 47.6 25.0 37.7 20.0 37.4 Total 100 100 100 100 100 100 Profileby sector o f activity Agriculture 11.5 70.5 3.0 49.7 1.2 30.9 Extraction 0.1 0.1 0.3 0.8 0.8 0.4 Manufacturing 35.0 9.7 42.9 18.4 39.7 29.7 Utilities 17.1 9.4 13.0 11.6 7.8 7.4 Construction 0.3 0.0 0.7 0.1 0.9 0.5 Commerce 4.6 0.7 5.3 1.5 4.8 3.5 Transportation 8.3 1.6 7.8 5.4 8.0 4.2 Financial Services 0.7 0.2 1.2 0.1 1.8 0.1 Services 22.5 7.9 25.9 12.4 35.0 23.3 Total 100 100 100 100 100 100 Source: Own calculations basedon ENIGH2002. Labor-market access is clearly linked to the poverty status of the individual (Figure 1.3). Employment rates - the share o f population aged 15-64 that i s currently employed - are similar for the extremely poor andmoderately poor, butmuch lower for both groups than for non- poor. This i s largely because a higher share of the poor i s inactive in the labor market but also because a higher share i s unemployed. Unemployment rates - as calculated from the household 8 Fi17 survey - are nearly twice as high for the poor as for non-poor. Inrural areas, unemployment rates are in fact lower for the poor than for n~n-poor.~ Figure1.3 The poor's participationinthe labor marketis muchlower than the non- poor's =Work =Searching =Not searching 100 40 30 20 56.1 0 Extremely poor M o deratelypoor Non-poo r Source: Own estimatesbasedon ENIGH2002. Higher dependence on labor income Labor markets are key for urban poor. The importance of labor markets for urban poor - and non-poor -i s clearly borne out in Figure 1.4, which shows income by source, location (urban or rural) and for the differentincome quintiles respectively (usingrural andurban quintiles separately). In fact, urban households show great resemblance in the pattern o f income sources, irrespective o f income level. Compared with richer urban household, the poor rely slightly more on labor income and on in-kindincome and receive virtually no income from physical assets like rental or financial income. Urban households receive a similar share o f transfers across income levels (about 4 percent). Urbanpoor rely much more on labor income and much less on in-kindincome and receive fewer transfers. Compared to their rural counterparts, labor income i s much more important for urban poor. Another striking difference i s that transfers make up a relatively much larger share o f rural poor and non-poor household income and remittances contribute virtually nothing to urban poor household income. This being said, transfers received by urban poor remain higher in absolute terms than for rural poor, although they receive considerably less (in absolute and relative terms) from Oportunidades.Instead, transfers receivedbyurban households, whether poor or non-poor, tend to consist much more o f pensions, severance payments, scholarships and gifts from other households6 (on the other hand, given higher poverty incidence in rural areas, the first quintile in urban area encompasses both urban extremely poor and 5 Note that the household survey is not an ideal instrument for analyzing labor market indicators and the unemploymentrates are therefore not consistent with official unemployment rates. An in depth study of the labor market situation for low-income groups, based on much richer employment surveys, is provided inchapter 4 of this report. 6 See World Bank (2004a): Mexico Poverty Assessment: Poverty in Mexico: Conditions, Trends and Government Strategy. 9 moderately poor, while the first quintile in rural areas only includes the extremely poor, which makes the comparison difficult). Figure1.4 Urbanhouseholds are muchmore dependent on labor income than rural ones income sources, by locationand quintiles, 2002 - 1 OLabor OBusiness/Cooperative .Transfers ORemittances .Other* Olmputed lncomc UrbanQ1 Urban Q2 UrbanQ3-5 Rural Q1 RuralQ2 Rural Q3-5 ~ Source: Own calculations based on ENIGH 2002. * Includes rental income, financial income, and other monetary income. Mexico actually stands out in Latin America for the remarkably low share of income that its urban poor receive from transfers. A recent regional study on urban poverty (World Bank, 2005) shows that while Mexico's urban poor only receive about 3.5 percent o f income from transfers the Latin America and Caribbean (LAC) average i s more than four times that amount. Only two other countries (Chile and Venezuela), out o f 18 comparators, exhibit a similar pattern o f lower transfers to urban areas, but even then the difference i s much smaller. This i s obviously not due to the fact that Mexico transfers less since Mexico's rural poor receive more than the L A C average in transfers. Pensions are about the same as inthe rest o f the region, significantly higher inurbanthan rural areas, but not enough to offset the lower levels o f transfers inurbanareas. Higher accessto sewices but quality and affordability aproblem Urban areas almost by definition have higher access to basic services than rural areas: electricity can be accessed (illegally or legally), there i s often some form o f water and sanitary infrastructure network in place, even for the poor, and there i s a higher density o f education and health services. All o f these aspects tend to improve gradually along the urban- rural continuum. Table 1.6 gives some municipality-level characteristics by population size. The larger the municipality, the greater i s its average access to basic infrastructure, and the higher are literacy rates. 10 Table 1.6 Urban municipalities have more services and higher education and wage levels Very Semi- Small Medium Large large Urban Characteristics Rural urban urban urban urban urban total % ofpersons indwellings with no pipedwater 26.4 29.2 30.1 26.4 22.4 13.6 25.1 with no sanitary drain 66.5 55.3 47.6 39.5 32.4 16.1 37.8 with no electricity 10.3 13.2 11.6 9.4 7.0 3.2 8.9 % ofpersons 15 and older that i s illiterate 20.7 20.4 19.4 16.1 13.5 7.3 15.6 with incomplete elementary school 32.2 30.2 27.9 25.4 23.1 15.6 24.4 % ofpersons employed earning 5 two min.wages 83.3 77.2 72.4 67.2 62.1 46.8 65.2 employed inmanufacturing 10.7 12.6 13.6 15.1 16.7 19.3 15.5 employed in services 28.8 34.8 40.3 46.5 52.5 66.4 48.2 Observations 381 998 430 239 183 165 1,017 Source: Own estimates based on census 2000. Population cut-offs are: rural: 2,500 inhabitants or less; semi-urban: 2502-15,000; small urban: 15,001-30,000; medium urban: 30,001-50,000; large urban: 50,001-100,000; very large urban: more than 100.000 inhabitants. However, average municipality characteristics do not distinguish the situation of poor from that of non-poor. Since population size is also inverselycorrelatedwith the incidence o f poverty, these average characteristics become increasingly irrelevant for understanding how the urban poor fare as one moves along the urban spectrum. Precisely because income poverty i s lower the more urban the area the poor will tendto be further away from the average and the gaps betweenpoor andnon-poor often more pronounced. The main infrastructure issue facing urbanpoor is morelikely to be the quality and affordability of services rather than actual lack of ~ervice.~ Nevertheless, a full 15 percent o f extreme poor in urban areas and 9 percent o f moderate poor do not have access to piped water and access to sewerage remains low. This, combined with the fact that water service i s intermittent and unreliable in many poor neighborhoods, has severe environmental and public health consequences in densely populated areas (see below). Note that some forms of network infrastructure services, such as sewerage, are only appropriate above a certain population density. As a result, their absence m a y not necessarily indicate as serious a deprivation in rural as inurban areas. 11 Table 1.7 The urbanpoor have moreaccess to infrastructure services than the rural poor Urban Rural % individuals living in Extremely Moderately Non- Extremely Moderately Non- households poor poor poor Poor poor poor Without electricity 2.6 1.1 0.0 10.5 8.2 2.3 Without running water or bypipe 15.0 9.0 1.3 42.8 36.4 14.6 track Without sewerage 29.4 16.8 2.8 86.1 76.7 40.1 Source: Own calculations basedon ENIGH2002. Within urban areas, access to services is also higher for older settlements. This is particularly true for the poor because o f the way inwhich they acquire housing -often settling in undeveloped areas where land can be acquired cheap and gradually buildinghouses. Thus, data from a survey o f poor barrios carried out by SEDESOL in 2002 shows that the share o f dwellings with walls o fthe bestmaterials goes from 57 percent for houses less than a year old, to 84 percent for houses more than 10 years old. As discussed in chapter 3, similar pattern i s found for piped water where both the age o f a settlement and the age o f a house matter in determining access, as well as for electricity and solid waste removal. The only service for which this was not the case was sewerage. Similarly, although education and health services are more widely available in urban areas, overcrowded- rather than absent - facilities limit effective access. Inaddition, poor are much more likely to receive low-quality services than non-poor.* As a result, actual health and education outcomes are more similar between urban and rural areas than would be expected. Inparticular, primary enrollment i s the same for rural and urban poor children. In fact, urban extreme poor do somewhat worse than rural extreme poor interms o f secondary enrollment (Figure 1.5). Thus, the key determinant for access to education appears to be income, rather than location, although this changes for higher levels o f education. In particular, rural-urban differences are significant at higher secondary education but enrollment rates o f the extremely urban poor are still only one halfthose o f non-poor urban children. The qualitative evaluation made at the time of Oportunidades' expansion into semi-rural and urban areas reports complaints with saturation of clinics in semi-rural areas (15,000 and 50,000 person settlements) and with the poor quality of schools inurbanareas (Escobar Latapiand Gonzalez de la Rocha, 2002 and 2004). 12 Figure1.5 Education:poverty rather than locationis what matters Enrollment ratesinprimaryand lower secondary education(ages 12-14) liL % Lower secondary enrollment J 83 96 0Rural 0Urban 95 80 0Rural 0Urban 66 T 90 90 I 1 - 54 88 86 I 30 84 20 82 10 80 Wremelypoor 0 I moderatelypoor non-poor Wremelypoor moderatelypoor non-poor I Owncalculationsbasedon uce: ENIGH2002. This rather surprising finding is confirmed by analysis conducted by SEDESOL using census data. Using intermediate poverty line ('pobreza de capacidades) which lies between the thresholds for extreme and moderate poverty, SEDESOL findings confirm that rural and urban poor enrollments are almost the same, with urban poor children somewhat worse off at lower grades and somewhat better off at higher grades (Figure 1.6). Whether this i s due to Oportunidades-which has had significant impact on enrollment inrural areas butnot incities - i s unclear. Limited health data available also suggests that there is not much difference in health status of the ruraland urban poor, at leastwhen it comes to infectiousdiseases(Table 1-9). This i s remarkable, given urban poor's much higher physical access to infrastructure and health services -about one in three rural people has to travel more than 30 minutes to a health facility, compared with one in six in urban areas.' On the other hand, data suggests a close link between income level and health status in both urban and rural areas. In 1999, incidences o f diarrhea and respiratory diseases for children under 5 years o f age were much higher for the first quintilethan for richer income groups. SEDESOL, Programa Nacional de Desarollo Social, 2001-06 13 IFigure1.6 School attendance is comparably low for rural and urban poor - S c h o o l attendance by age, s o c i o e c o n o m i c status, a n d p l a c e o f residence, 2000 6 0 'u pi 5 6 7 8 9 10 11 1 2 13 1 4 15 16 17 18 19 2 0 2 1 - N o n p o o r h o u s e h o l d s I+ U r b a n p o o r h o u s e h o l d s R u r a l p o o r h o u s e h o l d s Source: SEDESOL, 2003. However, data suggests that within urban areas location may matter even more than income. The 2002 survey o f poor urban barrios discussed in chapter 3 finds an even higher incidence o f diarrhea and severe respiratory diseases in the barrios (which are somewhat heterogeneous in terms o f income and so would include a group whose average income i s probably higher than the poorest quintile) than among the poorest quintile o f population (Table 1.8).1° Table 1.8 Incomeor location?Poor children are equally at riskfrominfectious diseasesin rural and urban areas SEDESoLBarrio Survey, 2002 National Health Survey, 1999. Poor barrios Poorest Qr2 Qr3 Richest Diarrhea Urban 22 15 10 10 6 Rural _ _ 14 11 8 Severe respiratory diseases Urban 48 21 19 16 13 Rural _ _ 24 22 10 Source: Adaptedfrom SEDESOL, ProgramaNacional de Desarrollo Social2001-2006 andown calculations. In sum, it does appear as if services (infrastructure, health and education) are failing, urban poor, despite the relatively high coverage... Inrelation to water and sanitation sector, this i s due to a combination o f the usual gradual way inwhich poor urban neighborhoods acquire services and a sector that i s generally dysfunctional, except in a few municipalities with well-run utilities (for a discussion on water sector see World Bank, forthcoming 2005~). Concerning education, quality remains an issue and one that i s central to poor's ability to escape poverty through the labor markets (chapter 4). As to health, the key issues concern environmental health issues (high density living, combined with pollution and poor sanitation) and saturation o f loAn in-depth analysis of the barrio surveys is provided inChapter 3 of this report. 14 clinics. However, these are part o f a broader set o f public health question related to the epidemiological transition that occurs with urbanization and that affects the urban poor worst (Box 1.2). Box 1.2. Public healthissuesfor the urbanpoor Across Latin America, a number o f key health indicators for poor urban children are as bad or close to what they are for poor rural children, despite the much higher availability o f services. This suggests an "urban penalty" due to high density living inunhealthy environments.' The urbanization o f Latin America has contributed to a dramatic change in its epidemiological profile: infectious diseases are being replaced by chronic, degenerative and cardiovascular diseases and violence as the leading causes o f death. Although there has been little systematic research on the different incidence in pathologies between urban and rural areas, certain pathologies like obesity, nutritional problems, sexually transmitted diseases (STDs) and HIV/AIDS, injuries from accidents, violence, drug addiction and mental health problems seem to be more prevalent in urban areas. In addition, they seem to hit the poor harder. In particular, obesity is more prevalent among the urban poor, possibly due to the lower price o f high calorie processed food. In addition to the well documented effect of inadequate water and sanitation services, some criticism i s emerging as to the fact that health care in Latin America have been slow to adapt to the epidemiological transition. Certainly, health issues o f the urban poor are an understudied issue, particularly in Latin America. The term "urban penalty" was coined in 19th century England when urban mortality rates, particularly from tuberculosis, were much higher than rural ones. Public health measures, improved water and sanitation, and socio- economic change ledto declines ininfantmortality rates (IMR) so that by 1905 rural andurbanIMRswere similar (See www.Urbanobservatrov.org for more details). Source: Based on Monkkonen, 2004. What about the indigenouspoor?" "Cuando uno llega aqui la ciudad te come. No te comes a la ciudad, ella te come y no te deja salir. Cuando llegas aqui todo te sabe sabroso, ves la luz del dia, de diay de noche; en elpueblo, no, ahi conpura velita, conpetrdeo, y llegas aqui, haz de cuenta que todo estaba oscuro y seprende la luz, asi es la ciudad, todo es bonito pero no sabes lo que te va apasar, no sabes 10s gokes que despuks te va a dar por el trasero, esta ciudad te come y asi como te come luego te tira y te desecha". Mazahuas Indian, Mexico City, August 2000, as quoted inMolnar, Carrasco andJohns-Swarts (2003) A recent study sheds some light on the situation of indigenous urban dwellers in Mexico. There i s not a lot o f information about the specific conditions o f indigenous people in the cities o f Latin America, although there are a few studies that shed light on the topic. A recent such study done inMexico by Molnar, Carrasco and Johns-Swartz (2003), i s particularly relevant for the present report since it actually asks the indigenous communities interviewed about l1This section is a summary of Molnar, Carrasco and Johns-Swartz (2003) unless otherwise specified. 15 differences in living conditions between rural and urban areas). The study i s based on focus groups and a survey of about 860 indigenous households in three cities - Mexico City, Cancun and Coatzacoalcos-Minatitlan, a traditional industrial center. Indigenous households include people from five different indigenous groups and included both first generation migrants, as well as second and third generation. Most households live in poverty or extreme poverty in urban areas. The studies show a wide variety o f experiences and situation across cities, indigenous groups and even individuals. However, some stylized facts emerge, summarizedbelow. Our understanding of the indigenousdimension of urban poverty is limited by data limitations. According to 2000 census, about 40 percent o f people speaking indigenous language live in urban areas, representing about 1.3 million individuals (INEGI, 2001). However, this seriously underestimates the urban indigenous population since many second and third generation indigenous migrants maintain their cultural identity but not necessarily their language. In addition, it appears as if official statistics about indigenous people in cities are biased downwards because o f their lack o f fixed address and legal residence - this is particularly common in large cities, touristic centers with seasonal employment and frontier cities with fluctuating population. In the study, this was particularly complex in Mexico City, but less so in Cancun and Coatzacoalcos-Minatitlan. Indigenous groups are characterized by strong social networks - but these are not so useful to get out of poverty. Indigenous migrants typically have very strong community organizations which provide a safety net and help negotiate access to services. However, most urban indigenous communities continue to live on the margin o f society, suggesting a lack o f vertical relations from the communities to service providers (municipality, federal or state administration in charge o f particular programs). This may in part be due to the fact that most recent immigrants come from situations o f extreme poverty and get to the city with low level o f social or even human capital. Inaddition, community organizations do not seem to help get out o f poverty. Those looking to get ahead seem to rely more on education and the use o f Spanish language and Spanish culture, than on traditional indigenous means. There is significant discrimination. The perception of discrimination exists particularly with respect to outward signs of indigenous culture, such as traditional clothing and language. As a result, many abandon indigenous costume in urban centers. Nevertheless, about 75 percent o f people interviewed felt that they had the same work opportunities as non-indigenous (although this varied significantly among ethnic groups). Between halfandthree-quarters ofthe interviewed felt their salary was fair. Those who spoke an indigenous language were more likely to rate their salary as unfair. Finally, informal workers felt more at liberty to publicly express their cultural identity. Gender discrimination is less prevalent than in rural communities, however. Across all five ethnic group included in the study, women declared that, contrary to what happened in rural communities, they participated in important family decisions. The majority (54 percent) declared that they themselves chose their work, when to visit their parents, whether to work outside home or use contraceptives. Inter-generational differences emerged however, with the second generation reporting higher levels o f autonomy, higher levels o f education, more formal employment (as well as less use o f their indigenous language or link with the village o f origin). In contrast with rural areas, the vast majority (three quarters and above, depending on the particular indigenous group) o f households consider it important to invest ingirls' education. The main reasonfor migration is a push factor. The impossibility inrural community o f origin o f ensuring access to education and a job with minimum income, or, sometimes, to 16 access to property. And whereas most indigenous migrant workers performs in the informal sector and earn little, they report earning more and having steadier, more secure work than in rural areas. The general opinion i s that even if wages are indeed low in the city it i s possible to secure a steady stream o f income -as long as illness does not hit. Traditional indigenous activities remain important sources o f earnings inthe cities, particularly for women, who often combine the production o f traditional handicraft with housework. Access to services is elusive. Better access to services i s a powerkl motivation for migration to the cities and many perceive urban life positively for the opportunities it offers to their children. Inparticular, educationis perceivedas the most importantmeanto get ahead andis much higher among youths or second generation than among first generations o f migrants or older people. Nevertheless, many children abandon school before finishing primary school and many repeat grades due to economic duress, the impossibility to attend regularly or the lack o f parental help for homework. Access to health services varies widely across ethnic groups from a low o f 7 percent - among the Mazahuas to a high o f 57 percent among the Zapotecas (who are mostly employed in the formal sector Coatzacoalcos-Minatitlan, a traditional industrial center). In addition, about 25 percent o f those with access to health services report not usingthem - due to their poor quality, a fear of discrimination and poor treatment, or to limited Spanish language skills. Overall, only about half o f pregnant indigenous women get pre-natal care although the majority gave birth in clinics or hospitals. City life has advantages and disadvantages for the indigenouspoor. People surveyed felt life i s better in cities in terms o f work, education, health, justice and housing, but worse in terms o f security and environment. Answering the question o f what was the most serious problem confronting the community, commonly used answer was crime and violence (35 percent) followed by lack o f services (20 percent), with menmore likely to worry about crime and women more concerned about services.12 Unemployment was mentioned by 9 percent and housing 6 percent. Interestingly, poverty was reported as a problem for the community in only 2.7 percent but with a range from 1.4 percent among the Mazahuas to 17 percent among the Zapotecas. Interviews showed the variance o f concerns across social groups. Women were more likely to worry about the excess o f work; time lost in transit and spent helping sick relatives. In contrast, youths complained about the lack o f supervision from parents, conflicts with alcoholic parents, time spent inthe streets after school, and culture of drug andviolence. Finally, although cities are perceived as places o f greater opportunities they are not seen positively: about 46 percent think that city people are "worse" than country people. In sum, poverty incidence and symptoms clearly differ between urban and rural areas, as well as among urban areas of different size and location. But how important are differences in well-being within cities? The fact that urban poor in many cases have more in common with their rural counterparts than with other urban dwellers hints at substantial heterogeneitywithin urban centers as well. l2"Services" seems to refer to infrastructure services, since "education" was a separate answer. 17 INEQUALITYAND HETEROGENEITY High inequality tends to dampen the effect of growth on p~verty.'~ Below we use four indicators o f inequality to illustrate patterns and trends in urban inequality: the Gini index, and three indices that are part o f the generalized enthropy (GE) class o f indices, and which can be varied so as to pay more attention to inequality at different segments o f income distribution. The GE(0) index (also known as the Theil index) gives more weight to inequality at the bottom o f the income distribution (Le. among the poor), the GE(1) index gives equal importance to inequality along the distribution, and the GE(2) index, finally, gives more weight to inequality among the richer groups. Inequality now lower in urban than rural areas Inequality indicators suggest an increasein inequality between 1992 and 1994, a fall in inequality connected with the macroeconomic crisis and a flattening out during the recovery period. As Figure 1.7 shows, the four indices had roughly similar trends for inequality inthe 1990s. However, inequality within the top income distribution (GE(2)) has been somewhat more volatile than other indices and also increased in the recent period. The trends in inequality are consistent with findings in the previous section on income vulnerability: although the mid- 1990s crisis increased absolute poverty, it was in fact progressive as it affected households in the top of the income distribution relatively more than those at the bottom. The decrease ininequality in urban areas coincided with a rapid increase in inequality in rural areas, especially between 1998 and 2000. As a result o f these disparate trends, inequality i s now higher in rural areas than in urban ones - the right panel o f Figure 1.7 shows the Gini index only, but in 2000, rural inequality was also higher for all the GE indices. :::u Figure1.7 Urban inequality has fallen and is now below rural inequality Urban areas: Giniand GE indices, 1992-2002. Urban and Rural areas: Giniindex, 1992- 2002 2 1 ORural OUrban 11 0 9 ".'- 0 10 0 7 0 5 0 3 0 00 1992 1994 1996 1998 2000 2002 1992 1994 1996 1998 2000 2002 Source: Own estimatesbasedon ENIGH(2002). Mexico's urban areas display fairly low inequality, at least by Latin American standards, where inequality is notoriously high. As shown in Figure 1.8 below for the Gini index and in Annex Table A-4 for other measures o f inequality, Mexico has among the lowest l3See Ravallion (1997)for a discussion 18 urban inequality for all indices discussed above. Regardless o f inequality measure used, urban areas are less unequal than rural ones, although differences are slight - except for the GE (2) measure that gives more weight to inequality among the rich and for which Mexico exhibits the highestrural-urban difference inthe L A C region. Figure1.8 Urban Ginicoefficients, Latin America and Caribbean Regio Brazil I Nicaragua II Chile I I G uate mala Colombia Jamaica Bolivia Ecuador Panama El Salvador Paraguay Dominican Republic I Venezuela I I Mexico Costa Rica I I Peru I 1 0 10 20 3 0 40 5 0 6 0 ource: For Mexico, own calculationsbasedon ENIGH2002. Other countries: World Bank (2004a). Vey heterogeneouscities14 Since cities are very diverse socio-economically, urban poor are part of a much more diversified economy. Land and rental markets result insome sorting by neighborhood yet, inMexico at least, considerable heterogeneity persists within neighborhoods. Thus it was found, inthe context of Oportunidades,that while 77 percent ofurban extreme poor live in"marginal" neighborhoods, extreme poverty rate in such neighborhoods i s only 26 percent (GutiQrez, Bertozzi and Gertler, 2003). Even the poorest barrios, such as the ones that were selected for SEDESOL survey discussed in chapter 3, only exhibit an average poverty rate o f 36 percent (in contrast, rural average for extreme poverty rate i s almost 40 percent). This complicates targeting for social safety-net programs; inparticular geographic targeting i s likely to involve high errors o f both inclusion and exclusion. We look at heterogeneity within and between municipalities that belong to same metropolitan area, using municipality-level census data for 1990 and 2000 (large Mexican cities are composed o f several municipalities). As a proxy for "well-being", two different indices o f marginality are presented in order to distinguishbetweenwell-beingfactors whose provision i s l4Based on Araujo (2004) 19 o f public nature (access to water, sanitation, electricity and literacy) and those which are privately determined (crowdedness and floor material o f the dwelling, as well as wage levels). Since these are indicators of marginality, a low level i s a good outcome. Within cities, municipalities differ considerably in terms of marginality (annex Figures2-A and 3-A). Many metropolitan areas display relatively low levels o f marginality (Le. high levels of well-being) for both private and public indicators, but the range among municipalities within each o f these zones i s large.15 The ranking o f these zones also differs between private and public indicators. The three largest cities o f Mexico - Mexico City, Guadalajara, and Monterrey -tend to have low average levels o f marginality for both private and public indicators, but these cities are also characterized by high inequality among municipalities, especially Monterrey. For each zone, the dispersion in the public marginality index i s generally higher than dispersion in private marginality, suggesting that there are larger within-city differences in well-being related to public provision o f services. This i s consistent with the findings of Boltvinik's (1997) study of four colonias around Mexico City which showed much greater heterogeneity in unsatisfied basic needs than in terms o f income.16 In addition, there i s also a weak tendency for cities with higher levels o f public marginality to exhibit greater degrees o f dispersion - in other words, poor cities are also more unequal -but no clear relationship exists for private marginality. Unequal cities tend stay that way What do we know about the dynamics of spatial heterogeneity within cities? The decline in urban inequality discussed above does not necessarily mean that inequality among people living in the same city has fallen, however. A key issue i s therefore whether cities have tended to grow jointly or disjointly, i.e. whether there has been convergence in well-being over time inmetropolitan areas between 1990 and2000. First, have poorer cities - Le. with high marginality - fared better on average than more affluent ones? There seems to have been some convergence between cities for average levels o f marginality related to public provision, but there i s no clear pattern for private-level deprivation (Figure 1.9 and Annex Figures 5.A and 6.A). That is, less well-off cities saw a faster improvement (larger reductions) in public marginality but private deprivation levels did not fall faster than for richer cities. Good news emerges from Figure 1.9 since all cities experienced a reduction in their average level o f private marginality. Most cities also improved public marginality, although a few relatively well-off cities (including Monterrey) actually saw an increase inthe average public marginality index. l5Note that average levels of well-being for each metropolitan zone is calculated as the mean of municipal averages - and thus does not take into account population weights. l6The indicator that varied most within colonias was access to health services. 20 Figure1.9 There has beensome convergenceinthe city averagepublic marginality index but not inthe private marginality index 0 - , , , I -2 1) 10 20 30 4c 0 -4 0 ~ 0 4 2 -6 ~ -8 ~ 7 a, -10 40 ** ~ 0) -12 ~ * * 66 -14 ~ -16 ~ Rblic margin%y index301990 10 4c -18 ~ Rivatemarginality index 1990 ource: Adapted fromAraujo (2004) basedon census data 1990 and 2000. Second, has inequality in well-being rising within cities? The range o f municipal marginality indices, whether private or public, has indeed fallen for almost all cities. Though range admittedly i s a very crude measure of variance, this indicates that within cities there was less inequality between municipalities in 2000 than in 1990. Yet, the ranking appears to be more or less preserved: cities that were relatively unequal in 1990, whether with high or low average levels o f well-being have in general stayed relatively unequal regarding private and public marginality. These include, notably, Mexico City and Monterrey (Annex Table 3-A). A third questionrelates to the extent to which growth - here meaning the change in marginality - differs within cities. Generally, cities that in 1990 had a highdispersion o fpublic or private marginality also saw widely differing rates o f reduction in marginality across their municipalities (Figure 1.10). An interpretation i s that relatively unequal cities tend also to grow disjointly, thus supporting the observation that unequal cities generally have stayed unequal. Figure1.10 Unequal cities grow disjointly Municipality marginality in 1990vs. changes of municipality marginality between 1990- 2000 0 s 16 0 0 5 0 141 25- 0 m .^ 10 zm 20, * 4 11 - * 4 0 10 20 30 4C 0 20 40 6C Rivate marginality range 1990 Public marginalityrange 1990 ource: Adapted fromAraujo (2004) basedon census data 1990 and 2000. 21 T o sum up, there is clearly great variation in the dynamics of changes in marginality over time. Importantly, cities that on average were less well-off in 1990 have seen a relatively faster reduction in public marginality, but there i s no similar catch-up process for private marginality. Inequality, finally, seems to persist over time, suggesting that unequal cities also grow disjointly, with differentsub-areas improving at very different speed. SUMMARY AND CONCLUSION Location is a key element for understanding the structure and trends in deprivation and therefore for formulating the policy interventions neededto reduce it. However, there i s a continuum between rural and urban areas in terms o f poverty and its characteristics. Also, regional differences may matter more than rurallurban ones. As such, one should exercise caution inusingthe ruralhbandichotomy. Income sources in particular differ between urban and rural poor, with the former more dependent on labor income and less dependent on transfers. Urban poor share many household characteristics with rural poor - they have larger families, less education and less access to services than the rich. However there are some notable differences. In terms o f consumption patterns, they spend relatively more on housing (twice the share), transport and education, but relatively less on food, clothing and health. They are much more dependent on labor market for income. The most remarkable difference i s the surprisingly low share o f urban income that i s derivedfrom transfers. Infact, this i s the only dimension on which the situation o f urban poor inMexico i s substantially different from the rest o f Latin America. Clearly, this i s not due to low overall expenditure on social safety nets, since rural poor's share of income derived from transfers i s actually higher than regional average. Public services are more available in cities but tend to fail for poor due to low quality and reliability. Another notable difference i s urban poor's greater access to infrastructure and other public services. But here again there are some worrisome caveats. First, despite much higher access to infrastructure and health services inurban areas, infectious diseases such as diarrhea and acute respiratory infections are equally common among urban and ruralpoor children. This suggests that water and sanitation systems have remarkably poor quality. While much attention i s being paid to expanding coverage quality i s not keeping up, making the improved access only nominal. It also implies that using access to services as a targeting instrument may lead to misleading conclusions about well-beingin urban areas. Second, school enrollment rates are equally low among urban and ruralpoor. This may reflect the positive impact o f Oportunidades in rural areas, but it does also show an urgent need for assessing the educational situation o f urban areas. Urban areas have become less unequal but remain highly heterogeneous. Urban areas today are somewhat less unequal than rural areas. However, they are highly heterogeneous, within cities and even within poor neighborhoods. This is true for income, whether measured in pesos or through an asset index. Heterogeneity in terms o f access to services i s however greater than interms o fhouseholdcharacteristics such as wage levels or house quality. The incidence and manifestation of poverty varies enormously between and within different urban areas, and even within particular neighborhoods. This has important implications for the targeting mechanisms o f social safety-net programs: in particular that geographic targeting i s likely to involve higherror rates for both inclusion and exclusion. 22 2. POVERTY,GROWTHAND PUBLICPROGRAMS FOR THE URBAN POOR^' Urban poverty inMexico is highly sensitive to macroeconomictrends. The peso crisis and its aftermath triggered a sharper rise inpoverty inurban areas compared with rural ones, but urban levels fell back again more quickly in the subsequent economic recovery, as shown in chapter 1. This higher sensitivity to growth implies both good and bad news for Mexico's urban poor: on one hand, they may be better able to take advantage o f periods o f highgrowth than those in rural areas where poverty is more stubborn. On the other, more people are likely to fall into poverty duringnegative macroeconomic shocks. This chapter examines the relationship between urban poverty and growth. It explores the link between household characteristics and income variability and discussing the coping strategies that urban poor uses to protect themselves from the effects o f economic shocks. Itthenreviewsthe programs specifically designedto helpurbanpoor. URBANPOVERTY - MORE RESPONSIVETO GROWTHTHAN RURAL POVERTY Urban poverty response to overall growth is strong, more so than rural poverty. Mexico's example i s consistent with international evidence that urban poverty generally reacts more strongly to macroeconomic swings - whether positive or negative -than rural poverty. A potential explanation for higher sensitivity to growth i s that urban residents tend to be more closely connected to the overall economy than their rural counterparts, through deeper integration into the labor market, reliance on a cash economy, etc. These differences in elasticity largely disappear if one considers relevant sectors of growth only: rural poverty is highly elastic in comparison to agricultural growth. This i s driven by the fact that growth in general is more closely related to non-agricultural growth, however. The companion study on rural poverty finds, perhaps unsurprisingly, that rural poverty's response to agricultural growth i s pretty much the same as urban poverty's responsiveness to non-agricultural growth. Estimates o f poverty elasticity to growth for Latin America from Siaens and Wodon (2003) are around -1.3 for urban poverty and -0.7 for rural poverty. Work by Soloaga and Torres (2003) for Mexico shows that the relationship between urban poverty and "urban'' (non-agricultural) growth i s very similar to that betweenrural poverty and rural (agricultural) growth (Table 2-1). Inboth cases, extreme poverty i s muchmore sensitive to growth than overall poverty. l7This chapter was written by Marianne Fay and benefited from inputfrom Anna Wellenstein. 23 Table 2.1 Urban poverty is very responsiveto non-agricultural growth Total Urban Rural Extreme PoverQ Agricultural growth -1.3 n.s. -1.5 Non-agricultural growth -0.9 -1.6 n.s. Poverty Agricultural growth -0.6 n.s. -0.8 Non-agricultural growth -0.5 -0.7 n.s. Note: Elasticity i s defined as the percentagechange inpoverty headcount given a percentagechange in output. Source: adaptedfrom Table 4 in Soloaga and Torres (2003), as quotedinWorld Bank, 2004a. The distribution of growth among consumption quintiles can explain why the growth elasticity of urban poverty was high (Figure2.1). Infact, only the poorest quintile saw a statistically significant increase inconsumption. Figure2.1 The pattern of growth between2000 and 2002was pro-poor inurbanareas 6.0 5.1 0Consumptiongrowthperquintile 5.0 +M ean consumption growlh 4.0 3.0 2 2.0 1.0 0.0 -1.0 -2.0 -1.5 Poorest 2 3 4 Richest Quintiles Note: Only consumption growth for the poorest quintile i s statistically significant Source: WB staff calculations basedon ENIGH. Income vulnerability depends on household characteristics, notjustpoverty What household characteristics affect vulnerability to shocks? As mentioned above, the downside o f high growth elasticity i s that negative income shocks have a devastating impact on low income households. From the perspective o f policy interventions and, especially, the design o f safety nets, it i s important to know how certain household characteristics affect vulnerability to shocks and, inparticular, whether some low-income households experience much larger income cuts than others. Moreover, it i s important to know whether income shocks (to which households are always exposed, e.g. income loss due to illness or loss o f job) are distributeddifferently during turbulent times. Maloney, Bosch, Moreno and Tinajero (2003) look at the magnitude and distribution o f income shocks in Mexico's urban areas, comparing a period 24 o f normal times (1992 to end-1994) with one o f intense macroeconomic turbulence (end 1994 to end-l995).'* This analysis focuses on the entire distribution o f shocks and not only on average changes as it i s more common inthe literature. This i s important as it mightbe that on average a particular category o f household does not see large falls in income, but that some households within this category are disproportionately affected. Annex Table 5-A presents estimation results for the pre-crisis and crisis periods. Results relate the performance of different households to a base case o f a typical high-income household, in this case a household headed by a married middle-aged college-educated male working in the formal sector with a smaller family. What do these estimations tell us about income vulnerability for low-income groups? First, even during normaltimes different types of urbanhouseholdsexperiencevery different income variability depending on their characteristics.Households whose head was less educated generally had less variability in income than their more educated counterparts. Upwards, this may be because they have fewer opportunities for income growth. Downwards, these households may lack savings or other means to smooth consumption and so are forced to smooth income, possibly byworking more or putting additional workers into the labor market. Another important determinant of variability is the employment category: households whose headwas self-employed or employed inthe informal sector had higher income variability than the formally employed or employers. This finding i s consistent with the typical behavior o f a dynamic small-firm sector, where some firms do much better or much worse than the average salaried worker." Household heads with no earnings at the beginning o f the period generally saw higher median income growth as well as higher variance, suggesting that some o f them findjobs duringthe next period, while those that do not endup decidedly worse off." Age also seems to matter, with older workers experiencing less income growth and larger income drops than the base case. Single women - possibly approximating youth - appear to have both few possibilities for income gains and greater possibilities for lay-offs (income losses), while single mothers appeared not to be worse affected by shocks than the typical high-income family. Poor households did not experience more variability in income during the macro-crisis comparedwith normal times Secondly, and perhaps surprisingly, when looking at the crisis period, it appears that macroeconomic crisis did not substantially change the distribution of shocks for different types of households.Inparticular, it did not disproportionately increase the risks (as measured by income variability) faced by more vulnerable groups such as the self-employed and informally salaried, single parents families, young and old workers. This may be explained by the fact that Mexican households greatly benefited from increased remittances from abroad as a coping strategy (McKenzie 2003a). There are two important exceptions to this finding:households whose head earned no income at the beginning o f the period, which were decidedly worse off duringthe crisis probably This section relies on panel data from the ENEU. l9This higher variability may still be consistent with overall welfare gains, however, since entry into self- employment generally tends to be voluntary. However, recent research discussed inchapter 4 suggests that Mexico's recent expansion of informality may have been largely involuntary or induced by labor market distortions. 20When estimated on 2000-2002 data, very similar results were obtained, suggesting that these findings are robust. 25 because the increased difficulty o f securing a job then; and the less-educated who in fact did better than in normal periods, experiencing smaller negative income shocks and greater positive income shocks. This suggests an equalizing effect o f the crisis which i s consistent with the finding for Mexico and for other Latin American countries that the premiumto higher education tends to fall duringmacroeconomic crisis (see chapter 4 on labor markets for more details).21 It appears that urban and rural households have fairly similar patterns of income variability. One exception is that in rural areas workers with no education seem to experience deeper negative shocks than their urban counterparts. At the same time, rural households endured on average lesser shocks and the differential would likely be even higher if their greater reliance on self-consumption than urban households was factored in. In sum, low-income families were not disproportionately hit by the macroeconomic shock, exception made of household heads with no income. Their susceptibility to income shocks did not change markedly during crisis compared with normal times. Some caveats are needed, however: First, low-income households may be less able than others at smoothing consumption to avoid drastic swings in welfare. In particular, very poor households may lack savings or other insurance mechanisms against, for example, loss in employment.22 As a result, even ifthey face smaller income shocks these could translate into higher consumption shocks than for other groups. Also, poor households, for lack o f other means, might adopt coping strategies which undermine their long-term well-being. Much attention has been given to taking children out o f school issue -but households can also adopt other strategies with possible long-term deleterious effects, such as postponing spending on primary healthcare (McKenzie 2003b). Second, low-income households will be much more severely affected in absolute terms even if they experience proportionally smaller changes in income and especially if they live close to the minimum subsistence level. The tripling o f the number of extreme poor in urban areas during the peso crisis is in fact evidence of how income falls for low-income groups can push a large number o f people under an absolute minimumwelfare standard. Third, macroeconomic crisis may have widely differing causes, consequences and policy responses, and their impact across the distribution o f households can therefore vary considerably. Analysis based on the particular trajectory o f the peso crisis, while informative, will therefore not necessarily apply to all macroeconomic shocks. Labor markets, source of vulnerability but also a key copingmechanism Idiosyncraticshocks are the major source of risk for households.Work done for this report looked at the sources o f vulnerability o f the urban poor as well as some o f their coping mechanisms, contrasting them with those o f the rural poor (Skoufias, 2004). One key finding i s 21See e.g. World Bank (2004a) and Schady and Shchez-Paramo (2003). 22Indeed, an analysis of rural consumption vulnerability showed that younger and better educated workers - who generally experienced larger income swings - tended to have better mechanisms to smooth consumption than other groups. 26 that the more common types o f shocks in urban areas are those transmitted through the labor markets - instead the rural poor are more vulnerable to weather-related shocks. Interestingly however, in both rural and urban areas, idiosyncratic shocks - those specific to a particular household, such as illhealth or loss o f employment- are the major source o frisk for households. Householdsresort to a variety of risk mitigating (ex ante) and risk-coping strategies (ex post), due to these large idiosyncratic risks combined with the absence of formal insurance markets and credit constraints. These strategies include the accumulation o f assets, income diversification, sending women and children to work, withdrawing children from school, migration, marriage and other informal risk-sharing arrangements. Understanding the coping strategies that households adopt and designing social programs that help them refrain from using the coping strategies that are detrimental to their future welfare should be among the primary objectives o f a proper social protection policy aimed at reducing household vulnerability as well as poverty. A key finding in this matter is that urban households have very effective income- smoothing strategies, mostly through an "added worker" effect, whereby the spouse, and sometime the children, enter the labor market if householdhead becomes unemployed.This added worker effect i s specific to urban areas and i s not found inrural areas, with the implication that labor markets' dual role as a source o f risk and as a form o f reducing risk i s an urban phenomenon. Note however that how much labor markets can reduce household vulnerability to shocks depends on the household's demographic composition (whether it includes a non-working spouse or teenage descendent). In addition, labor markets may be less effective at protecting households from specific shocks (sickness, death, divorce) than from wage or employment shocks. PUBLIC PROGRAMS FOR THE URBANPOOR Many public programs are relevant for the urban poor - several of which are discussed in the context of the analysis of labor markets (chapter 4) and asset building strategies (chapter 5). However, there are two that are most directly relevant and that are discussed here: Oportunidades,the government's flagship anti-poverty program that was recently rolled out to cities and i s confronting some challenges in adapting to urban environment, and Hhbitat, a new program that offers urban upgrading for poor neighborhoods and i s at the heart o f government's strategy to fight urban poverty. Oportunidades Human Development Program Opportunities (Programa de Desarrollo Humano, Oportunidades)was the first large scale conditional cash transfer program in Latin America and is one of most successful according to several external evaluations. Its main objectives are to promote the development capacity o f extremely poor households inthe areas o f education, health and nutrition. Monetary benefits are conditioned on human capital investment by its beneficiary families (school enrollment o f childredyouth and regular attendance at health clinics). Oportunidadesonly targeted the rural poor until a gradual roll-out to urban areas begun in 2002. The main challenges of this expansion into urban areas relate to the targeting strategy and to adapting the programto urbanneeds. 27 Oportunidades retains the basic principle of geographic and household targeting usedin ruralareas, but adjustedthe mechanismsto recognizethe greater difficultyand cost of targeting in urban areas. Census information i s first used to identify poor neighborhoods where modules are set up and advertising is carried out to let households know they can apply for benefits. Applicants can also come from outside the neighborhood. Individuals arriving at the module are administered a questionnaire about their socio-economic conditions, the answers to which are immediately entered into a computer to determine the household's eligibility. Households deemed eligible then receive a verification visit generally within the following two weeks. A recent evaluation estimates that this approach captures about 65 percent o f eligible households, as opposed to 77 percent inthe first year o f the urban roll-out when all households in eligible areas were administered a questionnaire. On the other hand, the cost o f the module approach i s about one third less than that o f the administration o f the questionnaire to all households inthe poor neighborhood (Gutikrrez, Bertozzi and Gertler, 2003). Evaluations suggest that the impact is smaller in urbanthan in rural areas. As ofthe end of2003 the package ofbenefits offeredwas the same inurban andrural areas, but evaluations show rather different results. In particular, the impact on school enrollments, graduation and dropout rates are much smaller than inrural areas (Parker and 2004). This i s more likely to be due to the fact that opportunity costs are higher in urban areas than in differences in enrollment between rural and urban poor - which, as discussed in chapter 1, are remarkably similar. In addition, street violence and distance from schools makes it necessary for children and youth to take a bus to school, the cost o f which i s reported to absorb as much as 80 percent o f the school grant (Escobar Latapi and Gonzalez de la Rocha, 2004). Unexpectedly, one of the most obvious and strongest impacts of Oportunidades on urbanhouseholds has been home improvements, which occurred to a much greater degree among beneficiarythan non-beneficiary households - improvements include payments for regularizationor acquisition of infrastructureservices or improved construction materials. The evaluation study argues that this i s becausethe irregular status o f ahome or its poor quality i s perceived as an obstacle to overcome or a source o f vulnerability to a much greater extent than in rural settings (Escobar Latapi and Gonzalez de la Rocha, 2004). The need to adapt the requirements imposed on participantsto urbanwork reality has been a particulardifficulty encountered duringthe urbanexpansion of the program.In particular, a number o f working mothers did notjoin or dropped out o f the program because they cannot attend medical appointments or educationaltalks which are heldduringworking hours. An additional problem i s saturation o f clinics in poor urban areas which poses a significant problem since regular medical check-ups are a program requirement (Escobar Latapi and Gonzalez de la Rocha, 2004). In addition, program administrators mention the need to adjust the content o f the educational talks on public health issues the urban reality - coping with drug use and street violence rather than boiling water etc. Habitat Until recently slum upgrading was occurring - if at all - without planning, neighborhoodparticipation, sequencingof investmentsor integrationof physical and social improvements.The fragmentation of these functions within different areas of SEDESOL - the government ministry in charge o f urban development - also impededan integrated approach that responds more precisely to the social physical and economic needs o f the poor (e.g. actions that helphouseholds' earning capacity such as more reliable access roads or day care centers). 28 Recognizingthis problem, the government began a series of diagnostic and strategic studies in2000 and 2001 which lead SEDESOL to launching its Hhbitat program in2002. The program's goal i s to help improve living conditions in marginal urban neighborhoods by addressing both infrastructure and social needs, as well as improve the overall quality o f cities. Hhbitat supplements local resources and aims to create incentives for increased and more effective investment by local government and beneficiaries: it channels federal funds (under Ram0 20) as matching grants to municipalities for programs benefiting families earning up to 1 minimum wage per family member. The grants are approximately US$ 500 per family and are meant to service as an incentive for more effective municipal and beneficiary expenditures inthis area, rather than financing for the total costs of the a~tivities.~~ Municipalities access Hhbitat through states and can request funding for the following activities: i)neighborhoodupgrading through the provision o fbasic infrastructure such as water and sanitation network expansion, paving o f roads, sanitation, public lighting, etc.; ii) access to land through land purchasing reserves; iii)prevention o f natural disasters through risk mapping and planning and investment for vulnerability reduction; iv) community development through community facilities, day-care centers for working mothers, job training among other activities; v) support to women head o f households through job training, child care and health education; vi) support to local development agencies for training and the undertaking o f strategic studies and investments in partnerships with communities and the private sector; and vii) urban equipment and improvement including protection and restoration of historic neighborhoods. In 2004, SEDESOL added the option o fusingHabitat support for property titling. Hhbitat includes several design innovations to improve targeting and program impact: Capacig. Hhbitat aims to create local capacity among municipal authorities, Non- governmental organizations (NGOs) and community based organizations (CBOs) related agencies. A capacity buildingprogram runbySEDESOL has been initiated. Poverg Targeting and Eligibility Criteria. Recent progress by SEDESOL in poverty mapping i s crucial to furthering our understanding o f the challenges faced by the urban poor and in better targeting public support. SEDESOL has developed detailed geo-referenced poverty data down to the block level for eligible municipalities that allows for accurately allocating Ram0 20 funds. Eligibility criteria for priority neighborhoods for investment include at least 50 percent o f the families earning less than the asset poverty line level (or 30 percent in special cases as specified in Program's regulations), density o f occupation (80 percent o f lots occupied at least 35 households per hectare o f land) and deficits in urban infrastructure, equipment and services. Multiyear program of improvement. The program supports the development and execution o f multi-year plans for neighborhood improvement, helping to overcome some o f transaction costs and capacity limitations o f progressive investment. In the absence o f such a strategy, investments tend to be driven by political and social pressures and often with a shorter-term perspective. 23The $500 grant per family grant amount does not apply to Hibitat land reserve and urban development agency activities. Instead, these activities are financed based on a matching basis, with maximum contribution from Hibitat of approximately US$250thousand. 29 Labor market particQation. Habitat's efforts in terms o f training, physical access to jobs (e.g. paving roads) and supporting social services (e.g. child care which i s crucial for women) may go a long way to overcoming the barriers faced by the urban poor interms o f access to quality employment. T o date, Habitat has shown a strong performance, rapidly moving to scale with support to 400 barrios in 2003 totaling MX$ 781 million. With the focus on working on implementation issues in 2004, 760 barrios were targeted with a total funding increased to M X $ 1.730 million. Although an important advance, Habitat requires further action for the program to fulfill its potential. The program design stipulates important procedures essential to its performance such as community participation, neighborhood investmentplans and an appropriate level and sequencing o f investment in each community. However, rapid expansion o f the program in its first years and the urgency of meetinginvestmentgoals have sometimes taken precedence over these other aspects. Specific areas for further work include the following: Possible addition of assistance for job search. Habitat's efforts in Superacibn de Pobreza and Mujeres Jefes de Hogares include child care and training, all of which should facilitate the poor's participation in labor markets. An additional possibility could be for the program to include labor intermediation services, such as job-search programs, to increase the access o f poor, low-skilled workers to labor market information. As discussed in chapter 4, current programs (Chambatel, Chambanet, Sistema de Apoyo Econbmico a Buscadores de Empleo -SAEBE-) tend to favor higher skilledworkers or the unemployedwho usedto work inthe formal sector. Any effort from Habitat to include job-search program should o f course be coordinated with existing programs of Labor Ministry(Secretaria de Trabajoy Previsibn Social, STPS). Greater efforts to provide multi-year support. The amount o f annual federal subsidy limits the scope of the project. The US$500 per family subsidy, matched by local funds for a total of US$l,OOO per family is sufficient for investment in minor infrastructure in a spatially well structured neighborhood. It i s not sufficient, however, for major restructuring (relocation, stabilization works) for neighborhoods located in high risk areas and o f poor layout with insufficient public access for infrastructure and right o f ways and overcrowding. A recent review o f upgrading programs inBrazil found an average cost per family closer to US$3,400. Strategic use of localfunds. According to Habitat regulations, the federal subsidy can be dedicated to the same families for a maximum o f three years, in which case the total amount of resources would be sufficient to address most upgrading needs. However, given the high turnover in municipal staff, and the lack o f poverty and upgrading planning, there i s likely to be a tendency for funds to be spread throughout the city rather than focused on upgrading particular neighborhoods over multiple years. In order to fulfill the programs goals o f promoting more strategic use o f local funds, rather than discrete one off investments, subsidy amounts should be increased or greater weight placed on investinginthe same neighborhood over several years. Increase emphasis on medium term upgrading strategies. A major benefit o f neighborhood upgrading i s how the process supports the poor's own approach o f gradually improving their neighborhood. To realize the impact o f the public funds 30 dedicated to upgrading, the gradual investment should fit within a medium term strategy for that neighborhood and an overall poverty reduction and urban development strategy for the city. Moreover, the operating rules for Hhbitat state that a neighborhood plan and participation inprevious years (but not more than the three year limit) are factors in classifying neighborhoods as `priority' for investment. However, the weight of this factor inresource assignment i s not clear.24 Budget figures from 2003 and 2004 indicate that Habitat may not be spending enough on strategies. During the first two years o f program implementation, approximately 2 percent o f resources were dedicated to Urban Agency activities, for which poverty strategies and slum upgrading plans are only one o f several eligible activities. Summingthis 2 percent with SEDESOL administrative costs o f 13 percent, the program funds dedicated to software i s low comparedto other programs of lesser scope and geographic scale in the region.25These experiences point to the need to increase the funds dedicated to neighborhood plans and city poverty strategies and to place greater weight on the process o f approving and prioritizing investments. Review capacity-building plans and/or bring in external partners for training. Municipalities are already investinginneighborhood upgrading via funds from Ram0 33 (see below). One o fkey goals o fHhbitat should therefore be to encourage a better use o f Ram0 33 investment funds - through a medium term vision that includes a better link to social needs o fneighborhoods andto the overall needs and development potential o f a city. This will only occur ifHhbitat can help boost the capacity o f local governments to foster greater poverty and strategic planning. The Hhbitat operating rules state that National Institute for Social Development (Instituto Nacional de Desarrollo Social, INDESOL)will train SEDESOL state level offices (delegaciones), which inturn will train municipalities and communities. Giventhe level o f staffing in the delegaciones and the fact that they must support all SEDESOL programs to the state, they are unlikely to have the necessary number o f staff with the appropriate skills to foster a major change in local capacity or practice in urban upgrading. External partners - INDESOL could be a good candidate- may therefore need to be brought infor the capacity-building objective. In addition, Hhbitat is a geographicallytargeted instrument which therefore cannot help the poor that do not live in marginal neighborhoods.As discussed in chapter 1, cities are very heterogeneous with large differences in welfare also at neighborhood level. It is unclear what share o f the moderatepoor live outside neighborhoods eligible for Hhbitat, but it i s likely to be substantial.26 One first step might be to estimate this share. Another is to examine the possibility o fnon-residents benefitingfrom the programs offered (such as training, job placement, and childcare). 24SeeHibitat Reglas de Operacion, pp. 31-32. 25InSanJose, CostaRica, the costofFundacionPromotoradeVivienda's(FUPROVI) services asapercentageof the actual costs of the houses it helps people buildrangesbetween 16 and 23 percent, depending on location, size of the units and complexity of the project. Ina similar project in Sao Paulo, Brazil, the cost of project software was about 20 percent for upgrading projects that cost $2,500 per households. 26Work done inthe context of Oportunidades, it was found that 77 percent of the urbanextreme poor (pobreza de alimentacion) live in "marginal" neighborhoods. No such data was available for moderate poverty (pobreza patrimonial) that determines eligibility for Hibitat. 31 Overall, while the approach of Habitat is promising, careful evaluation and follow- up on progress and impact will be important to further strengthening the program. At this time, there is little data available on the outcomes or development impact of the program. Developing a comprehensive evaluation system to track implementation and impact issues should be a priority. Finally, while Habitat's support for land reserves provides a useful short term response, broader urban land strategy is needed. This i s discussed inmore details in chapter 5 inthe context ofthe broader debate onhousingpolicy for low-income housing. DECENTRALIZATIONOF SOCIAL INFRASTRUCTURE SPENDINGUNDER RAM033- FAIS The Contributions Fund for Social Infrastructure (Fondo de Aportaciones para la Infraestructura Social, FAIS) provides federal funding to the states to improve the basic social infrastructure o f marginalized areas. A key feature in Mexico i s that most social service delivery responsibilities are decentralizedto the state level - for example, close to 74% o f basic education spendingand 75% o fpublic health spendingfor the non-insuredpopulation is decentralized. Most o f the resources for physical capital o f poor areas are distributed through FAIS under the budgetary branch Ram0 33. 27 This i s the result of a series o f decentralization reforms initiated at the beginning o f Zedillo administration in 1995 under the title o f New Federalism (Nuevo Federalismo). In addition to various fiscal reforms providing new sources o f revenue for the states, these included the creation in 1998 o f a new modality o f federal transfers implemented through the budgetary branch 33, Ram0 33 (R33), that comprises seven funds2*.O f these, the education fund (FAEB) and health fundtogether absorb some 76 percent o f the total R33 budget, while FAIS,the social infrastructure fund, accounts for about 9 percent o fthe total envelope. F'S has been considered an important improvement over previous methods. FAIS funds are intended to strengthen basic infrastructure: clean water, sewerage, drain, municipal urbanization, electricity for rural and poor urban areas, basic health and education infrastructure, housing, rural roads and rural productive infrastructure. The allocation o f resources i s partly based on a poverty-weighted formula, using indicators such as illiteracy, education, drainage, electricity, housing, and income as eligibility criteria (Scott, 2000). However, FAIS still faces important challenges in terms of transparency of how resources are assigned, in terms of a compact that promotes accountability from operators to beneficiaries and in terms of developing spatially neutral targeting between urban and rural areas. Thus, some important issues relating to transparency and incentives remain in evaluating whether FAIS constitutes areal improvement over its predecessor PRONASOL: Not all resources within FAIS have been distributed through poverty formulas, though part o f the fund has been distributed on the basis o f a so-called "equity 27See World Bank (2004a). 28 Fondo de Aportaciones para la Educacion Bbsica y Normal (FAEB); Fondo de Aportaciones para 10s Servicios de Salud (FASSA); Fondo de Aportaciones para la Infraestructura Social (FAIS): Fondo para la Infraestructura Social Estatal (FISE) and Fondo para la Infraestructura Social Municipal (FISM); Fondo de Aportaciones para el Fortalecimiento Municipal (FORTAMUN); Fondo de Aportaciones Mtiltiples (FAM); Fondo de Aportaciones para la Educacion Tecnologicay de Adultos (FAETA); and Fondo de Aportaciones para la Seguridad Ptiblica de 10s Estados y el Distrito Federal (FASP). 32 criteria," guaranteeing every state a fixed and equal proportion of the fund, independently of population size or poverty mass. This can be interpreted as a transitional mechanism, which responds to political constraints by protecting the smaller and richer states (and municipalities within these states) from a sudden loss o f resources. There is limited transparency in the use of funds at sub-national level. The decentralization process was not accompanied by mechanism to monitor the effectiveness o f the allocated resources. As a result, there i s no clear accountability at different levels o f government. Recent efforts to improve transparency at the federal level (the Federal Law o f Transparency and Access to the Governmental Public Informationhave yet to be fully implementedat sub-national levels. Resource trickle-down from state to munic@ality level may be flawed. States are required to distribute only the municipal component o f F'IS (FISM) through a federal formula called Municipal Mass o f Lacks (Masa Carencial Munic@al) or through a simpler formula that i s utilized when household level data i s not available. The MCM formula has a number of important technical and political limitations. Some technical limitations are associated with the weights and variables chosen to structure the formula.29 There are adverse incentives inherent in the allocation process. If a state performs better in terms o f poverty it gets fewer resources inthe next budget cycle leading to adverse incentives for the improvement o f state programs. The targeting criteria are biased against urban areas. The criteria -the proportion o f the population that i s illiterate, lives in houses without sanitary drain, does not have electricity and who earns less than two minimum salaries - are systematically lower inurban areas. Electricity connections inurban slums are usually illegal and as such are not a good measure o f an individual's needs or living conditions. Similarly, access to drainage in urban and rural areas measure different things - urban neighborhoods that do not have piped sewerage, or at least some form o f sanitary drainage are in seriously dismal conditions. Intervention i s justified on public health criteria alone. On the other hand, having pipedsewerage i s not generally appropriate inlow density areas (less than 25 persons per hectare) where on-site systems such as septic tanks are better. More generally, when comparing two areas without sanitary drainage or sewerage priority should be given to the area with greater population density. As to illiteracy rate it is always much higher in rural areas, in part because the illiterate are less likely to migrate. Better indicators that capture well the needs and living conditions o f a population group regardless o f whether they reside inurban or rural areas are: access to pipedwater, quality o f houses (durable material on roof or walls), and for education infrastructure, school rooms to school-age children ratios. M o r e work i s needed to assess the allocationand uses of FAIS at the municipallevel in order to evaluatewhether truly represents an improvementover PRONASOL.This forms part o f the third phase o f the World Bank programmatic work on poverty in Mexico, which 29For a more extensive analysis see Scott, 2000. For example, the author argues that there is ambiguity inthe construction of the weights in the formula that are meant to reflect the costs to families of accessing the service. 33 concentrates on service delivery in the context o f decentralization. More specifically this work will concentrate on (i)incentives and accountability in the use o f resources, (ii) incentives generated by the formula, and (iii) allocation o f resources at the local level. SUMMARY AND CONCLUSIONS Urban poverty i s more responsive to growth -the flip side of this i s that the urban poor are very vulnerable to macroeconomic crisis. Household characteristics condition income vulnerability- inparticular self-employment or employment inthe informalsector and ayounger household head are associated with greater variations in income. The recent macroeconomic crisis did not change the risk distribution for the poor, although it certainly led to a massive increase inurban extreme poor number, which tripled between 1994 and 1996. The analysis of vulnerability highlighted the importance of large idiosyncratic risks and the crucial role of labor markets in household coping strategies. Even though it i s through labor markets ('job losses, pay cuts) that poor urban households are most vulnerable to macro-shocks. The main implication of this chapter is that growth is key. Urban poverty will decrease if Mexico grows and jobs are created in urban centers. But safety nets are needed. As such the recent expansion o f Oportunidades to urban areas i s opportune. The program i s strugglingsomewhat with the needto adapt to the urban reality, although evaluations suggest it is having a positive, although somewhat unexpected, effect. The key challenges confronting it concerns increasing coverage while keeping targeting costs to a reasonable share o f overall program costs; and adapting co-responsibilities to the schedule of urban working parents. In addition, the recent creation of Hhbitat, a program focused on urban upgrading with a careful emphasis on social issues is a very positive development. The program is innovative, although it could benefit from further action to enhance its impact. 34 3. PLACES OFTHE POOR: THE CASE OFMEXICAN POOR BARRIOS3' Many poor people are disadvantaged and endangered by the places and physical conditions where they live and work. ... Most poor people canfind only `places of the poor '' in which to live. Theseplaces keep thempoor. [Voices o f the Poor] The quote above from Voices of the Poor draws attention to the fact that where the poor live shapes the opportunities they face and the challengesthey encounter in daily life. This chapter looks more closely at some important features o f the "places o f the poor" in urban areas and provides insights on how these affect poor people's lives and coping strategies. Places o f the poor -which we will also refer to as poor barrios or colonias to use a common terminology in Mexico- are places where the majority of inhabitants are poor though by no means they exhaust the areas where poor people live. Looking at life in these barrios gives us insights on how disaggregated urban categories are useful to understand the challenges that poor face in urban contexts. Within cities location makes a difference above and beyond individual characteristics, though individuals and their characteristics, alone or through their interactions, contribute to shape the environment they live in. In particular, we will emphasize the role played by a special set of non-market relationships between individuals - their social interactions - in shaping that environment. The social capital literature emphasizes the role o f social interactions as an asset on which individuals can draw upon, but in many other areas the potential for social interactions, information flows and the importance o f civic attitudes have been highlighted. This chapter analyzes some of the salient features of the places of the poor in cities, the differences among them and the inequalities within them by using the first round of a new survey collected by SEDESOL in 2002. The survey was fielded in one poor barrio in the capital cities o f every Mexican state, exception made o f the nation's capital. The barrios were identified on the basis o f a lack o f basic infrastructure and amenities as well as o f the characteristics o f people who live inthem. All are areas with indices o f marginalization o f 5 and 4, although there are substantial differences among them: the extreme poverty rate, for example, varies between 18 percent and 55 percent for an average o f 36 percent (in contrast the overall 30This chapter was written by Caterina Ruggeri Laderchi. The author is grateful to Michael Woolcock and Berk Ozler for their comments. 35 urban extreme poverty rate i s 11percent); andthe proportion o fhouses with access to pipedwater varies from 17 percentto 93 percent.31 The survey providesa unique opportunityto lookin depth at the livelihoods of poor in urbanareas across the country.At the same time, however, the sample designmakes itmore representative o f poor places across the country, rather than o f poor people themselves. This can complicate the comparison with quantitative evidence offered by the census data discussed in chapter 1. WHY "PLACES OF THE POOR" MATTER Interest in disaggregated urban poverty analysis stems from two different sets of concerns. The first is that urban subunits and how they are identified become relevant for policymaking, either because of decentralized decision-making and/or because o f targeted interventions. Relatedto this, the interest in smaller urban areas i s essentially descriptive - certain neighborhoods appear to be correlated with high concentration o f poverty independently o f why this is the case, though differentcausal explanations mightbe put forward to explain the poverty o f certain areas. This seems to be behindMexico's focus on small area characteristics highlightedby the computation o f the marginacibn index, recently extended to measuring marginalization at the local level so as to be able to classify census tracts/localities "according to global impact o f the shortfalls suffered by the population residing there." (p.11, CONAPO 2002) A composite indicator o f marginalization (comprising measures o f "some forms, intensity and demographic and territorial implications" o f marginalization) had already been computed for Mexican states and municipalities since 1990. A new index has been calculated to better reflect barrio level characteristic^.^^ A different motivation is that some inherent characteristics of the location may shape opportunities and constraints of the poor.33Literature rooted in economics of growth has hypothesized the existence o f geographical poverty traps, based on differences in technologies across space or alternatively in initial differences in endowments o f local public goods if these offer increasing returns (Ravallion and Jalan 1999). Households living in poor areas enjoy lower returns on their factors of production because they are in poor areas. This emphasis on different endowments o f local public goods in the case o f urban areas can focus for 31Selection was a two-step process. First, barrios with marginalization index of 4 or 5 (the highest) were identified in each of the state capitals (except for Mexico City) based on SEDESOL's poverty map. Second, local authorities were selected for their views as to which of the eligible barrios should be selected and what exactly were the boundaries of the barrio (since that includes more than one census tract, the basis of the SEDESOLpoverty map). 32The ten indicators on which the index is based relate to different aspects of health (access to health services, child mortality), education (school attendance, lack of secondary), housing (access to water, access to sanitation, material of the roof, lack of a refrigerator, overcrowding), incomes (share of population on with low payingjobs), and gender (rate of teen-age pregnancy). 33The U S literature has been very concerned with such effects. "School delinquency and dropout, teen pregnancy, out-of-wedlock childbirth, violent crime and drug abuse are magnified inneighborhoods where the majority of residents are poor and, increasingly, minorities." (Goetz 2003). These effects, together with those on access to jobs and employability, often compounded by stigma (see also Perlman, 2003) have led to policy efforts to "deconcentrate" poverty. 36 example on the availability o f local infrastructure (at least if congestion costs are not an issue).34 These concerns are also reinforcedby evidence, for example, on the role played by transport costs as a barrier to access tojobs and on the problems o f network provided services.35 A different line of argument has been built on evidence from sociology and social psychology to focus on the concept of group membership, of which inhabiting a poor neighborhoodis one example (Durlauf 2001). Inthis view, the spatial concentration o f poverty appears to result in specific cultural and social features which, in turn, can often be linkedto the social make-up o f the neighborhood itself (e.g. the educational levels o f adults in the community or its ethnic composition). Durlauf lists the main mechanisms through which social composition o f a group (and more specifically o f a neighborhood) can have on individual behavior: (1) peer group effects and role model effects by which choices that some members make -or made- influence the valuation o f those choices by others; (2) social learning by which information on some o f the choices available i s derived from the experiences o f some group members; and (3) social complementarities (group members' outcomes are directly affected by each other's outcomes). All these types o f interactions fit the typical description o f an e~ternality.~~ Similarly related to the importance of socialinteractions is the idea of social capital. Such literature builds on the idea that long lasting patterns o f relations which are brought about by social interaction constitute a form of "capital" available to individual actors. What defines the asset value o f such resource i s its productive use, while it also shows public good like features o f not being appropriable by a single individual. Around this concept a rich literature has developed, documenting a variety o f positive impacts such as the association between quality o f public services and participation in "horizontal organizations" (found in the Italian case by Putnam 1993), and the positive impact o f trust and civic cooperation on economic performance (Knack and Keefer 1997).As the literature adopts a variety o f indicators to capture social capital (ranging from to the individual or to the community, to the resources which are shared through social relations or to the relations themselves, etc.) Grootaert et al. (2003) suggest seeing them as relating to either structural characteristics o f social capital in a given context (this includes cognitive aspects such as individual perceptions o f trust and solidarity, and structural aspects like the existence o f groups and networks) or to its operational aspects (this relates to evidence on collective action and cooperation). In the case of Mexico, the qualitative studies of the barrios which complement the quantitative data gathering point to the importance of bridging the apparent divide between local endowments and social relations. The physical characteristics o f neighborhoods such as the type o f infrastructure available, regular lightingin the streets etc., seems to influence social interactions, for example, by influencing perceptions o f security, participation in communal activities etc. This complements common findings that more closely knit and stable communities are also more conducive to more investment in common resources andpublic goods. 34Similar arguments could however be developed, for example, with an emphasis on local institutions 35E.g. detailed discussion of the case of Bogota inBaker and Lall, 2003 36Depending on the characteristics identifying groups, membership may or may notbe endogenous. Gender for example, identify groups whose membership is exogenous (arguably ethnicity falls into a similar category), unlike for example location of residence. Even in that case, however, the literature on land markets points out how market mediated choices might be strongly constrained, so that individuals are "sorted" into homogeneous neighborhoods (e.g. Alonso, 1964and Muth, 1969). 37 POOR PEOPLE OR POOR PLACES? Poor people choose to locate in poor areas. Are people poor because they live in poor area or are deprived places because poor people choose to locate there, or both? This chapter argues that location matters for determiningpoverty. But as the literature on growth and poverty traps suggests, the geographical concentration o f poverty i s an endogenous process. Poor people choose to locate inpoor areas. They do so because living there i s affordable to them and possibly to draw on social capital - they know people who live there and can take advantage of existing networks. At the same time, the concentration of poor people has some negative externalities which reinforces and reproduces poverty over time. These effects stretch well beyond commuting times and rental values. For example, Durlauf (2001) suggests that the main effects which the social composition o f a neighborhood can have on individual behavior can be seen as: peer group effects when individual choices are influenced by the choices o f others; role model effects when the preferences o f older members of a neighborhood influence younger members preferences; social learning when information on some o f the choices available i s derived from the experiences o f others; and social complementarities when group members outcomes are directly affected by each other's outcomes. All these types o f interactions fit the typical description o f an externality and suggest that policies encouraging socio-economic mixing might have important "social multipliers." Neighborhood effects might also be compounded by stigma, affecting poor people's access to jobs and other forms of discrimination. For example, inRio de Janeiro, 85 percent o f those sampled in a slum perceived discrimination against anyone living in a favela (Perlman, 2003). A recent study found that residents of Rio'sfavelas earn between 10% and 47% less than other similar individuals (in terms o f education, age, and gender) in similar occupations, but living in other neighborhoods (Cardoso, Elias and Pero, 2003). One the other hand, a study of public housing participants in Toronto suggestedlittle or no influence o f neighborhood quality on resident's long-run labor market outcomes, while family differences appeared to matter a great deal. Inwhat follows we will explore two aspects of urbanpoverty inMexico. The first will beto look at inequality inurban areas andhow it i s affected by barrio characteristics. The second will be a more specific focus on social interactions within the barrio as one o f the specific "structural" characteristics, through which the urban poor themselves contribute to shaping the barrio environment. A CLOSER LOOKAT URBANINEQUALITIES It has been found elsewhere (Fay and Ruggeri Laderchi, 2004) that in Mexico, inequality in urban areas i s lower than in rural ones. The availability o f barrio level data allows us however to refine this insight and draw attention to how varied the urban picture i s and how important differences are within urban areas. Inequalities within barrios Inequality in poor barrios varies across different aspects: it i s relatively smaller in the income dimension, but high in terms of assets. Living in a poor barrio in Mexico might 38 mean very differentthings and inequality varies across different dimensions. Interms o f income, for example, the Gini coefficient for household per capita income varies between a minimum of .24 to a maximum o f 30 implyingthat inequality i s fairly similar in all these barrios and lower - actually than for urban areas as a whole.37 On the other hand, greater differences characterize other non-monetary indicators o f well-being such as years o f schooling o f the household head or a composite indicator o f household ownership o f consumer durables and assets.38In the case o f education for example, the Gini varies between .14 and .31. In order to compare these findings with the differences in terms of durables and assets between more and less equal barrios, we calculated relative inequality measures for all o f them and found them almost 9 times higher for the asset index than for income.39 Inequalityis much higher within these poor barrios than between them, pointingto the great heterogeneity that exists even within poor neighborhoods. A more formal way of pointing out the differences that characterize the lives o f poor people in different barrios i s to decompose overall inequality in the sample in its between barrios and within barrios components. As shown inTable 3-1 below income inequality within the barrio i s at least 20 times higher in a barrio than across barrios4', when decomposing the GE(2) index (an index of the GeneralizedEnthropy family, where the index denotes a greater sensitivity to inequality at the top o f the distribution). 37 Income captures consumption opportunities or can be seen as a proxy for consumption levels and hence welfare. As adopted in this paper, it is not strictly comparable with other published national figures as income information inthe survey is not fully satisfactory. Shortcomings include the non-comprehensiveness of questionnaire (only income from primary jobs is recorded; the contribution of government programs to household income cannot be estimated due among other reasons to the impossibility of isolating the credit from the subsidy component of some programs; information on food transfers programs is not detailed enough), and the lack of price information at the barrio level. Monetary values are therefore transformed in Mexico City constant prices using price indexes from the capital of the state. In order to identify the concentration of poverty at the barrio level, the 40thpercentile for the aggregate sample has been adopted as a country level poverty line. While not comparable with other poverty lines in the literature, it provides a useful way of identifying the barrios in which the "tail" of the income distribution lies. Note also that in order to arrive at some lower bound, but supposedly believable estimate, of inequality of the barrio we have first discarded extreme values. We have then further restricted the analysis to the 8 intermediate deciles to further disregard the tails. 38 The asset index, created following Filmer and Pritchett (2001), can be seen as either a measure of consumption potential ina multiperiod setting where shocks are possible (it amounts therefore to a measure of household resilience) or as a proxy for overall "household economic status" (Filmer and Pritchett 2001). In our case the index was obtained by principal component analysis of a set of 15 variables including household durable goods, ownership of financial and real assets and housing characteristics. 39 Relative inequality measures such as the standard deviation are needed as it is not possible to compute a Gini coefficient for the asset index as by construction it assumes negative as well as positive values (McKenzie 2003). 40See note n.33. 39 Table 3.1 Inequality is greater within poor neighborhoodsthan across them Income Years of schoolingof householdhead Total inequality (GE2) 0.123 0.128 Withinbarrio 0.118 0.123 Betweenbarrios 0.006 0.005 Source: WB staff calculations basedon SEDESOL survey Pockets ofpoverty One important aspect of inequality within poor areas is that it often translates discontinuities into pockets of poverty. Attention to this phenomenon has been brought by the literature on urbanpoverty inthe U S and its emphasis on the concentration o f poverty. Inorder to explore how evenly poverty i s distributed across the barrio, we looked at differences in poverty incidence indifferentparts o f the barrio. Pockets of extreme poverty exist in the poor barrios. Figure 3.1 below documents the variation o f headcounts in different census tracts part o f the same barrios.41 The variation i s captured by the spread between the minimum and maximum headcounts registered in different census tracts in the barrio.42Panel (a) adopts the extreme poverty line (food based poverty line) and shows that headcounts o f around 40-45 percent are compatible with differences o f as much as 20 percentage points in the headcounts o f relatively better off and relatively worst off census tracts. Panel (b) replicates the analysis with the highestpoverty line adopted inMexico and shows that while incidence i s higher, spreads in general are lower as the majority o f the population in each barrio i s considered poor. Some barrios are more prone than others to having pockets of poverty, but the location of pockets of extreme poverty differs depending on the degree of spatial disaggregation. Three interestingpoints are made by these graphics: one i s that in some barrios more than others pockets o f poverty exist. These are the barrios in Figure 3.1.a with mid-range headcounts but high spreads.43 Second, depending on where the poverty line i s drawn, the diagnosis on the extent o f the segmentation o f the barrios in poor and non-poor areas changes. Third, and directly related to the general point on the heterogeneity of urban areas and the need for disaggregated assessments o f poverty, depending on the level o f disaggregation adopted (barrio level or census tract level in this cases) different assessments o f the geographic concentration o fpoverty and its incidence emerge. 41 Because of the nature of our data, for each barrio we focused on differences in census tracts containing at least 20 percent of the population in order to arrive at some plausible estimate of the census tract level variables. This way of proceeding is different from the one adopted inthe US literature, inwhich an absolute cut off is used to identify areas of concentrated poverty (i.e. census tracts in which at least 40 percent of the inhabitants are poor, National Academy 1990).Using this definition we find that in our sample there are both barrios with pockets of concentrated poverty and barrios uniformly very poor. Out of 7 barrios with poverty incidence between 40 and 60 percent, 2 of them have all their census tracts as areas of concentrated poverty, while the others have a varying proportion of census tracts with concentrated poverty. We do not find, however, examplesof barrios with low average incidence and pockets of poverty. 43 By construction inthese graphics pockets of poverty are unlikely to be found in barrios with very high or very low headcounts; on the other hand, there is no reason for barrios with mid-range headcounts to exhibit highvariations. 40 Figure3.1 Povertyincidence varieswithin barrios, particularly for extremepoverty Figure 3-1.a Extreme Poverty Figure 3-1.b Moderate Poverty Barrio level headcount (pll) and spread of headcount between census tracts .. Barrio level headwunt (p13) and spread of headwunt between census tracts 0 N4 0- 0 a . a I a 0 0 . 0 0 0 0 % 0 a 0 . 0 0 Source: Own calculations basedon Barrio Survey Concerns for the concentrationof some types of deprivationcan be extended to non- monetary dimensions of deprivation - Figure 3.2 plots differences in access to sewerage (panel a) and electricity (panel b). These two types o f services are very different in the structural requirements for their delivery, and appropriate sewerage in particular might require high retrofitting costs once some other system has been adopted. Differences in sewerage provision between different areas can be particularly important given the externality element - in terms for example o f risk o f infectious diseases - that are associated with this utility. Figure 3.2 shows that access to electricity i s almost universal and that variations betweendifferentparts o f a barrio are very limited. On the other hand, access to sewerage represents a much larger challenge for many barrios andthe invertedUshaped pattern which emerges i s compatible with the gradual expansion o f the network to different parts o f the barrio, so that inequality in service provision first increases then decreases. Figure3.2 The more commonthe service, the smaller differences existswithin a barrio. Figure 3-2.a Access to sewerage i s gradual, Figure 3-2.b Access to electricity i s quasi universal implying great differences across census tracts and does not vary much across census tracts, within within any given poor neighborhood poor neighborhoods Barrio level lack of access to sewerage and difference 8n lack of access between census tracts --Barrio level lack of access to electricity and difference 8n lack of access between census tracts 0 - ;z a 0 . a 0 a 8 0 tPO 0 7 7 0 Barrio level lack of4access to sewerage 2 8 0 02 Barrio level lack of04accessto electricity 06 08 Source: Own calculations basedon Barrio Survey 41 What is driving different outcomes inpoor urban neighborhoods? What determines the high inequality in well-being within and across barrios? Large variations in indicators o f well-being and access to services across poor urban contexts and different degrees o f discontinuity across groups within those areas raise the issue o f what i s determiningthese differentoutcomes. Differentfactors seem to play a role inthis issue, linkedto the structural and institutional features o f poor neighborhoods and their integration with the surrounding economy, or to the interaction betweenpoor people characteristics and those o f the areas they live in(regression results are presented inAnnex Table A-6). Several barrio level characteristics, as well as those of the municipality the barrio belongs to, are significantly linked with the barrio's income level. Inparticular, more isolated barrios as measured by the average time that people inthe barrio take to get to work, tend to be poorer -though own distance i s positively correlatedwith income (people are willing to travel for betterjobs). Similarlythe age of the barrio measured as the modal age o fthe existing houses also matters with newer barrios exhibiting lower income, all else being equal. Lower access to public sewices in the barrio relative to access within the municipality i s also associated with lower incomes and so i s a higher propensity to natural disasters according to self-reported assessments. Among the characteristics o f the broader environment in which the barrio i s situated as captured by municipal level variables, the structure of the local economy matters: with higher shares of constructionjobs across municipality and lower share in services associated with higher incomes. Public expenditure per capita i s associated with higher incomes. Larger cities in 1990 and cities which have grown faster since are also correlated with higher incomes. High illiteracy i s associated with negative effects. To assess the relative importance o f different sets o f factors in determiningaverage living standards at the barrio level we have run some simulations. Contrasting the simulations with a baseline based on the real characteristics o f the sample provides an estimate o f the effects o f the various factors we are considering4 The detailed results are presented inAnnex Table A-6. The range o f effects that different factors can have on the barrios i s presented in Figure 3.3. The results refer respectively to the simulated effects o f removing inequality in household characteristics at the barrio level (i.e. all household have their characteristics equal to the barrio average); removing inequality o f household characteristics across the sample; removing inequality across barrios in barrio characteristics; and forcing returns to household characteristics to be the same across barrios. 44 It is important to note that standard decompositions adopting variations of the Oaxaca-Blinder methodology (e.g. Ravallion and Wodon 1999)identify the differences due to different characteristics and a residual component, which can be attributed to differences inreturns (if the specification of the model seems robust and comprehensive). In this case we have proceeded differently, running separately regressions for each of the barrios and contrasting those with a pooled model where by definition the returns are forced to be the same across barrios. The difference, therefore, provides a direct estimate of the effect of differences of returns to characteristics across barrios. 42 Figure3.3 Observedlargestpositive and negativevariations in average income by barrio, under different assumptions(as % of food basedpoverty line) 060T 000 ~ equality within each effects o ' " m i n g i qualify across allthe effe its of remo~ "on returns to bardo haracfedsfic of the bardos ot individuals The largest simulated variations in average income are those obtained simulating the income distributions with equal barrio characteristics across the sample, pointing to the importance o f factors such as distance from jobs and type o f economy in the municipality to which the barrio belongs. Also the variation in returns to individual characteristics are substantial, though the span o f variation observed was overall less than for changes in concentration o f characteristics. Barrio level characteristics are associated not only with differences in average income but also to the availability of services. The age of the settlement (together with the age o f a particular house) i s a key determinant o f the physical quality o f life since infrastructure, both public (water sanitation, transport etc) and private (the housing stock) improves over time inpoor neighborhoods. This i s due to the way the poor acquire housing, typically in a progressive manner, first buildinga shack on an unsecured piece of land and gradually improving the house as well as possibly legalizing their tenure status (which conditions access to public infrastructure services). The process of improving quantity and quality of housingdiffers according the age of the settlement (defined as the modal age of the houses in the settlement) (Figure 3.4). The availability o f water in the house or in the land can vary by as many as 47 percentage points depending on the house age in a given type o f settlement and by as many as 34 percentage points for same age houses age across settlements o f different ages. Settlement age seems to matter less inthe case of sewerage possibly due to high costs of retrofitting any sewerage arrangement the house might have been equipped with already. Differences across barrios are not limited, however, settlement age. When grouping barrios in broad geographical regions, those in the Northare more likely to offer access to water provision, sewerage, and garbage collection. 43 Figure3.4 Access to services inpoor neighborhoodsimproves with bothsettlement age and that of a house Figure 4a. Access to water Figure 4b. Access to sewerage Access to public water by age ofthe house and age of Accesstopublic seneragenetmrk the settlement milovean 1, lejstbmlyear 1-5yean 5lOyean 1 0 + p I a s than 1y m 1-5years 6-10years morethan 10years ageof thehousngutllt urce: Own calculations basedon Barrio Survey Prioritiesfor poor neighborhoods What are the perceived needs of these barrios? Inparallel to data collection in these poor neighborhoods, SEDESOL commissioned qualitative analysis based on focus groups. The information that emerged was analyzed by the Centro de Investigaciones y Estudios Superiores en Antropologia Social (CIESAS). In particular, the analysis enabled a hierarchy o f needs to emerge. Access to the city is the most frequently need invokedamong the five most pressing priorities, as well as the one that most often emerges as number This coincides with findings from a qualitative analysis o f Oportunidades beneficiaries by Escobar Latapi and Gonzalez de la Rocha (2004) which argues that people living in Mexico's poor urban neighborhoods live in the city but do not benefit from urbanization (``est& en la ciudadpero no gozan de la urbanidad'?. Access to water and sanitation where they are a problem are usually considered the highestpriority for those lacking them. Public safety and education infrastructure are also often mentioned as an important need, though rarely as number one. 45At the same time, publictransportation is rated as a low priority. Itis unclear what is behind this seeming contradiction. Perhaps cost or distance, whereby public transportation is deemed adequate, but takes too muchtime, becauseof distance and traffic, or isseenasunaffordable. 44 Table 3.2 Prioritiesfor those livinginpoor urbanneighborhoods- access to the city followed bywater, sanitation, education and safety from crime andviolence Need Priorityranking Frequency 1 2 3 4 5 Access to the city 16 17 4 3 3 43 Water 15 3 2 2 2 24 Drainage 11 6 7 2 1 27 EquipmentinEducation 7 6 11 5 7 36 PublicSafety 5 8 4 5 2 24 Landtenure 5 3 4 1 2 15 Equipmentinhealth 5 5 5 3 5 23 Pollution 5 5 2 0 5 17 Housing improvement 4 0 0 3 0 7 Electrification 4 5 2 1 0 12 Public lighting 2 2 4 3 0 11 Sidewalks and garnitures 2 4 2 0 2 10 Refuse collection 1 1 1 6 2 11 Recreational spaces and green areas 1 3 0 5 8 17 DrugAddiction 1 2 0 1 1 5 Equipmentinsupply 0 1 0 1 2 4 Public transportation 0 0 1 0 5 6 Source: CIESAS(2004) SHAPING ONE'S PLACE, THE IMPORTANCE OF SOCIAL CAPITAL Social interactions have been highlighted in the literature as important characteristics of the environment poor people live in and contributeto shape. Inthe case of Mexico, attention to these non-market relationships and their role as a resource for the poor in urban areas i s not new. Already in the 1950s, Oscar Lewis, for example, noted that urbanization was taking place without the expected breakdown in traditional social ties. Poor newcomers to Mexican towns maintained an intense web o f relations with their rural family members, and had close social ties intheir new urban location (Beezley 2003). Social relations withinpoor barrios The barrio surveys point to the importanceof social capitalin poor neighborhoods. Evidence from the barrio surveys allows us to analyze at least some o f these relations. As many as 31 percent of respondents affirm that they would trust someone in the colonia to deal with personal or neighborhood related issues, and 28 percent say that they would resort to the help o f this person in case of a crisis such as illness, need for accommodation or lack of money. The potential for resorting to this type o f informal transfers for coping with shocks i s clearly high, given the relatively limited coverage o f formal insurance systems (for example, only half o f our sample i s covered by formal medical insurance) - and indeed 15 percent of respondents resorted to this type o fhelp inthe last month. These networks are clearly characterized by a highdegree o f reciprocity, as in 51 percent o f cases respondents reveal that they have themselves helpedin the last month the individuals whom they consider potential sources o f materialhelp. Aside from informal support networks, formal horizontalorganizations exist in the barrios. Furthermore, almost 70 percent o f the barrio inhabitants had attended at least one 45 meeting o f one o f the 6 major types o f organizations listed over the previous year.46Effective membership in these groups is somewhat lower. Church or religious groups and parents' school clubs were the most attended social activities andthose with widest membership. Indicators of willingness to participate in communal activities in the neighborhood are high. They differ significantly, however, between activities to create or improve different services in the colonia (90 percent would be willing to participate in at least one) and collaborating in activities to improve the environment (less than 70 percent). One possible difference between the two sets o f activities i s that the latter seems to be more demanding in terms o f regular commitment o f time as it includes, for example, help ingarbage collection, rather than one off types o f efforts (e.g. help in create or improving public parks). Overall 95 percent o f the household declares that they would want to engage in at least one o f those activities. Informal community organizations do exist to help solving problems related to e.g. public services or conflicts over property rights. Finally, we have information about community informal organizations. 55 percent o f respondents declared that in their colonia there was either a person or an organization which would solve problems related to public services, or that if there was any problem with public services people would gather to solve them. Even greater emphasis seems to be put in organizing the regulation o f property rights as only 28 percent o f the respondents reported that no mechanism existed to deal with problems arising in this respect. But the extent of social capital varies greatly across barrios. Data show however a large amount o f variation across barrios in terms o f these non-market relations: the share o f households who would trust someone in the colonia to help, for example, ranges between 11 percent and 58 percent. Literature provides many pointers to household and community determinants o f the different intensity o f social capital. Significant effects have been repeatedly found for example for variables such as homeownership which can be expected to influence individual incentives to be socially connected and provide local amenities (DiPasquale and Glaeser 1999). Others, more connected to individual experiences such as recent traumatic shocks, belonging to a discriminated group, low income, low education, strong racial division and income inequality can in contrast be seen as determinants o f trust (Alesina and la Ferrara 2002; interestinglythey do not findreligious beliefs and race to matter). Yet another group of variables has been found to be influencing participation in different kinds o f groups; these include age, years o f schooling, and number o f children - supposedly through a time availability mechanism. O f these variables, community heterogeneity along racial, ethnic or income lines appear to have a strong impact, though inthe case o f participation o f differenttypes o f social activities inthe U S it has been found that it matters less for activities with a high degree o f exclusion or where interaction i s limited. What determines the access to social capital? Annex Table A-8 presents the results o f probit analysis aimed at identifying key correlates associated with the set o f social capital indicators available. As these indicators capture different aspects o f social capital we would not necessarily expect them to be associated with the same variables. The two indicators relating to informal support networks, in particular, seem to capture an aspect o f social interaction quite different from the other indicators, as their correlates show a pattern of associations often at variance with the others. 46Parents' school clubs, churchor religious group, cultural or social or ethnical organizations, neighborhood or community organization, trade union, political party -in the case of attending meetings/ gatherings there is also a 7* option ie sports club. 46 Owning one's home, havinglivedinthe barriofor some time, and havingsome level of education increases access to networks. Overall, results presented in Annex Table A-8 are broadly in line with those found in the literature on social capital. Among household characteristics, for example, homeownership significantly affects all indicators but those for informal support networks. Beingnewcomers inthe barrio (as capturedby the share o fhousehold members who moved in the last 5 years) decreases the chances that reliable networks of reciprocity have been established and decreases all types o f social activities with the exception o f the willingness to participate in activities aimed at building something.47 A dummy for being catholic (the religion most practiced in our sample) has been introduced and consistent with evidence from other countries we find it i s mostly not significant, with the exception o f significantly decreasing the likelihood o f being a member o f an organization. Years o f education household head are positively associated with all indicators aside from knowledge on the existence o f informal organization, and significant in the case o f having used reciprocity networks, willingness to participate in providing services inthe community and membership and attendance o f organizations. Gender, age and ethnic patterns emerge in the responses. Women are significantly less likely to have and to use reciprocity networks and to want to build something. In contrast they are more willing to help inrunningcommunity services and assist to meeting. Interestingly, they seem also to be less involved or aware of the workings of community organizations or actions to solve public services issues or property rights issues. As far as age i s concerned an interestingsplit emerges: social attitudes are stronger in older people, though social connections through reciprocity networks i s less, possibly because older people are less likely to be able to reciprocate for the help received. Finally, indigenous people have less access to reciprocity networks and declare to be less likely to want to collaborate in one off projects for the community. At the same time they are more likely to be part o f some group/organization. What barrio characteristics determine the build-upof social capital?As far as barrio characteristics are concerned, variability o f income inthe barrio decreases social connections, but increases other form o f social activities. The effects o f more racially mixed environments are rather similar, while religious heterogeneity inthe barrio (catholic and non-Catholic) increases all forms o f social interaction. Older settlements have fewer support networks and, interestingly, while in the oldest settlements social connections and informal organizations are lower, they appear to be higher than in the newer ones in settlements 5-10 years old. We try to capture the overall social integration inthe barrio usingthe barrio average o f the "having someone to help" variable, in order to test whether the social structure o f the barrio has an independent effect on other aspects o f social capital. We find that it i s extremely significant and that while it decreases the desire to do one off projects, it increases all other forms o f social capital, and in particular informal organizations. Finally, these results highlight also regional differentiations: for example, households in barrios located along the Northern border are less likely to ask for help o f neighbors while the opposite i s true inthe South. There are also differences across cities size and growth rates - inlarge cities there aremore informal support networks andless ofother indicators o f social capital, the latter finding also found in fast growing cities. I s social capitalan importantresourcefor the inhabitants of poor urban barrios? As mentioned before, several indicators discussed here have been found to be associated with a variety o f positive outcomes inboth macro and micro studies. The literature offers however some 47One can speculate that this might be due to some form of reverse causation, with new comers occupying spaces where more amenities are needed, or, alternatively, pointing to a willingness to create ties within the community. 47 more specific hypotheses and inwhat follows we consider the impact o f social capital indicators on two variables that are key for poor people -employment and the incidence o f crime and violence. Uses of social capital:finding ajob The role of informationflows across social networks in increasingthe probabilityof finding a job has been subjected to increasingscrutiny.U S and Britishstudies findthat asking friends andrelatives is the most effective search mechanism (e.g. Blau and Robins 1990), and the role o f networks in finding a job in the context o f migration has been well established (see Munshi 2003). A recent paper based on detailed data from Egypt (Wahba and Zenou 2003) focuses specifically on the characteristics o f the network through which information i s shared. Crucial inthis context are the homogeneity o f the network (and inparticular whether it extends to individuals with diverse backgrounds), the quality o f information which can be passed and the spread o f the network itself, which mightresult in congestioninthe use o f information. While we lack the detailed information needed to test these hypotheses, some analysis can be performed focusing on indicators o f individual connections as proxies for belonging a network. Note that we cannot use indicators such as involvement in community organizations since the fact they explicitly require a time commitment makes them endogenous (unemployed people are more likely to belong).48 Network information (quantity and quality) has an impact on job search Annex Table A-9 details results o f our analysis o f the effects o f access to networks and the probability o f being employed. Having someone inthe barrio whom one would trust to solve a problem or ask for help i s found to be positively associated with employment, while the negative coefficient on the barrio average o f the same variable hints to the possibility o f some congestion effects (when everybody i s well connected, connections do not help in finding ajob). We also control for other indicators o f informationavailability (number o f householdmembers inthe workforce) and of its quality (the share o f the barrio labor force which i s unemployed). While the latter might well be capturing broader economic factors which are reflected in the individual probability o f having employment, it i s interestingthat the interaction o f the indicator o f household access to network o f reciprocity with the unemployment share leads to a dampening o f the effect o f the reciprocity network (the coefficients are jointly significant at the 1 percent level). This suggests that, as found elsewhere, the quality o f the information available in the network matters - where unemployment i s high, the networks are not capable o f relaying much information about jobs. Similarly, the effectiveness of network in helpingwith ajob i s dampened by its interaction with the household members inthe labor force -this couldpoint either to congestion effects within the household or the importance o f alternative ways o f obtaining information about vacancies. Exposure to crime and violence Exposure to crime and violence in urban areas is a key concern in Mexico as in other Latin American countries. Economic theory would suggest that crimes, at least those related with property, should be more frequent in urban areas due to higher concentration o f transferable assets found there. And indeed, evidence from the Mexican capital (Distrito Federal) suggests that urbanization, though not in itself responsible for the surge in crime over the 1980s 48Note the difficulties of identifying plausible identification restrictions 48 contributed to social and economic changes (lack o f housing, jobs and public services) related to a rise in crime, which the economic crisis exacerbated. Exposure to crime is high in barrios, but few are reported to the authorities. Data from our survey shows high shares o f reported crimes: 10 percent o f respondents report that a member of their family has been victim of a crime inthe last 6 months. O fthese crimes, only 32 percent have been reported to the authorities. Theft i s the most frequent crime with 8 percent o f the respondents reporting one. These data can be put in context by considering the crime incidence by neighborhood, as detailed by Table 3-3, which provides summary statistics on the incidence o f different crimes across slums. Comparison with international evidence shows that these neighborhoods suffer from a high density o f criminal activities. For example, at 20 per 100,000 persons, the homicide rate i s comparable to the L A C average o f 30 per 100,000 persons, but is five times higher than the Mexican average of 4 per 100,000. This is reflected in the widespread sense o f insecurity reported by the inhabitants o f the urban slums (in 21 o f the 31 barrios the most frequent answer given i s that the colonia is not very safe, and in 7 more the answer i s that the colonia i s unsafe). Table 3.3 Crimeincidencevaries across neighborhoods,butis generallyhigh No. of Crimesper 100,000 Min Max persons(*) Total crimes 9852 944 28846 Assault 8194 3226 18453 Injuries 1023 0 2472 Abuse of authority 113 0 806 Threat 296 0 1020 Fraud 27 0 223 Homicide 20 0 219 Breach o f confidence 30 0 202 Damage inalien property 454 0 1590 Sexual crime or attacks to the modesty 142 0 674 Abduction (express) 19 0 197 Abduction 18 0 375 Others 94 0 410 (*) Data obtained from questions detailing the nature of the last crime of which one of the household's members was victim - overall numbers might therefore behigher Source: Own calculationbasedon Barrio survey We explore the determinants of number of crime reportedin each barrio over the last 6 months, testing for the importance of social capital type of variables. A negative relation between social capital type o f indicators such as trust and civic attitudes and crime has been found in cross-sectional analysis (e.g. Lederman et a12000). Most pertinent for the analysis o f our barrio data, the importance o f "neighborhood ties, social control, mutual trust, institutional resources ..." has beenhighlightedby literature on urbanpoverty inthe U S (Sampson et a12002). Giventhe limitednumber of barrios available inthe survey, any empirical analysis is somewhat limited in scope and in possible sophistication, but some interesting associations can be highlighted(see results inAnnex Table A-10). Indicators of the social fabric and civic attitude in the barrio have a significant impact on crime rates. The density of support networks is negatively related to the number of crimes and so i s the share o f respondents who would like to participate in activities to improve 49 the infrastructure o f the barrio. Another interesting finding i s that local amenities, such a public lights inthe street, decreases the number ofreportedcrimes. Policingis positively associated with crime, suggesting that patrols are more frequent in areas with higher incidence. Finally, income per capita in the slum -a proxy o f the amount o f transferable assets which exist in the barrio- seems to increase the number o f crimes though at a decreasing rate. These results are consistent with those found inthe literature (Moser, Winton and Moser 2004). CONCLUSION This chapter has relied on a recent survey of poor neighborhoodsin urban Mexico to discuss characteristics of the places of the poor. What emerges i s a picture o f great diversity: while the poor are indeedconcentrated inpoor neighborhoods, not all poor live inthem and, most importantly for the topic at hand, poor neighborhoods include both poor and non-poor. Indeed, the incidence o f poverty across our sample o f poor barrios varies from 18 percent to 55 percent, for an average o f about 36 percent. And income inequality i s much greater within poor barrios than across them. Data shows also that within a given poor neighborhood there can be pockets o f concentrated poverty. What matters for barrio income levels? Unsurprisingly,the average income level in a poor neighborhood i s found to be negatively correlated to distancefFom centers of employment, the age of a settlement, lower access to public services, andpropensity to natural disasters. The characteristics o f the city to which the barrio belongs such as the structure o fthe local economy, - public expenditure per capita, city size and growth- are also found to matter. The gradual way in which the poor acquire housing shows up in a positive correlation betweenthe age o f a settlement on one hand and access to services and quality o f housing on the other. What are the most pressing needs perceived by barrio residents? An analysis of the needs expressed by the residents o f poor neighborhoods shows that they feel physically isolated from the city they live in - access to the city i s by far the most frequently mentioned need. Other highpriority needs that emerge are water and sanitation, as well as education infrastructure and public safety. Social capital i s found to be fairly high in these poor barrios, although great variation exists across barrios. A variety o f indicators o f individuals' level of social connections and o f neighborhood level organizations indicate highlevels o f social capital. However, these are found to vary significantly across barrios. Greater social connections are found to be positively associated with employment, suggesting that social capital does indeed play a role in the economic well-beingo f poor households. Greater social capital appears to reduce the incidence of crime and violence - which i s high in these poor neighborhoods, much more so than in the rest of Mexico. Other factors that appear significantly associated with crime include higher income per capita (which i s a proxy for the amount o f transferable assets) and local amenities (with public lights in particular acting as a deterrent). Policingi s positively associated with crime. What then are the implications of this chapter? First, where people live matters in determiningtheir opportunities, as well as the challenge facing them. This impliesthat the spatial focus o f Hhbitat, the new urban poverty program developed by the Mexican government, i s appropriate. Second, the potential negative externalities arising from segregation need to be considered in the context o f housing policies which tend to cluster low-income households 50 together. Third, poor neighborhoods are tremendously varied in their structure and composition, which argues in favor o f programs that leaves room for self-determination. Finally, social capital and social structures are important in these neighborhoods - indeed Hhbitat seeks to build on local social structures. It i s known from other studies that the benefits o f social networks, particularly in terms o f procuring employment and securing benefits, are greater when these social networks can span social groups and geographic areas (Woolcock, 2004). This suggest that one a role o f public intervention, whether through Hhbitat, or through other programs active in poor urban barrios such as Oportunidades,couldbe to helpthese networks reach out. 51 4. POVERTYAND LABORMARKETS INURBAN Employment is the main source of income for poor urbanfamilies. This dependence accounts for about 57 percent o f the income o f the poorest urban quintile in 2OOL5' Enhancing the ability o f the poor to get and holdbetterjobs, both in terms o f wages and quality is, therefore, central to reducingurban poverty inMexico. This chapter explores the role of the labor market in the urban poor's lives. This chapter does not pretendto be a poverty profile.Itidentifies the constraints and challenges the poor face in order to recommend policies that enable them to better take advantage o f the employment opportunities offered. The evidence shows that the urban poor were working more, but for lower wages in2003 than in 1991, as they found less opportunities inmanufacturing and the public sector, and more inthe self-employment sectors On the positive side, real wages of the poor showed important growth since 1996, but especially in the later years studied (2000-2003). This, combined with higher employment and participation rates, has eased extreme poverty, but many urban inhabitants still find themselves in moderate poverty. Explanations can be found in the structure of the Mexican economy, such as the type o f jobs it offers and its international competitiveness, as well as in the human capital and other characteristics o f the urban poor that condition access to betterjobs andthe labor market ingeneral. Work done for this chapter relies on the Encuesta Nacional de Empleo Urbano (ENEU),Mexico's main labor survey, which provides apanel of considerable size that includes several Mexican cities for the period 1987-2003.51Since this survey appears quarterly, this paper usedthe second quarter o f each year. Official poverty lines, as constructedby the Comitk Tkcnico para la Medicibn de la Pobreza-CTMP- (2002), are used. O f the three nested poverty lines used by the Mexican Government, we only use two - food poverty (poverty line 1) and asset poverty (line 3), which correspond to extreme and moderate poverty respectively. Any mention o f the poor as a whole refers to all individuals living inasset poverty, i.e. the moderately poor. At this point,it is importantto introduce a key methodologicalnote. Given the fact that the purposeofthis chapter is not to provide informationabout poverty incidence or the number of poor, but rather to explore how the poor fare in the labor market compared to other groups, an adequate way to identify the poor and their personal and labor market 49This chapter was written by Gabriel Montes, Mauricio Santamaria and Magdalena Bendini and edited by Mary Morrison. The authors relied on research produced by them as well as by Orlando Gracia, David Kaplan and William Maloney. Marianne Fay provided extensive comments. Using the methodology proposed by the Technical Committee for Poverty Measurement to calculate income. Urban localities are defined as those with a population of 15,000 or more people. Labor income in the ENIGHincludes categories POO1-POO9. 51Despite the fact that the coverage and scope of the Encuesta Nacional de Empleo (ENE) is broader than that of the ENEU, the latter was chosen because it provides longer series, which was necessary to carry out the analysis. 52 characteristicswas needed. Thus, the methodology used consisted o f considering "poor worker" that who belongs to a household with a per capita labor income lower than the poverty lines definedbythe CTMP. This methodological choice i s dictated bythe fact that the ENEU,which i s the instrumentthat contains the relevant labor market information, only includes data about labor incomes in the main job o f the individual^.^^ In any case, given the fact that non-labor incomes are more important among the richer households, we believe (and empirically checked) that the classification o f workers according to this method i s very accurate.53Also and as mentioned above, since the purpose o f the chapter i s not to provide information about the incidence or evolution o f poverty itself, this seemed as the soundest choice. Additionally, it should be noted that since data for the period 1987-2003 are used, to maintain comparability we had to restrict the sample to the 16 cities present in all the surveys. Thus, all the results hold for this universe. The chapter is organized as follows. The first sectionpresents an overview ofthe labor situation o f the urban poor: where they work, in what condition, and how this has evolved over the last 15 years. Section I1builds on this information to discuss the changing capacity o f the Mexican economy to provide decent jobs for the urban poor. Section I11 then examines the characteristics o f the urbanpoor, and how these affect their labor market access and performance. Section IV discusses the structure o f Mexico's labor market, looking in particular at how its ability to absorb shocks through lower wages has changed over time, as well as the impact o f labor regulations and active labor market policies. The final section summarizes the main findings and presents policy recommendations for improving the labor market performance o f the urban poor. URBANPOOR,WORKING MORE FOR LESS PAY The analyses below were carried out by the authors o f this chapter and are based on academic research or research conducted by them, unless otherwise specified. Methodological details can be found inBendini, Montes, and Santamaria (2005), which i s available uponrequest. Employment rates have risen among urbanpoor but remain low, especiallyfor women Employment rates in Mexico have risen fairly steadily in the last 15 years (Figure 4.1). The increase has been particularly large among the poor, whose employment rate rose from 36 percent in 1991 to 48 percent in 2003, largely inresponse to the crisis o f 1994/1995 -indeed most o f this increase occurred in 1995 (Table 4.1). This i s because, as discussed in Chapter 2, the urban poor tend to rely on an "added worker" strategy, whereby additional household members enter the labor markets to cope with income shocks. While this strategy has helpedmitigate the effect o f the crisis, particularly the fall in real wages, a negative aspect has been an increase in employment among poor youth. This may imply a curtailing o f education, hence a decrease inthe probability o f a future transition out o f poverty. The poor also increased their weekly hours o f work by about two hours during the period, (from 42.5 to 44.5), but this was in the same proportion as for the non-poor. 52 Labor is the most important source of income for the poor inurban areas. As discussed in the companion reports, there are other sources of income and household strategies to complement the labor income of the poor. 53This analysis does not pretend to be a poverty profile, only a way to rank households within the ENEU. 53 Figure4.1 Source: ENEU. EmploGent rate defined as persons employed as a share of persons of working age (12 years and older). Overall, the poor participate much less in labor markets than the non-poor, although the gap narrowed over the decade. This i s the main reason behindthe lower rate o f employment for the poor (48 percent) than for the non-poor (70 percent). The poor's unemployment rate (5 percent) i s low - although higher than that o fthe non-poor (3 percent). Table 4.1 Mexican poor have lower employment rates and higher unemployment rates 1991 1995 1999 2003 Non-poor Poor Non-poor Poor Non- Poor Non- Poor poor poor Employment rate (%) 65.4 35.8 72.0 45.3 73.5 47.1 70.4 47.8 Unemployment rate (%) 1.4 5.0 5.6 9.6 2.3 3.7 3.1 4.7 Participation rate (%) 66.8 40.8 77.6 54.9 75.8 50.8 73.5 52.5 Employment rate (%) for youth aged 14-17 26.0 15.7 25.8 22.8 24.0 23.9 22.7 21.5 Source: Own calculationsbasedon ENEU. The increase in female participation rates has been an important factor driving the overall increase in the employment rate over the decade. Female participation rose from 35.6 percent in 1992 to 39 percent in 2003. In contrast, male participation remained constant over the period. Nevertheless, female participation in Mexico remains among the lowest in Latin America (itis higherthan 50 percentinmost large countries inthe region). Thepoor's employment opportunitiesare increasingly concentratedin lower-quality sectors Access to good jobs has fallen for the poor: lower quality jobs are on the rise. Among the main sectors o f employment for the urban poor, manufacturing, which offers relatively good jobs, has ceded ground so that its share in employment o f the extreme poor fell from 26 percent in 1991to 19 percent in2003 (Figure4.2). As a result, an increased share o f poor workers now work in construction, commerce and personal services - sectors characterized by lower-than-average wages, high informality, slow growth and a tendency to decline sharply in recessions. 54 Figure4.2 Becauseof manufacturingdecline, poor are increasingly employedin sectors that offer lower qualityemployment 100% ~ 90% ~ 80% ~ 70% ~ 60% ~ 50% ~ 40% ~ 30% ~ 20% ~ 10% ~ 0% 1991 2003 10Manuf.0Comrce 0Pers.Sew. 0Const.0Others 1 Note: the data refers to urban extreme poor (under poverty line 1). Source: Own calculationsbasedon ENEU. Similarly, self-employment has increased. Another notable change is the sharp drop in the proportion o f the poor employed in the public sector, and the increase in self-employment (Table 4.2). Thus, it appears that the increase in overall employment o f the urban poor after 1996 was mostly through self-employment o f the lower category (self-employment without investment or SEW154),which i s typically precarious and low paid. This was particularly notable for the poorest, 35 percent o f which are now self-employed, mostly inthe SEW1category. Nevertheless, the majority o f the urban poor continue to work as private employees, with the proportion flat at 65 percent for the moderately poor over the period as a whole, but falling to 59 percent from 64 percent for the extremely poor. 54 This corresponds to what is defined in ENEU as "cuenta propia sin local." 55 Table 4.2 Self-employmentis on the rise for the poor Employmentshares 1991 1995 1999 2003 Totalworkforce Employer 4.7% 4.9% 4.1% 4.2% Self-employment 16.9% 18.2% 17.3% 19.2% With Investment 5.4% 5.6% 4.4% 5.2% Without Investment 11.5% 12.6% 12.9% 14.1% Public employees 16.8% 14.6% 13.6% 13.8% Private employees 61.6% 62.3% 65.1% 62.7% All poor (povertyline3) Employer 2.5% 2.8% 2.8% 2.4% Self-employment 19.7% 21.9% 20.6% 25.0% With Investment 5.2% 5.4% 4.6% 5.5% Without Investment 14.5% 16.5% 16.0% 19.6% Public employees 12.5% 9.5% 8.0% 7.2% Private employees 65.3% 65.7% 68.6% 65.3% Extremepoor (povertyline 1) Employer 2.0% 2.3% 2.3% 1.9% Self-employment 24.3% 28.6% 26.0% 34.8% With Investment 5.7% 6.4% 5.3% 6.4% Without Investment 18.6% 22.2% 20.7% 28.4% Public employees 9.6% 5.9% 5.1% 4.6% Private employees 64.0% 63.2% 66.6% 58.7% Note: Consistent sample across periods. Source: own calculationbasedon ENEU. Mostjobs seem topay less as comparedto 1991, especially those available to thepoor Wage levels fell most for self-employed poor workers. The impact of this reallocation of employment is reflected in the evolution of wages. Not surprisingly, wage levels fell most in the categories that experienced the highest surge in employment. Thus, although wages plunged across the board after the Tequila crisis, this was strongest for self-employed poor workers. Overall, pay levels have recovered since 1996, although the improvement was not sufficient by 2003 to regain the value lost since 1991. This is particularly true for poor workers, driven by the notable lack of a recovery in wages of the self-employed. Overall, average wages fell by 2 percent over the period, while S E W I wages dropped by 9 percent (Table 4.3). For the poor as a group, wages eroded by 5 percent overall and 8 percent for SEWI, while for the extreme poor the losses were 12 percent and 22 percent respectively. Realwages also fell by more than the average in the industries inwhich most of the poor work (construction, commerce and personal services), as job growth in those was outstripped by an expansion of the pool of workers seeking them. 56 Table 4.3 Realwages have fallen most for the types of employmentwhere poor predominate Wage trends 1991 1995 1999 2003 Employer 1.oo 0.96 0.91 0.98 Self-employment 1.oo 0.87 0.84 0.92 With Investment 0.88 0.88 0.96 1.oo 1.oo Without Investment 0.86 0.84 0.91 Public employees 1.oo 1.01 1.02 1.os Private employees 1.oo 0.93 0.92 1.oo All poor (Poverty Line 3) All 0.91 0.90 0.95 1.oo 1.oo Employer 0.94 0.91 0.95 Self-employment 1.oo 0.87 0.84 0.87 With Investment 0.87 0.94 0.86 1.oo 1.oo Without Investment 0.89 0.88 0.92 Public employees 1.oo 0.98 0.98 1.01 Private employees 1.oo 0.92 0.90 0.97 Extreme poor (Poverty Line 1) All 0.89 0.88 0.88 1.oo 1.oo Employer 0.94 0.91 0.90 Self-employment 1.oo 0.87 0.85 0.79 With Investment 1.oo 0.87 0.88 0.88 Without Investment 0.87 0.85 0.78 1.oo 1.oo Public employees 0.94 0.94 0.95 Private employees 1.oo 0.90 0.89 0.93 Source: own calculationbasedon ENEU.Note: Consistent sample across periods A further indication of the falling standards of jobs for the poor is given by the increasing share of the poor who work in small firms. In 2003, firms with less than 5 employees accounted for 65 percent o f the very poor's employment, as opposed to 40 percent of total employment. Small firms are less likely to comply with employment regulations or provide benefits or stable longer-termjobs. They also pay less than bigger firms: in 2003, wages in firms with fewer than 10workers paidbelow 1991 levels, whereas larger firms saw increases o f about 4 percent. In sum, the situation of the urban poor in Mexico appears to have deteriorated in the 1990s, although a recoveryis evident at the end of the period. As a consequence o f falling incomes (and growing unemployment during the Tequila crisis), the poor moved to very precariousjobs in sectors and occupations that offer low salaries and slow growth, and are highly vulnerable to recessions. At the end o f the period (2003), poverty had declined slightly because the poor were working more, but mostly in those "low quality" jobs. In this context, it i s important to stress that a characterization o f households according to their labor market attachment showed that there i s a strong correlation between the occupation o f the members o f the household and the likelihood o f that household o f being poor, being those with workers in SEW1 far worse-off than any other category. We now turn to a discussion of the structural changes inthe Mexican economy that may help in explaining these trends. 57 FEWER GOOD JOBSFOR THE POOR^^ The county's competitiveness has declined due to weak totalproductivity growth Mexico's ability to compete in international markets, especially in the US, has been diminishing. This i s largely due to the country's poor performance in improving total factor productivity (TFP), which also constitutes the main engine o f economic growth. Loayza, Fajnzylber and Calderbn (2005) find that, in the 1990s, TFP growth in Mexico was a low 0.4 percent per year, below the Latin-American average o f 0.7 percent, and well below o f the 1.1 percent observed among the seven largest countries inthe region. In fact, among this latter group Mexico only outperforms Colombia and Venezuela, characterized by a very poor performance during the decade. Additionally, TFP growth rates were above 1.5 percent in the 1960s and 1970s, and then plummeted to negative levels in the 1980s as a result o f the profound crises suffered by the country and the region in that decade. Thus, while there i s some recovery in the 1990s, the observed levels o f TFP growth are still below those observed 30 or 40 years ago. In this context, the authors find a matchingtrendwhen analyzing GDP growth: Mexico belongs to a group o f countries that experienced an increase in economic growth rates in the 1990s relative 1980s but, if the performance i s compared against previous decades (1960s or OS), the situation i s the opposite (Mexico's GDP growth rate in the 1990s was less than half o f that observed in the period 1960-80).56 In the period 1999-2001, a combination of slow productivity growth and faster rising wages, especially in dollars, has increased unit labor costs. The country's continued competitiveness abroad in the nineties, can be attributed to low wages, which declined in dollar terms duringthe Tequila crisis, helpingto offset differences inproductivity with the U S and other competitors. However, since 1999, and intensifying during 2001, unit labor costs for manufacturing have increased in Mexico due to a significant increase in wages (in dollar terms) and relatively slow productivity growth. In comparison, unit labor costs in the U S have steadily decreased since 1993 due to the faster expansion o f productivity over wages. This trend, however, seems to have reverted since 2003 due to a combination o f higher labor productivity growth (see below) and slow growth o f mean wages. Labor productivity growth was high in the first half o f the 1990s, low in the period 1999-2001, and somewhat picked up after 2002, although it did not reach the levels observed before 1996. As a result, industrial labor productivity increased by 4.8 percent per year in the period 1994-2003 and 3.3 percent in 1999-2004 (INEGI, Encuesta Industrial Mensual). Thus, industrial productivity growth was low when compared to the U S in the period(5.8 percent). 55 This subsection is based on a number of papers or Bank reports, mentioned throughout the text, which mainly constitute abody of analytical research and should not be interpreted as official statements. 56 It should be noted that Loayza, Fajnzylber and Calderon (2005) constitutes academic research that, by no means, should be considered as providing official estimates of the evolution of TFP. The estimates used in this Report correspond to what Loayza, Fajnzylber and Calderon call "model 1" in their paper, which is a very standard way of estimating TFP trends. This methodology is based on the idea that TFP is the "unexplained" portion of observed GDP growth. That is, the portion that cannot be explained by increments inthe factors of production (capital and labor) alone. Thus, the authors econometrically "fit" a production function and subtract from observed GDP growth the observed increments in labor and capital (weighted by their estimated elasticities) and the result constitutes TFP growth. For more methodological details, see the paper at: http ://www-wds.worldbank.org/servlet/ DSContentServer/WDSP/E3/2005/ 41271 000090341~2005042713205%/Rendered/PDF/321%0.pdf 58 Consistent productivity growth is essential for sustained wage increases, especially for the poor. Low productivity growth i s a source o f concern because, besides the factors mentioned above, productivity has a positive effect on wages inMexico, as shown by Montes and Santamaria (2004). Greater productivity facilitates sustainable higher wages for more workers, while wage hikes without productivity growth further damage competitiveness and hinder export growth and employment. The slowdown in sectors producing tradable goods (food and manufacturing) - which offer relatively better jobs - is already reflected in a decline in their share in the employment of the urban poor, from 39 percent to 30 percent for extremely poor workers (Table 4.4). This represents a loss o f good jobs for the poor, as exports tend to come from manufacturing and larger companies, both o f which offer comparatively higher salaries, more benefits and greater security, as discussed above. Table 4.4 Sectors producing tradable goods are employinga smaller share of urban Door Extreme Poverty Moderate Poverty' 1992 2002 1992 2002 Share o f urban poor working in: Tradable 39% 30% 24% 22% Non-tradable 61% 67% 76% 78% Source: Own estimates using ENIGH 1992, 2002 following SEDESOL's urbadrural definition and the official poverty measurement. The share of involuntary informal workers may have risen since the mid-1990s. Until the early 1990s, unemployment and informality in urban Mexico were negatively correlated, suggesting that an important share o f workers were voluntarily choosing the informal sector, rather than being pushed into it by a lack o f formal opportunities. The trend appears to have reversed (at least partially) after 1992, as informality began to follow a somewhat similar trend to unemployment, more compatible with a conventional pro-cyclical view in which a decline in labor demand results in a growing informal sector (Figure 4.3). This trend was especially evident during the Tequila crisis, when labor demand shrunk in an important fashion. Since 2001, however, the pattern i s not so clear, fact that may be indicating somewhat improved conditions inthe labor market. Thus, inthe period 1994-2000 the informal sector inthe Mexican labor market may have become more heterogeneous, with an increasing share o f workers joining the informal sector because o f deterioration inthe demand for formal labor.57 57The definition of informality used inthis report usually includes the self-employed plus salaried workers working in small firms with no access to social insurance (typically health). This definition is very close to the one adopted by the ILO and allows some international comparisons. However, it is not the official definition of informality inMexico, to which reference is made below. 59 Figure4.3 Since 1991, informalitv has risenwith unemDloyment Fgure 1 Unempbymentand Informalty In Mexco I 1 4 Unemployment -Formalllnformal Note: Informality includes both salaried informalworkers and self-employed. Source: Own calculationbasedon INEGI. Limited formal employment possibilities during that period pushed poor workers into lower quality employment. Recent research on workers' motivations in becoming self- employed also supports the view that limited formal employment possibilities increased the share o f involuntary workers in the informal sector. Academic research by Fajnzylber, Maloney and Montes (2003) find that transitions from salaried to self-employment are determined by education, experience, and wealth. Ifthis i s the case, then workers with low education, experience and capital are unlikely to be moving voluntarily into the informal sector, so that high rates o f self-employment in poor households may also be considered involuntary or induced by labor market distortions. Furthermore, the fact that significant increments in SEWI were accompanied by a large fall in wages, as shown above, suggests that the involuntary hypothesis regained importance in later periods. Moreover the fact that the share o f employment in SEWI increased 30 and 60 percent for the poor and extreme poor, respectively, provides further evidence o f the increased segmentation that has taken place inthe Mexican labor market, at least among the poor. Thesedevelopmenntshurt thepoor and the economy These trends are bad for workers, for whom they may imply lower wages, fewer benefits, little job security and uncertain working conditions. The regressive pattern o f Mexico's social security coverage disadvantages even the poor in formal employment, as coverage rates are positively correlated with education and income. But most informal sector workers, who, in any case, are concentrated in the poor, are totally excluded. Both the social security system and the unemployment protection scheme (limitedmostly to severance payments) currently in place cover only formal workers (see the companion study, Mexico: an Overview of Social~ r o t e c t i o n ) . ~ ~ But informality can also be damaging for the economy as a whole if it hinders productivity growth. Informal firms tend to be less efficient than formal ones as they rarely 58 Though the legal framework does not intend to discriminate between formal and informal workers, in practice, only formal workers receive the benefits established by the law, while informal workers (i.e., self- employed and informal salaried), by the nature of their occupation, are precluded, totally or partially, from receiving severance payments when the job ends. Of course, in the case of salaried informal workers they can recourse to legal suits to obtain what the law establishes, if they consider that they have been unjustly fired. However, this entails a long and complicated legalprocess that is rarely pursued by these workers. achieve economies o f scale or employ much capital. And they are less likely to exploit or develop new technologies and productive processes. Mexico's share o f employed population in firms o f less than six employees has hovered around 40 percent in the last five years, which makes it among the highest in the region.59Importantly, this percentage i s higher among the poor and increased over the 1990s. Traditionally, these smaller firms tend to concentrate in the informal sector. Recent estimates made by the INEGI in Mexico indicate that around 30% o f the labor force was employed in "non-structured jobs" in 2000 (trabajo no estructurado), which corresponds closely to informality, as defined by ILO. For a country with already low total productivity growth, such highlevels o f informality are therefore a serious concern. LIMITED EDUCATION OF THE POOR -AN OBSTACLE TO BETTER JOBS The methodologies used for this section allow for the estimation o f wage differentials across educational levels. Consequently, the sample was divided into several educational cells, corresponding to the main educational levels in the country. In addition, the methodology proposed by Mincer (in various papers produced in the 70s and 80s) and Katz and Murphy (1992), widely usedby economists were utilized. The results are consistent with other estimations produced for the case o f Mexico but they do not constitute official estimates. Low educationlevels are still a binding constraintfor thepoor Despite the recent improvements in Mexico's educational achievements, the poor still have much lower education levels than the non-poor. And while the poor have become better-educated, with the average years o f schooling for a household head rising to 8 years in 2003 from 6.6 years in 1991 (Table 4.5), the non-poor have maintained their advantage in educationwith a similar increase. The lower education level of the poor remains the main reason for their low relative earnings. Indeed, higher education i s associated with a lower probability o f being poor, even after controlling for other factors, and a higher probability o f having a better job. Higher education in itself may not directly lead to better paying jobs, but highly educated individuals typically have access to information about better jobs or unobservable characteristics that facilitate access to such jobs (e.g., inherited ability, family history, contacts and networks, location) (World Bank, 2004). 59 This was estimated with the sample of 16 cities in the ENEU panel. Though this indicator is not informality per se, it is used because it allows for an international comparison. 61 Table 4.5 The educationlevelof poor hasincreased, but non-poorhave maintained their advantage 1991 2003 Average level Poor Relativeto for the poor relativeto non-poor (average) non-poor Years o f schooling o f poor household head 6.6% 71% 8.0% 70% Share o f poor households with at least one college graduate 9.4% 25% 13% 30% Source: Own calculationbasedon ENEU. However, educationdoes not fully explain the lower income of the poor. An exercise conducted to evaluate the returns to education inpoor households compared with non-poor ones, found that poor households receive lower wages than non-poor households, even after controlling for education and experience. That is, conditional on having the same level o f education, poor households still receive lower wages. Moreover, the gap increased over time for all categories, so that in 2003, poor workers with only primary education earn 40 percent less than non-poor workers with the same level o f education, while poor workers with college education earn 100 percent less than their non-poor peers. The same patterns are observed after controlling by experience. The non-educational factors that thus curb the earning power o f the poor are likely to include inferior social capital, information and resources, the stigma often associated with slums and other poor areas, lack o f connections, ineffective networks, inadequate access to transport, medical and other services and a lack o f access to affordable childcare facilities. But low-skilled wages are up in both real and relative termssince 1997 How much wages reward education - and penalize the lack of it - varies with changes in the economy that affect the supply and (especially) demandfor different kinds of labor. Such changes include fluctuations within sectors or jobs, employment reallocation across industries and demographic factors. Figure 4.5 charts the evolution o f wages for different educational groups, while Figures 4.5a and 4.5b show relative wages - or skill premia- for high school and college relative to primary workers, and college relative to high school by gender. This second form o f presentation controls for the business cycle and highlights changes in inequality, very relevant inthe context o f this study. 62 Figure4.4 Reallow-skilled wages up since 1997, but still below their 1991 level I.a The skillpremium fell after the macro-economiccrisis. The decomposition of changes o f the differentwage premia into within and betweeneffects, shows that inthe first period (1991- 95), the largest increases in wages occurred in sectors with more educated workers (so that the shifts in employment betweensectors increased inequality). On the other hand, these same sectors started recruiting more workers with primary education (so that the evolution within sectors was inequality decreasing). This may be the result o f biasedtechnological change. A similar pattern i s repeated inthe subsequent periods, but with between components loosing importance, making the overall change more favorable to low skilled workers. As a result, the skill premiumi s increasing ineducationfor the first period, butdecreasing duringthe latter. Since all wages increased during the 1999-2003 period, this may indicate that a reallocation of resources in favor of low skilled workers actually occurred in that period. That, i s the lower skills gained more than the higher ones. In terms o f experience, the pattern described above i s clearer for younger workers. The North American Free Trade Agreement (NAFTA), thus, may have had a positive impact on the incomes of poor workers, who have a skeweddistribution o f skills toward primary education. Real wages are still lower for all groups than in 1991, but the low-skilled gained ground in the economic recovery since 1997. The pay o f all educational groups dropped precipitously following the Tequila Crisis and thenrebounded after 1996, at a faster rate for less- educated workers (Figure 4.4). The recent equalizing trend in skill premia i s more apparent from relative wages (Figure4.5). After increasing until 1996, the relative wage advantage o f menwith completed high school or college over those with a lower qualification has been falling (since 1997). The pattern i s similar, but less pronounced for women. Both women with high school and college educations saw their wages decline relative to their primary-educated peers. But college- educated women managed to maintain - though not increase their advantage over the secondary group. While for men the different premia had returnedto roughly 1991 levels in 2003, college- educated women finished the period earning greater relative wages than both the less-qualified 63 groups. It also seems that the recovery was less equalizing for more-experienced workers than younger ones inthe periodstudied. Figure4.5 Relative skill premia have fallen since 1997 to near 1991levels, except among better-educatedwomen Wage premia, Men Wage premia,Women - - 1990 - 1995 2000 2003 1990- 1995 2000 2003 Year Year -High Sch I Primary College1 Primary -High Sch I Primary College I Primary College1High Sch College1 High Sch Source: Own calculationbasedon ENEU. The change intrends ofworker remuneration by skill level can be largely explained by shifts in labor demand before and after 1996. From 1991 to 1996, labor demand favored high-skilledworkers, due to a technological upgrade related to trade liberalization, affecting the entire economy. As a result, the relative wages o f low-skilled workers fell, as demand for their labor declined, amid a contraction inthe sectors inwhich they are concentrated. The consensus o f many studies on the period (see Revenga 1994, Cortez 2001, and Cragg and Epelbaum, 1996) i s that tariff reductions and trade expansion, and their interaction with technical change, caused the increase inpremia, which closely follows the pattern observed inthe US. Other authors (Feenstra and Hanson, 1997) studiedthe effect o f foreign direct investment(FDI) in the form o f maquilas, and they showed that those investments can account for more than 50 percent o f the initial increase in wage inequality. After 1996, low-skilled workers saw their relative wages increase amid an expansion o f specific sectors which employ them heavily, such as commerce and construction. Growth in the relative supply of highly educatedworkers, due to a marked increase in educational attainment, also put downward pressure on the returns to this group, contributing to the equalizing trend after 1996. Figure 4.6 presents the relative supply o f high school and college workers with respect to primary workers, as well as the ratio o f college to high-school workers, by gender. The trend only falters immediately after the crisis, which prompted the higher employment rate discussed above. For men, the ratio o f high school to college actually decreased, while for women it displayed a weaker growth rate. However, the fact that for the first half o f the 1990s skill premia and relative supply are positively correlated illustrates the strength o f other influences, particularly labor demand, on relative wages. This apparent reallocation o f resources in favor o f low-skilled workers, especially during the 1999- 2003 recovery, i s clearer for men, for whom the different premia regained their initial level in 2003. However, for women and overall, the reduction in skill premiainthe second periodwas not enough to offset the increase duringthe first one. 64 Figure4.6 The relativesupply of educatedpeoplehasexpanded Relative Supply, Men Relative Supply, Women - College/ High Sch A College / High Sch Source: Own calculationbasedon ENEU. STRUCTURAL ELEMENTS OF THE LABORMARKET AFFECT THE POOR'S ABILITY TO GET JOBS Mexican unemploymenthas historically been low Unemployment in Mexico has been remarkably low, mainly due to the quick response of real wages to changes in output. Evidence suggests that falling real wages, rather than employment levels, accounted for much o f the adjustment after the crisis o f 1994/1995. Real wages fell by almost 30 percent (Figure4.7), with the largest declines experienced by those with low skills. With the exception of the Tequila crisis in 1995, Mexico's unemployment rate was never over 5 percent during the period, despite constant increases in labor participation. Unemployment rates are in fact very low when compared to those observed in Latin American countries and most developed economies, including Colombia and Chile, which experienced similar or stronger episodes o f growth than Mexico. Even when compared to Asian countries, which also heavily compete in the U S market, Mexico's unemployment remains low. Moreover, unemployment duration has been also low, with most skilled and unskilledworkers finding ajob within a six-month period. 65 women 6000 ~ 2000 ~ 0 4 , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , ,. . . . . . . . . . . . . . . . . . . . -Male Female ,, ,,Total ... ~ Source: Own calculationbasedon ENEU. However, lowerflexibility may herald higher equilibrium unemployment The ability of wages to adapt to changes in economic activity and absorb shocks in times of recessions seems to have declined since the late 1990s. While Mexico experienced a decline inboth Gross Domestic Product (GDP) and employment growth between2000 and 2001, mean real remunerations kept increasing. During the period 1987-99 the employment rate increased by about seven percentage points, partly because households responded to declining incomes by expanding labor supply, as discussed. But after 2000 this trend was reversed. It appears therefore as it wage flexibility hadundergone a structural change when inflation declined to single digits in the late 1990s. If it i s indeedthe case, this will limit the ability o f the labor market to keep unemployment as low as during the previous decade. This poses a challenge to policy-making since it could imply labor markets will increasingly adjust via quantities (i.e. by creating unemployment). This is further supported by an analysis of the "Okun coefficient", which measures the degree of responsiveness of wages and employment to changes in output.60Figure 4.8 shows the 8 year moving average o f this coefficient to have declined significantly over time, meaning that any given decline in output translates into smaller declines inwages and thus larger falls in employment. More importantly, although the trend i s downward throughout the period studied, it becomes steeper around 1999. This suggests that the ability to adjust via wages (represented by the size of the wage Okun coefficient) has diminishedduringthe last decade and, importantly, that an abrupt change occurred around 1998/9 (for details see Bendini, Gracia and Santamaria, 2004). 60The Okun coefficient is defined here as the change in wages or employment associated with a change in employment. 66 Figure4.8 The responsivenessofwages to changesinoutput has declinedsharply 1 2 4 The wage Okun coefficient measures the change in wages associatedwith a change in output. The dates in the x-axis correspond to the preceding X-year periodas these are X-year rolling coefficients. Source: Bendini, Gracia and Santamaria (2004). Institutional and regulatoy constraintsfurther limitpoor workers' accesstoformal jobs Institutional factors may be distorting Mexico's labor market, including stringent firing regulations, hiring modalities, promotions and provisions for shutdowns and downsizing. Maloney (1999) shows that inefficiencies in formal sector protections and low levels o f labor productivity make the informal sector a desirable and optimal alternative to formal employment. International evidence supports this view, by showing that excessive or very rigid regulations - even if well-intended - curtail the creation o f formal employment, as employers seek to circumvent costly and complicated requirements by hiring workers informally. Workers also, if they perceive that formal employment is too costly while the benefits are not clear, may prefer informal employment accompanied by other forms o f social insurance. This particularly affects young and less-educated workers, whose productivity tends to be lower, making them less likely to be hired in the formal sector (Pages and Montenegro, 1999). Furthermore, the "insider/outsider" literature, asserts that strict rules protecting job security strengthenthe position o f formal unionized workers (the insiders), which they exploit to increase their wages, which in turnboosts pay inthe broader formallabor market, encouraginginformalhiring. International experience suggests a strong link between labor market rigidity and the degree of informality. Figure4.9 shows a strong positive association, confirmed by a highly significant correlation coefficient o f 65 percent, between different countries' informality rates and a labor rigidity index based on the costs (implicit and explicit) o f firing workers. Most Latin American countries appear in the upper right-hand side o f the graph. Mexico i s above the regressionline, suggesting informality i s actually somewhat higher than expected given its degree o f labor market rigidity. 67 Figure4.9 Informality rates rise with labor market rigidity 0b o 0 0 5 1 1 5 2 2 5 3 3 5 5 l a b o r m a r k e t rigidity Source: Heckmany PagCs (2000) and ILO-LABORSTA database. High non-wage costs also appear to reduce total employment. Statistical analysis o f employment trends inthe manufacturing sector61 produces a negative correlation: a decline inthe proportion o f non-wage contributions brestaciones sociales) over total remuneration (the non- wage ratio) i s accompaniedby growth in employment. Results from the estimation show that a 10 percent increment in the non-wage ratio produces a contemporaneous effect on employment o f about 0.4 percent. The long run effect rises to 4.4 percent. However, controlling for the endogeneity o f non-wage costs, the estimates increase 10-foldfor both the short run and long run; while the former reaches 3 percent, the latter exceeds 40 percent. These results suggest that non- wage costs do have a significant impact on employment Estimates using total number o f hours worked as the dependent variable show that, controlling for the endogeneity o f non wage costs, the coefficient o f interest becomes insignificant, which suggests that firms avoid hiring new workers and substitute hiringwith a heavier workload per worker. The 1997Social Security System reform may have had apositive impact The 1997 Social Security reform, which replacedthe pay-as-you-go pension systemwith one based on privately managed individual accounts and overhauled the health financing system, appears to have stimulated formal employment generation, mainly because o ftwo reasons. On the one hand, the reform increased public contributions to the financing o f both systems (as a basic pillar), effectively reducing the burden on employers and employees and introducing clearer incentives for them to make contributions "on top" o f this basic pillar. In line with this, contributions in general were made more uniform across wage levels, scrapping the existing "levy" on human capital under the former system. On the other hand, the reform strengthenedthe link between contributions and benefits. For example, the portability of benefits across jobs increased inthe pension system, which raisedthe incentives to contribute to the system and, thus, to become formal. Statistical analyses conducted for this report show that the formal sector grew more than the informal sector in the period right after the reform, which provides evidence that the reform may have had a positive impact on formal employment generation. Other studies 61 The Encuesta Industrial Mensual (EIM) was used for the analysis, which provides information about medium and small firms in the manufacturing sector. This survey contains data from 1987to 2004 on nine aggregate sectors (Food, Apparel, Wood, Paper, Chemicals, Non-metal, Metal, Machinery and Equipment and Others). The sample allows the use of a long time-series (more than 200 observations by sub-sector). conducted in the country, however, do not find evidence o f this fact. Indeed, the statistical analysis shows that the formal employment grew 9 percent more than that o f the informal sector in the period right after the reform (Table 4.6). While there are no significant differences by gender, estimates decrease with experience. The cost o f becoming formal seems to increase with age. Primary and secondary educated workers appear to have benefited the most. Since poor workers are over-represented in those educational categories, these results imply that the poor may have been disproportionately affected by pre-reform rigidities and benefited more from the introduction o fthe reform than non-poor workers. The reform was expected to primarily affect formal workers already receiving IMSS benefits. Using informal workers as a control group, a simple differences-in-differences estimator was applied to the data. Strong assumptions are needed in order to maintain the validity o f this estimator. In particular, it i s assumed that no changes in formal versus informal employment would have occurred without the reform. In other words, the relative cost o f employment varies only because o f the reform. A panel from the ENEU, with observations before and after the reform was used. The panel starts inthe last quarter o f 1996 and covers all o f 1997. Table 4.6 The socialsecurity reform ad apositiveimpacton formalemployment Differencesin differencesestimata s ofthe 1997reform,by sub-groups 0 v e r a l l 0 . 0 9 2 M e n 0 . 0 9 7 W o m e n 0 . 0 8 2 S c h . I E x p . S c h . I E x p . P r i m a r y 10 - 9 0 . 0 9 2 P r i m a r y l o - 9 0 . 1 2 0 1 0 - 1 4 0 . 1 1 9 1 0 - 1 4 0 . 1 9 8 1 5 - 1 9 0 . 1 4 4 1 5 - 1 9 0 . 1 8 6 2 0 -. 0 . 0 7 4 2 0 -. 0 . 0 5 9 S e c . I n c 0 - 9 0 . 2 0 8 S e c . I n c 0 - 9 0 . 1 8 4 1 0 - 1 4 0 . 1 5 3 1 0 - 1 4 0 . 2 0 5 1 5 - 1 9 0 . 1 4 8 1 5 - 1 9 0 . 0 6 8 2 0 -. 0 . 1 2 1 2 0 -. 0 . 0 1 5 S e c . C o m p . 0 - 9 0 . 1 7 6 S e c . C o m p . 0 - 9 0 . 1 2 8 1 0 - 1 4 0 . 3 3 1 1 0 - 1 4 - 0 .O 6 1 1 5 - 1 9 0 . 0 1 9 1 5 - 1 9 0 . 0 2 8 2 0 -. 0 . 1 0 3 2 0 -. - 0 . 1 3 7 C o l l e g e I n c . 0 - 9 - 0 . o 1 0 C o l l e g e I n c . 0 - 9 0 . 3 5 8 1 0 - 1 4 0 . 3 8 3 1 0 - 1 4 0 . 7 9 8 1 5 - 1 9 0 . 2 6 8 1 5 - 1 9 0 . 1 7 9 2 0 -. - 0 . 2 6 8 2 0 -. 0 . 6 1 5 C o l l e g e 0 - 9 0 . 0 4 7 C o l l e g e 0 - 9 0 . 1 0 3 1 0 - 1 4 0 . 1 7 2 1 0 - 1 4 - 0 .O 0 6 1 5 - 1 9 0 . 0 2 2 1 5 - 1 9 - 0 . 1 7 3 2 0 -. - 0 .I5 9 2 0 -. - 0 . 1 5 2 Coefficients show the % change in ormal employment att butable to the so( r l security reform. Source: Own calculationsbasedor ZNEU. Box 4.1. The 1997SocialSecuritySystemReform Launched in February 1997, the Social Security Law or (Ley de Seguridad Social, LSS) introduced important and positive changes to the hnding and provision o f social security services. Estimates suggest that the reforms, which linkedcontributions more directly to expected benefits, had a positive relative impact on formal employment, especially o f low-skilled and young workers, thereby improving the relative position o fthe poor. 69 As a result o f the reforms, health insurance system established an almost-uniform contribution based on the minimum wage, eliminating the tax on human capital implicit in the former system, which varied payments according to salaries paid, even though the same services were available to all. Contributions above the minimum wage level were therefore seen by employers as an extra tax with no benefit attached. In scrapping this "levy" and sharply raising its own contribution (from 0.6 to 13.9 percent o f the minimumwage per contributor) the government created a powerhl stimulus for formal employment. The pension system was also overhauled. Underthe previous collective system the value o f a pension dependedon base salary and length o f contribution time according to a complex and somewhat arbitrary table that was largely independent o fthe actual amount paid in.Under the new scheme, the country's private-sector employees must contribute to individual accounts through the private capitalization program. The contribution made by the federal government is an important innovation as it reduces incentive problems associated with combined systems that introduce individual accounts but still guarantee minimum pensions. Mexico's privately managed pension system i s currently the largest in Latin America due to country's size, the major transition plan in place and the scope o f the new program. By the end o f 1997, Mexico's 17AFORES (pension fund managers) had enrolled more than 10.5 million workers. Despite the positive impact o f 1997 reforms further major changes are neededto reduce Mexico's high non-wage labor costs and further tighten the link between contributions and benefits to promote formal employment. Non-wage expenses paid by the country's employers are estimated at 47 percent o f the payroll (Hernandez Laos 1998), the heaviest such burden in Latin American apart from Colombia (53 percent). Chile (21 percent), Argentina (around 40 percent), and Brazil (35 percent) all display lower non-wage costs. In Mexico, the areas that require attention include housing programs which constitutes a sizeable 5 percent o f the payroll and premia to cover professional risks. PROGRAMSFORTHE UNEMPLOYED-BETTERTARGETING COULD IMPROVE IMPACT The government has a number of active labor market policy interventions, mostly offered by the Secretaria del Trabajo y Previsibn Social (STPS). These include occupational training programs, labor intermediation services and direct and indirectjob creation programs. A study of these programs, based on official and academic research already produced, was conductedby Montes and Santamaria (2004) and i s summarizedbelow. OccupationalTrainingPrograms The poor have relatively little access to training programs. Occupational training programs are key for helping labor supply keep up with the changing demand. Because firms' investment in training is generally insufficient, the role of the state is important in providing financing and a regulatory framework that provides workers, in particular the poorest, with tools to remain competitive in the labor force. However, in Mexico, the distribution of occupational training services i s very unequal (World Bank, 2000): from the poorest 10 percent o f the population, only 1.5 percent has attended some training program, whereas among the wealthiest 10 percent, the participation rate is 32 percent. This can be partly explained by the fact that the poor are mainly employed in small or informal firms, which rarely provide training. Evidence suggests that only 17 percent o f the poorest workers who received training did so in the firm 70 where they worked. Results also show that those programs with larger private sector participation (either privately offered training courses or courses offered in firms) obtain the most positive results. The Programa de Apoyo a la Capacitacidn (PAC) subsidizes the hiring of independent instructors. Efforts are made to match firms with local training providers that correspond to their needs. Participant firms exhibit increased investment in training, improved labor productivity, higher capacity utilization rates, reduced personnel rotation, and improved production techniques, among others. The effectiveness of the Sistema de Capacitacidnpara el Trabajo (SICAT) varies greatly across sub-programs and regions. For instance, where there is a large participation of private institutions in the provision o f training courses, results have been very positive (Calderbn and Trejo, 2003). In medium and large enterprises, SICAT has had a positive impact on participants' income and employment and the implementation of training programs has helped these firms' courses' curricula to remain relevant. However, evidence suggests that in some cases, firms have used these programs as a substitute for personnel rotation, and that the evaluation o f the programs objectives' fulfillment has not always been properly conducted, inwhich cases, the implementation o f operational rules should be enforced. On the other hand, results have not been so positive in small and medium enterprises, which are precisely the firms that need the most assistance to innovate and promote workers' training, as well as the ones exhibiting very high rates o f personnel rotation (De Ferranti, et al., 2003). Inthis type o f firm, the emphasis should be placed in the selection o f instructors, which seems to be more relevant than in larger firms. The self-employed do not seem to have benefitedas much from the program. Evidence indicates that a larger participation o f private institutions would be desirable, as they offer more flexibility to choose training courses that best suit individuals inthis highlyheterogeneous group. The government has taken the right steps in promotingprograms that are based on the distributionof vouchers since that allows choice by beneficiaries.Although no evaluation has yet been conducted of voucher-based programs, the introduction o f this element in the programs mentioned above could greatly improve their quality and relevance by introducing competition among private and public providers. More generally, it would be beneficial to improve the targeting and relevance o f occupational training programs, so that they effectively facilitate poor workers' access to better jobs. Training programs for youth should be designed to address the specific needs o f young workers and ease the transition betweenschool and the labor market. It i s imperative that such policies stipulate a transparent use o f training funds to avoid corruption, as well as promoting competition among private and public training providers. State monopolization o f occupational training results intraining courses that do not respondto demand. Labor Intermediation Sewices Efforts in education and training can be complemented by the strengthening of labor intermediationservices to increasethe access ofpoor, low-skilledworkers to labor market information and give them access to networks and connections that may boost their chances of finding goodjobs. Labor intermediation services should be designedto facilitate the matching of candidates with vacancies and the referral o f workers not only inthe formal sector, but also more unregulated forms o f employment, such as micro enterprises, self-employment, and small business development, which, as mentioned in the Chapter, employ a large share o f the urban poor and typically have more restricted access to labor market information. Though the experience with current programs and the innovations introduced by the government recently, 71 such as CHAMBATEL or CHAMBANET, has been generally very positive, evidence suggests that in practice, these programs are more accessible to skilled workers than non-skilled workers, or are designed to cover the unemployed who used to work in the formal sector (SAEBE). The utilization o f public resources by people most likely to find jobs, even without these programs, underminesthe effectiveness o f labor intermediation services andreduces their social benefits. Worker placement in current programs is fairly high, but targeting could be improved.Results indicate that the level of worker placement of current programs is 33 percent o f the effective demand, a percentage that would be lower if the population that has no access to these services was taken into account. Compared with other countries' experiences (30 percent in Chile, 10 to 25 percent in Europe), this level o f placements can be considered a success. An evaluation o f the SAEBEin 2003 showed that workers with very low education levels, the youth, the elderly and those with the largest number o f dependants (a possible proxy for the poor) were among those who benefited most from the program. These results suggest that a broader application o f the benefits o f SAEBE could increase the level o f effectiveness and transparency in the use o f resources. Such change in the program will probably present operational challenges, however, that should be resolved in the framework o f the Sewicios Estatales de Empleo and access to an increased budget. More generally, a greater emphasis should be placed on targeting the neediest groups, such as the low skilled, the poorly educated, the poor or victims o f discrimination, who typically face the highestbarriers tojob market entry. Direct and Indirect Creation of Employment Many of the labor market policies currently in place have been designed in conjunction with policies to combat poverty, such as the support of micro-enterprises. Programs designed to finance micro-enterprises in general yield positive results, although they present operational challenges. The private sector needs to play a larger role in advising and training firms chosen by the state to receive financial help. Programs designed to finance micro- regional projects also produce positive results, as proved by the Proyectos de Inversibn Productiva and would also benefit from active private sector participation. The promotion o f lucrative private activities in areas where funds are applied, such as the coordination between micro-enterprises and larger firms, could also be beneficial. Employment creation should focus on those groups that have the biggest difficulty in accessing job opportunities. This practice should be carried out with caution, though, as a broad application o f the same could lead to inefficiency and corruption. CONCLUSION The poor's employment situation worsened in some respects during the 1990s. The preceding analysis o f the relationship between labor markets and poverty in the main 16 urban areas o f Mexico shows how developments in the labor market affected the employment opportunities o f poor workers between 1987 and 2003. The poor's human capital endowments and characteristics (both observed and unobserved), combined with economic trends, low total factor productivity, and institutional factors, caused a significant deterioration in poor workers' employment - and consequently income - at least until 1996. The second half o f the period in view presented an improvement inthe situation o f the poor. Butthe recovery i s not complete and further attention andpolicy interventions arerequired. 72 During the period analyzed, poor workers were increasinglyconcentrated in sectors with meager performance and slow growth. Most o f the poor were in low quality occupations, such as low-end self-employment (SEWI). A 40 percent increase inthe share o f poor households inthis categoryresultedinsignificant decline inreturns to labor. Although income levels dropped significantly across the board after the Tequila crisis, the wage and unemployment declines were larger than average for low-skilled workers, triggering a surge in poverty that peaked in 1996. Results showed that labor demand fluctuations largely account for the two very different trends that low-skilled workers' remunerations followed before and after 1996. Between 1991 and 1996, the relative demand for high-skilledworkers increased, mainly as a response to technical change. Additionally, sectors that employed low-skilled workers contracted, resulting in a drop in the demand for low-skilledworkers as well as intheir relative wages. After 1996, relative demand for low-skilled workers expanded and fell for the high- skilled. This change in labor demand, coupled with a large increase in the supply o f skilled workers (especially women), resulted in a significant improvement inthe relative remuneration o f the poor. Wage levels recovered after 1996, but the improvement was not sufficient to regain the levels o f 1991, especially for the very poor. However, lower wages were partly compensated for by an increase inpoor workers' participation inthe labor market. Institutional factors in Mexico's labor market appear to have played an important role in curtailing formal employment opportunities, particularly for the poor. However, the 1997 social-security reform may have relaxed labor-market rigidities somewhat. The analysis o f worker andjob flows in the formal labor market shows them to have accelerated in recent years coinciding with the introduction o f the reform, thereby suggesting that costs may be falling. These results, combined with evidence that transitions to and from the formal sector are becoming more common (Budar and Garcia-Verdu, 2004), indicate that the labor market i s becoming increasingly fluid. Furthermore, although formal employment growth rose substantially in 1997 when the reform was implemented, this phenomenon was not observed in large or manufacturing establishments, indicating that the changes in employment occurred largely in firms more likely to employ poor workers. In addition, the country's low total productivity growth is affecting poor workers' ability to access better quality jobs and thus improve their labor earnings. Sluggish productivity growth i s slowing overall economic growth, hampering competitiveness, and jeopardizing Mexico's ability to take advantage o f the opportunities created by NAFTA. The fact that Mexico's labor market seems to be losing its ability to adjust to output shocks through wages combined with Mexico's declining ability to improve factor productivity may seriously hurt the country's competitiveness ininternational markets, particularly the US. Rising unit labor costs in 1999-2001 threatened competitiveness and long-term sustainability of job creation. In that period, real wages have been steadily rose in spite o f declining GDP and low labor productivity. As discussed in Section 11, Mexico was able to remain competitive in the U S market until 1998, despite slow TFP growth, thanks to its low dollar- denominated wages. However, the country's increasing vulnerability became evident during the last recession, when wages and unemployment increased in spite o f a large GDP fall, affecting export-oriented firms in a disproportional manner. If factor productivity i s not improved, it will become increasingly difficult for Mexican exporters to remain competitive abroad, which can have negative repercussions for the labor market as a whole, and particularly the poor, who have greatly benefited from export growth since 1996. With reduced wage flexibility and continued low TFP growth, the low unemployment that characterized the country untilthe late 1990swill be 73 more difficult to sustain, right at a time when labor participation, particularly among the poor, i s increasing. Policies that foster the expansion of export sectors, together with targeted interventions to favor productivity growth in small and informal firms, could increase the better jobs open to poor workers. However, for any such policy to work, it must address issues that constrain productivity growth. Furthermore, because the current low levels o f productivity growth are a major hurdle for formal sector expansion, improving productivity would go a long way toward increasing formality and, consequently, enhancing the access o f poor workers to better qualityjobs. An important step in this direction is to improve the investment climate by raising the quality of institutions and services for business, and simplifying the regulations and procedures for company registration, functioning, and growth. Deepeningthe 1997 social security reform in the areas where there i s still room for action could further smooth out labor-market rigidities, while continuing to facilitate formal employment (one such areas is, for example, the contributions to the housing fund, which are almost completely de-linked from benefits). In addition, labor provisions that hinder productivity growth, such as hiring modalities and promotion-related provisions, dispute settlement mechanisms, and termination o f employment and severance payments (both individual and collective), should be addressed soon. Current hiring modalities, which reflect the excessive emphasis of Mexican labor regulations on job stability, depress formality and productivity growth. More flexible hiring modalities could be introduced to eliminate disincentives against hiring young workers and women and facilitate their insertion into the labor market. In addition, extending fixed-term contracts couldbe beneficial, particularly for informal workers and firms that couldtransition into formality. Regulation o f promotions should be eliminated, at least for private workers. Regarding the dispute settlement mechanism, there i s ample agreement that it i s not working properly, but little consensus on the best way to reform it. Among the agreed priority areas o f action are: changing the composition o f the Juntas de Conciliacibn y Arbitraje and making them more independent; introducing effective incentives for agreement between employers and employees (deadlines and some regulations on how the decisions should be made, for example); and revising the current regulations to discourage the initiation o f disputes. The simplification of onerous and costly dismissal and severance-payment regulations is crucial for improving labor productivity and transparency. Such regulations tend to generate high degrees o f uncertainty for employers and workers and effectively hamper the efficient use o f labor along the business cycle. The distinction between just and unjust dismissal should be discontinued and severance payments should be set at lower levels. Given that the system o f unemployment protection currently in place (severance payments) neither protects informal workers, nor provides adequate service to its target population, formal workers (Dada, 2001), a more inclusive labor protection scheme should be developed, that covers the most vulnerable and does not distort the functioning o f the labor market. This could be jointly funded by employers, workers, and the government through individual accounts to provide supplemental income duringjob-search periods, thus facilitating easier labor market adjustments. If eligibility is well defined and costs are reasonable, this type of system may encourage employers and employees to register contracts that they mightnot otherwise. M o r e generally, it i s vital to develop a system of social insurance that effectively protects the poor. The current health and pension systems, which are designed to cover formal workers and exclude most o f the poor from sufficient income protection during a health shock or 74 in old age, should be re-designedto ensure that poor workers -formal and informal - receive adequate coverage. This calls for exploring ways to expand coverage, either by adding benefit programs for the informal sector, or by de-linking social security benefits from labor-market status and moving to a single benefit system with universal coverage. Funding could involve a combination o f redirecting subsidies away from the non-poor (Le. reform o f IMSS/ISSSTE) and public funding. Increasing household-member participation in the labor market effectively improves the household's income level, yet female participation rates in Mexico remain among the lowest in Latin America, despite some increase that took place during the period considered. It i s imperative to implementprograms that foster the integration o f female workers into the labor force, as well as to expand those currently in place to ensure provision o f services, such as childcare, which facilitate women's participation in the labor force. As such, the introduction o f training programs for women and childcare facilities in the Mujeres Jefas de Familiapart o f the overall Hhbitat program i s a welcome initiative (see chapter 2 for discussion). Beyond these policy solutions, the root causes o f the low female participation rate should be examined, as it i s possible that some degree o f cultural bias, discouraging women's participation inthe labor force, is further obstructingprogress neededonthis front. Providingpoor workers with opportunities to improve their skill levels, as well as facilitatingtheir access to informationonjob prospects could greatly expand their access to decent jobs. As mentioned inthis chapter, the ability of poor workers to get betterjobs is greatly hampered by their endowments and limited access to labor market information. Evidence presented in the Chapter indicates that returns to higher education helpedto improve the poor's relative situation in distributional terms, in spite o f their limited ability to access this level o f education. To a large extent, the limited enrollment o f people inhigher education results from the low demand for it, which in turn reflects problems in secondary education, particularly in terms o f coverage and quality, which should be addressed. Educational policies should be designed to improve the quality and coverage o f secondary schooling, as well as to deepen efforts already underway that promote access by the poor to tertiary education. These efforts could be complemented with incentives to make it more feasible for the poor to obtain higher education, such as the provision o f a combination o f credits and subsidies to cover tuition and associated costs along the lines o f the process carried out under the loan financed by the World Bank. Jbvenes con Oportunidadesi s intendedwith this purpose (Chile and Colombia also offer some interestingexperience inthis regard). Occupationaltraining and labor intermediationservices targeted to the poor need to complement efforts at increasing education, especially in the short to medium term. Increasing access to higher education alone, however, will not improve the situation o f the poor, as the benefits o f education take time to appear. Moreover, since the skill premiumhas fallen since 1996, such initiatives cannot be considered a silver bullet for improving the situation o f the poor. Thus, occupational training and labor intermediation services can be instrumental inhelping the poor in the short to medium terms to access decent jobs. A recent analysis o f current government programs (Montes and Santamaria, 2004) suggests that the poor are not the main beneficiaries o f current programs. A staged approach should therefore be adopted, including a more comprehensive evaluation o f existing programs, with an assessment o f what works best to help the poor access better jobs in order to help determine which programs have the greatest impact, and therefore should be continued and/or revamped, and which ones are contributing insufficiently, andthus should bereformed. 75 Safety nets are needed that protect the urban poor and help them mitigate risks, particularly in times of crisis, given that the current unemployment protection system only protects formal employees and that the Mexican labor market may be increasingly absorbing output shocks through employment rates (and consequently higher unemployment rates). A scheme to protect workers confronted with unemployment mentioned above couldbe part o f this. In addition, programs that particularly target the poor in times of crisis should be designed. Workfare i s one approach commonly used - indeedMexico had one such program operating in rural areas until quite recently (Programa de Empleo Temporal or PET). Chile, Argentina, Thailand and Korea are countries that have successklly relied on workfare programs to cope with the impact o f macro-economic shocks. Box 4.2 describes Argentina's Trabajar program, which was widely considered an effective design, while Box 4.3 discusses what experts agree are the key features ofgoodworkfare programs. Box4.2. Trabajar:A GoodExampleof a WorkfareProgram Trabajar, a workfare program, was established by the government o f Argentina in 1996 to help address a sharp rise in unemployment. The program aimed to help through the provisions o f short-term work at relatively low wages as well as through the execution o f small infrastructure facilities inpoor neighborhoods. Subprojects were proposed by local governments and NGOswho had to cover non-wage costs. Subprojects were designedto be labor-intensive and relatively small with average project size o f less than AR$100,000 and employing an average o f 20 workers. The projects were chosen on a competitive basis according to a points system set by the center. Points were given according to the poverty rate for the municipality in which the proposed project was located, the type o f project proposed (usually rehabilitation, expansion, and new construction o f community or public infrastructure such as water, sanitation, housing, roads, schools and health centers), willingness to work for a wage below the maximum and how much the area had already received from the program. The wage rate was set at the same low level for urban and rural areas -AR$200/month which was lower than the average monthly earnings for the poorest 10 percent o f households in Greater Buenos Aires (AR$263). In 2000, wage rate was further lowered to AR$160/mo, which i s below the minimumwage. There was no legal impediment as the payment was termed "economic assistance" rather than a wage. Skilled and semiskilled workers needed to execute the projects were hired as "foremen" and paid a slightlyhigher wage. The program was initially restricted to heads o f households butthis was changed so that the only requirement for participationwas to not be receiving unemployment benefits or be participating in any other employment or training program. Trabajarwas found to be well targeted with the distribution o f gains decidedly pro-poor: 80 percent o f participants were from the poorest 20 percent o f families nationally and 50 percent from the poorest decile. The relatively low wage clearly made the program unattractive to non- poor. Targeting performance was evaluated early on in the program, and adjustments were made to improve inthe less well performing provinces. Trabajar encountered a number o f difficulties in larger municipalities (e.g. 100,000 or more). First, large municipalities found it to difficult to insert Trabajar projects -the review and evaluation o f which occurred on a monthly basis- into municipal and master plans that had already long been formulated and which mostly included fairly complex works. In addition, Trabajar projects were small and larger municipalities found they neededlarger projects that could fit into the large-scale projects that made up their capital investment plan. Also, in larger municipalities, infrastructure projects fell under the purview o f the public works agencies, not the social assistance agencies, as was the case for smaller municipalities and often the public works agencies simply felt the benefits from a Trabajar project did not outweigh the cost of proposing and implementing one. Finally, larger municipalities found it easier to contract out the work rather than employ low skills workers inneedo fmore supervision. 76 The Trabajar program staff had developed a proposal to address these issues which involved changing the project cycle for larger municipalities so they would have an opportunity to work with a projected financial envelope o f Trabajar funds, and integrate them into master plans, as well as to allow financing o f a series o f small stand-alone projects that could be part o f a larger infrastructure project. These changes were never implemented, however, since the new government replaced Trabajar with the Jefes de Hogares program. It should be noted, however, that Trabajar program was still successful in large urban areas: there was always 100 percent uptake and demand for projects outweighed supply. Source: Fay, CohanandMcEvoy (2005), Ravaillon (2000), JalanandRavaillon (1999). Box 4.3. The KeyDesignFeatures of a Good PublicWorks Program The following are the key features that needto be included in a workfare program for it to realize its full potential as a poverty-reducing and risk-coping instrument: 0 Wage level. The wage rate should be set at a levelwhich i s no higher, and preferably slightly lower, than the prevailing market wage for unskilled manual labor in the setting inwhich the scheme i s introduced. Eligibility. Restrictions on eligibility should be avoided; the fact that one wants work at this wage rate should ideally be the only requirement for eligibility. In particular, eligibility should not be restricted to household head - which constrains families own adjustment (Ravallion, 1999) and reduces workfare's effectiveness in cases where the shock is felt through a decline in real wages rather than through unemployment. In case where resources are limited some clear secondary targeting or rationing rule might be needed. Options include limiting eligibility to one person per family (but still allow the family to pick the person) limiting duration o f job; limitingjobs to families with dependents; community based targeting for who gets thejobs; periodic lotteries, etc. Labor intensity. The labor intensity (share o f wage bill in total cost) should be as high as possible. The level of labor intensitywill dependon the relative importance attached to immediate income gains versus (income and other) gains to the poor from the assets created. This will vary from setting to setting. Generally speaking, unskilledlabor costs account for 40 to 60 percent o f total project costs on a large and diverse portfolio o f high value works (the balance being skilled and semiskilled workers, equipment and materials and administrative costs). It i s possible to raise this share but this usually implies restricting the portfolio of works andor limiting thereturn onthe work. Female participation. Provision o f childcare or preschool services can improve participation by women (as well as provide employment opportunities for them). Also, women tend to benefit fiom piece rates or task-based wages, since that allows themto combine the work with their responsibilities at home. 0 Targeting of projects. The projects should be targeted to poor areas and try to assure that assets created are o f maximum value to poor people in those areas. Any exceptions -in which assets largely benefitthe non-poor- should require co-financing from the beneficiaries and this money should go back into scheme's budget. 77 Timing. In larger municipalities projects design needs to take into account the fact that municipal and master plans are annually determined and include fairly complex works that municipal authorities preferredto contract out. Options to address these issues were developed in the Argentine Trabajar program and included changing the workfare cycle to allow the municipalities to work with a projected financial envelope o f Trabajar funds and developing a series o f small stand-alone projects that could be part o f a larger infrastructure project (Fay, Cohan and McEvoy, 2004). Sustainability. Sustainability o fthe assets created requires that the program includes an assetmaintenance component. Source: Adanted from Box 1inSubbarao(2003) excent where otherwise noted. 78 5. BUILDINGUPBUFFERSAND STEPPINGSTONES: ASSETACCUMULATIONANDACCESS TO FINANCIAL SERVICES~~ Assets are at the core of households' strategies to survive, meet future needs, improve their lot, reduce exposure to shocks or minimize their consequences. And despite being constrained by limited income and few adequate saving mechanisms the poor do save - how else would they cope with shocks and the occasional need for lump sums o f money?63It can be potentially devastating for a poor household to stand without savings. Indeed, analysis o f spending adjustment during the peso crisis suggests that poorer households cut primary health care spending in order to deal with other needs deemed more pressing (McKenzie, 2003). Apart from the direct negative effects on well-being this could have potentially serious effects on individual andhouseholdsurvival over the long run. Savings are particularly important to Mexico's poor, who, as discussed in the previous chapter, are still largely excluded from public social insurance. Unemployment benefits and pensions are usually directed at employees in the formal sector and as such exclude the vast majority o f the poor - indeed only about 22 percent o f poor urban workers have social security. Thus, the poorest ten percent in urban areas receive only two percent o f all public resources spent on pensions insurance and less than one percent o f resources spent on hospital care; they receive only seven percent o f monetary transfers (Figure 5.1). As a result, the household's own assets remain the principal mean to preserve consumption against income shocks or over the life cycle. 62 This chapter was written by Marianne Fay and Sara Johansson, and benefited from input from Anna Wellenstein. 63 Other motivations for accumulating assets include savings for predictable future expenditures related to lifecycle needs, to accumulate resources for potential future opportunities such as investing in a business or acquiring ahouse, or to leavebequests to friends and family (Matin, Hulme and Rutherford, 1999). 79 Figure5.1 Social programsdo not favor the poor Urbanand rural poor's share intotal(urbanand rural)public expendituresin social sectors 301 0 Urban poor li 0 Rural poor21 25 6 25 ~ 20 ~ 6 9 6 7 7 3 56- r 6 2 5 0- 3 9 3 6 19 0 2 L Total Education Primary Hospital Maternal Pensions Monetary health care health Transfers 1. First decile. 2. 3ecile 1-3. The first urban decile and the bottom three rural deciles correspond to roughly the same population share (about 7.5 percent), all of whom are people under the extreme poverty line. While this graph cannot be usedto conclusively argue whether the allocation of resources i s optimal between rural and urban poor (other measure such as the poverty gap may be needed) it does suggest that it i s probably roughly appropriate. More importantly, it does show that the poor still receive a small proportion of overall social expenditures - especially in the form of pensions andhealth expenditures. Source: own calculationsbasedon data from the Public Expenditure Review 2004 (World Bank, 2004). As discussedin the companion study on social protection, the Mexicangovernment i s aware of this problemand i s trying to improve the reach of safety nets. In addition, it i s placing increasing emphasis on asset building in its newer anti-poverty programs. Thus Oportunidades aims to develop human capital in both rural and urban areas; and Hhbitat i s an urban program that focuses on physical assets (homes and infrastructure) while developing social capital inpoor urban communities. This chapter focuses on the physical and financial assets of Mexico's urban poor, the role they play in their coping strategies and how government programs can foster more and better asset accumulation by the poor.Other types of assets are discussed elsewhere inthe report: social capital and its importance in poor barrios in Mexico was discussed in chapter 3; human capital, which conditions poor urban households access tojobs andthe quality ofjobs they can access was addressed in chapter 4 on labor markets as well as in chapter 2, in relation to vulnerability and coping strategies. As to natural assets, since they are less important inthe urban economy than in the rural one - common property resources are rare and households are integrated into the monetary economy -they are not directly discussed here.64 The perspective on asset building differs between urban and rural areas. First, as shown in chapter 2, urban residents tend to be particularly vulnerable to macroeconomic shocks and in need o f some type o f buffer against such fluctuations. Second, cash availability i s crucial to coping with shocks (to ensure food security or pay for health care), more so than inrural areas where the local economy tends to be less monetized. Third, policy challenges differ. For example, informal savings or micro-credit institutions may find it easier to enroll clients in urban areas 64 There is nevertheless evidence o n the importance of urban and peri-urban agriculture inproviding access to food and incomes for the poor (Bakker 2000), pointing to the importance of access to land for more than housing purposes. 80 compared to rural areas due to higher population density; on the other hand, enforcement problems are greater due to greater mobility and looser social ties within urban areas. These differences call for differentiated approaches to institutions and policies. A main tenet of this chapter is that the poor's asset portfolio could be improved to provide better insurance and higher returns. Clearly the poor's capacity for buildingassets i s constrained by their lack o f resources. But the mix o f assets and their return i s also at a sub- optimal level due to constraints on the supply-side. Some assets are more suited to the specific needs o f the poor than others, but are presently not accessible, and the returns, risk-level and liquidity of others could potentially be improved. This chapter looks at the two main forms of assets available to the poor -physical assets, mostly in the form o f housing, and financial assets. Because o f financial exclusion, the poor tend in fact to over-invest inhousing and under-investin financial assets that, iftailored to their needs, couldbetter meet their particular needs (Box 5.1). The chapter is organized as follows. Section Iexamines physical assets, focusing mostly on the housing market from the perspective o f asset building.Section I1reviewsthe poor's access to financial services identifyingthe key barriers to better access. The final sectionprovides recommendations as to how public policy can support better asset building strategies o f the poor byremovingsome ofthe most important supply-sideconstraints. Box 5.1. What makes for a good asset for the poor -and are the assets of the poor good? For assets to perform their primary function as a store o f value they should provide adequate returns. Inflation presents the largest threat o f depreciation for financial savings. For physical assets, such risks dependon the depth and volatility o fthe secondary market on which the asset i s liquidated. Poor households generally have a higher share o f low-return assets than others, because high return assets may not be available to them, because o f risk-aversion, and because some o f these assets offer non-monetary returns, including building social capital and the possibility to access credit inthe future. Good assets for the poor also allow for highpequency operations of limited size without high transaction costs, andforflexibility, to help the poor manage small balances and deal with cash flow problems. This i s particularly important as they have limited access to formal financial savings and credit instruments. Examples o f assets which offer the possibility o f high fiequency operations are rotating savings pool such as the tandas to which 20 percent o f urban Mexican households belong (see discussion below). Financial exclusion and the reliance on informal financial tools and physical assets are likely to result inportfolios whose return and liquidity characteristics do not compare favorably to those o f the better-off. The poor -generally save in physical assets - housing as well as consumer durables - and those financial assets available to them. How do these correspond to the requirements listed above? Consumer durables (furniture, consumer appliances, bicycles) offer the opportunity o f investingrelatively small balances, and are highly liquid as they can be easily pawned or resold. Further, where labor is cheap, durables life and its value as an asset can often be extended with small improvements and repair. Housing is likely to be the most valuable single possession owned by poor urban households. Most poor urban homeowners rely on "progressive housing", whereby they first acquire a plot, legal or not, build a shack and then gradually improve upon it as resources become available. This process allows households to make small incremental investments over time. It 81 also frees them from the constraint o f having to generate a fixed sum for rent every month - especially important at times o f crisis. Furthermore, housing services can be monetized quite easily, by taking renters or additional household members in. Yet, as an asset, housing has some important negative aspects. Most importantly, it is unclear how liquid or buoyant housing markets are, particularly inpoor neighborhoods. The main motivation for usingJinancial assets tends to be risk management rather than expected returns. Lacking access to formal insurance, the poor typically rely on a combination o f savings and credit as alternatives. Formal financial institutions are generally not geared towards low-income clients who instead rely on various informal savings and credit mechanisms. Although these may offer important advantages in terms o f flexibility, returns tend to be low. In fact, while the micro-finance "revolution" has increased the poor's access to credit somewhat, no similar progress has been made for savings instruments. Yet, surveys show that a saving account i s the product that un-banked people are most interested in (after payment services -World Bank, 2003a and 2003b). Living in poverty in an urban slum it i s very difficult to protect savings from theft, inflation or the demands o f everyday living. And because the poor tend to hold their financial savings in cash or in informal arrangements they tend to be less protected than the rich against macroeconomic instability. SAVINGS STRATEGIES OF POORHOUSEHOLDS Savings are notoriously hard to measure, and even more so for the poor. First, because income i s hardto measure ingeneral (it i s usually underreported in surveys) and the poor rely more than the rich on self-employment and self-produced goods, the value of which is difficult to estimate (Kochar, 2000). Second, assets may have both consumption and investment value -the poor save by accumulating anything from consumption goods (such as food) to semi- durable (such as clothing) to durable goods (such as furniture, equipment or housing), as well as cash or contributions to informal institutions or networks o f reciprocal obligation^.^^ a As Matin, Hulme and Rutherford (1999)put it, the poor have three common methods inwhich they get access to the lump sums they need: selling assets they already hold (or expect to hold); taking a loan by mortgaging or pawning those assets; turning their many small savings into large lump sums -through savings deposits, through loans or through insurance. 82 Table 5.1. The poor are less able to save out of their incomethan the non-poor Median savings rates by level of education 1994 and 1996 1.Savings excludingspendingon consumer durables 1994 1996 %change N o schooling 5.8 -0.1 -101.7 Primary incomplete 6.8 0.7 -89.7 Primary complete 6.9 1.5 -78.3 Secondary incomplete 6.9 2.4 -65.2 Secondary complete 7.6 5.8 -23.7 Highereducation 13.0 7.0 -46.2 2. Savingsincluding spending on consumer durables 1994 1996 %change N o schooling 7.4 1.5 -79.7 Primary incomplete 8.0 2.6 -67.5 Primary complete 8.3 2.7 -67.5 Secondary incomplete 8.6 3.5 -59.3 Secondary complete 11.4 9.9 -13.2 Highereducation 15.6 10.3 -34.0 Note: Doesnot capture forced savings (e.g. contributionsto pensionschemes) or savings inthe form of housing. Source: Adaptedfrom SzCkely (1998). Keeping these caveats in mind, the available evidence confirms that indeed, in Mexico, even the very poor do save. Table 5-1 shows that prior to the peso crisis, the savings rate o f the very poor (as proxied by education) was around 6 percent, increasing by about one percentage point if consumer durables are included.66This i s clearly much less than better off households with savings rate more than twice that amount. Nonetheless, it does confirm the central premise o f this chapter - which i s that the poor do save even in the face o f limited resources and savings instruments. Table 5-1 also shows the poor less able to maintain their saving behavior in the face of serious crisis. Median savings rates were decidedly lower in 1996 after the peso crisis for all households, but the reduction was much more severe for poorer households, whether consumer durables are included or not. The remainder of this chapter looks at housing and financial assets in the asset management strategies of Mexico's urban poor. While consumer durables, as mentioned in Box 5.1, are likely to play an important role in the savings strategies o f the poor, there i s very little evidence on Mexico. What i s available, suggests that during the peso crisis adjustment in durables spending played an important part in the coping strategies o f the poor: the poorest households saw significantly larger reductions in spending on consumer durables relative to non- durables than didricher households (McKenzie, 2003). 66 Education levels, which tend to be strongly correlated with income and consumption levels, can be considered a better proxy for permanent income levels than current income, due to measurement issues arising from e.g. underreporting. Szkkely (1998). 83 HOUSING- THE KEYPHYSICAL ASSET OF URBANPOOR Homeownership, at 66 percent, is high among Mexico's urbanpoor, almost as high as for non-poor (70 percent) (Figure 5.2). These findings are corroborated by an analysis of homeownership in 31 poor barrios across Mexico, where on average some 82 percent o f households owned their houses (Box 5.2).67 Most however acquire housing through informal markets. This i s due to the fact that in Mexico formal housing i s only affordable to people who earn more than 3 minimum wages - the middle class and above. As a result, half o f all newly constructed housing in Mexico, and two thirds of the existing stock, are self-constructed (World Bank 2002a). Figure5.2 The poor are homeowners,too Percentageof populationlivinginhouseholdsaccordingto type ofhouseoccupancy Oownproperly Orenled Oborrowed .other I I Extremely poor M oderafelypoor N"np""r I Source: Own calculations basedon ENIGH2002. The value of a house as an asset depends on the existence and characteristics of the secondary market for housing -which conditions its exchange value and liquidity - as well as on the flow o f returns that it can generate. These are discussed inturn below. Box 5.2. Distribution of homeownership and different types of tenure in Mexico's poor barrios The survey o f 31 poor barrios in Mexico's 31 state capital cities carried out by SEDESOLin2003 and discussed in chapter 3 gives important insightsinto housing tenure inpoor urban neighborhoods. It confirms the notion o f poor neighborhoods developing informally with propertyrights and infrastructure acquired gradually over time. Home ownership i s very high, with 82 percent o f households living in a dwelling owned by a member of their household. Most of this ownership is informal, however, with less than half (45 percent) o f households with formal property rights - escrituras -to the land on which the house i s built. Some regional differences emerge with the region along the Northern Border standing out with a higher percentage o f home ownership than in other regions (86 percent) but fewer o fthemwith titles (27 percent Northern border against 52 percent inthe Centre). 67 Note that we would expect homeownership to be higher in poor neighborhoods than among poor households -in Latin AmQica poor neighborhoods are dominated by newer settlements on the outskirts of town where self-help housing is the norm. On the other hand, as discussed in chapter 2, only about three quarters of the poor live in poor neighborhoods and the rest, who live inbetter off neighborhoods, are less likely to own their home. 84 Time, whether measuredby the age o f the house or the maturity o f the settlement affects the probability ofhaving formal ownership. Thus, the older the house, the more likely it is that the household owns both the building and the land it stands on: joint ownership o f home and land increases fiom 37 percent for houses which are less than 1year old to 71 percent for houses which are 10 years or older. Similarly, settlements with a smaller share o f recent settlers (a proxy for the maturity o f the barrio) have a greater proportion o f fully owned houses, built on titled land. Curiously, households who received credit or subsidies from government programs related to the dwellingwere more likely to own their house, but not more likely to have a full title o fthe house. Finally, less than one-third (30 percent) o f the households in the sample report that some form o f informal mechanism (individual or organization) exists to solve disputes over property rights.Thus, a large share o f informalhomeowners cannot resort to some arbiter institution iftheir propertyrights are contested. Housing exchange value There is unfortunately not much information on the resale housing market in poor neighborhoods - whether in Mexico or elsewhere. The presumption is that it is not very developed given poor households preference for progressive housing. The limited research available suggests that low-income settlements tend to be dominated by a land rather than a housing market as low income households prefer to acquire land for self-help housing rather than finishedhousing (Gough, 199Q6* In addition, studies suggest that few of the homeowning households in these low- income settlements actually want to sell: the hardships suffered during acquisition and consolidation result in a strong attachment to the property (Gough 1998, Datta and Jones, 2001). Inparticular, research on low-income settlements in Querktaro and Toluca suggests that if low- income financing solutions were to come forth most poor households would prefer to improve their conditions inthe present settlement rather than use the resources to move elsewhere (Datta and Jones, 2001). Nevertheless,data is emerging that suggests that a resale market does exist for low- income housing even if it still is in its infancy. Thus, a study of informal settlements and regularization in Tijuana found that the share o f houses that were paid and titled according to the census (49 percent) exceeded the share registered with government organisms (42 percent). This difference between"self-reported" title and formally registeredtitle couldpossibly be evidence o f the existence o f an informal secondary market where houses are bought and sold inparallel to the formal market (Alegria and Ordbiiez, 2004). In addition, SEDESOL survey of poor Mexican barrios shows that a good share of new arrivals and recent movers purchase used housing:more than a quarter o fthe households The relative importance of the land and housing market may be linked to the age of a settlement Turnover is quite high in recently occupied settlements (especially invasions) where residents attempt to cash-in on the value of the land by selling the rudimentary shelters they have built to establish their claim. Subsequently, little exchange of houses takes place for decades after that because even in later stages of consolidation, most newcomers acquire land rather than finished housing (Datta and Jones, 2001; Gough, 1998). 85 that migrated inthe last 5 years and own the home they live in are in a house that pre-dated their arrival; even more encouraging, three quarters o f the home-owning households that moved in the last year moved to a pre-existing home. Note, however, that given the small number o f households in the sample that migrated or changed homes recently this represents a maximum o f about 3.5 percent o f all owner-occupiedhomes inpoor neighborhoods The development of a secondary market for low-income housing is likely to be affected by the extent of limitations on new constructions and illegal settlements. The latter working against new arrivals but in favor o f the older settlers. Other important factors are the availability o f housing finance and possibly the strengths o f property rights, discussed in turn below. Most housing finance systems in Latin America work against the development of a secondary market for low-income housing -they typically exclude the poor and/or often financing for "used housing". InMexico, the public housing institutions, such as the Instituto del Fondo Nacional de Vivienda para 10s Trabajadores (INFONAVIT), are starting to provide mortgage for usedhousing financing but they tend to target clients belonging to the middle class and above. Security of tenure increases the exchange value of a house in several ways. First, recognized property rights contribute to creating a market. The barrios data discussed in chapter 3 points to this as the share o f households who own a house that predated their arrival rises in older settlements - where supposedly more mature institutional arrangements prevail and property rights, formal and informal, are better established. Secondly, there i s a presumption that prospective buyers would be willing to pay a premium in order to purchase something whose ownership i s clearly established. This i s supported by evidence from many surveys across Latin America and elsewhere. In Peru, for example, 64 percent o f homeowners who acquired a title duringthe massive titling campaign that occurred in 1996-2001 consider that the title increased the value o f their home (Mosqueira, 2003).69 Security of tenure also has additional indirect effects on the value of a home, through "neighborhood quality." This i s particularly true in Mexico, where utilities, road and transport services cannot be provided until a settlement's status i s legalized. More generally, evidence from Peru and elsewhere shows that more secure homeownership tends to result in homeowners investing more in their home, thereby contributing to the general improvement of the neighborhood. Other factors that affect the exchangevalue of homes owned by poor people have to do with the characteristics of poor neighborhoods. As discussed in chapter 3, Mexico's poor urban neighborhoods are plaguedby much greater crime and violence than cities ingeneral. They are also more vulnerable to natural and environmental disasters. This i s anotable issue inMexico, which i s particularly vulnerable to natural disasters, having experienced three yearly between 1980-1999. The 1985 earthquake in Mexico City, for example, cost some 10,000 lives alone. The urban poor, who tend to live inhigh-densityareas with poor infrastructure and whose houses are 69Formal tenure can also affect the returns of the overall portfolio of resources by freeing up labor otherwise engaged inprotecting insecure property rights and allows home businesses to move into more appropriate location. In Peru, formal titling increased labor force participation due to the reduced need for constant presence in the house (to demonstrate ownership), for participating in community groups and for filing administrative claims for formalization. This resulted insubstantial increases infamily income as well as in a significant decline inchild labor, for which adult labor was substituted (Field, 2002; Mosqueira, 2003). 86 o f worse quality are disproportionately affected by such disasters: more than two thirds o f the people affected by natural disasters are poor.70Finally, as discussed inboth chapter 1and chapter 3, access to services tends to be much lower in poor neighborhoods although access improves over time. But even where access i s high quality tends to be low in poor neighborhoods. The barrio surveys point to this fact as three quarters o f the households did report having water services on their property but only half o f them actually got water every day o f the week and every hour o fthe day. Other returns to housing Returns to housing as an asset include the flow of "housing services" it provides, some of which can be monetized in case of need by taking in new tenants or extended household members who share into the upkeep of the household. Indeed, the bulk o f landlords inMexico, as inthe rest o f Latin America, are homeowners who let out a room or a part o f their house (Rakodi, 1995). While surveys and interviews with these small scale landlords suggest that this i s not a very profitable way of investingtheir resources, rentingi s perceived as offering a number o f benefits (Box 5.3). Returns to housing can also materialize in the form of access to credit by providing collateral, although this is generally dependant on having formal tenure. In Peru access to formal credit increased from 7 percent to 42 percent among beneficiary households while recourse to informal credit decreased from 3 1percent to 9 percent following the titling campaign (Mosqueira, 2003). No similar data i s available for Mexico. Box 5.3. Small-scalelandlordism, how profitable is it? Surveys and interviews with small-scale landlords in Guadalajara and Puebla, Mexico found them to believe renting is not a very profitable activity, yet one that offers a number of advantages. Renting out a room or a floor of their house generates resources for housing improvements. It also provides temporary income in times of needs and permits use of accommodations built to one day house a child's or relative's family. A number of landlords felt that rental accommodations could provide a modest income duringthe landlords' old age. Finally, many seemed to admit that beyond investingin bricks and mortar they did not know what to do with their limited savings. Thus a mixture of motives revolving around the family, old age and a lack ofperceivedalternatives characterizesthe small-scale landlord. Source: Gilbert andVarley, 1991. 70Other countries also demonstrate the stronger effect of natural disaster o n the poorest. Thus, in Honduras, Hurricane Mitch was more costly to the poor than the rich: the poorest quintile saw average asset losses of 18 percent, compared to 3 percent for the richest quintile. Inmetropolitan San Salvador and Tegucigalpa, about one fifth of the poor report having suffered damage from landslides inthe last five years compared to 5 and 9 percent for the richest quintile (World Bank, 2002). 87 THE POOR'SACCESS TO FINANCIAL SERVICESINMEXIC071 A vast majority of Mexicans - rich or poor - do not use formal financial services. The banking crisis o f 1994-1995 is still present inMexico -bank credit to the private sector is still only a quarter o f what it was before the crisis- and the financial market remains shallow relative to comparator countries. The banking sector's credit to the private sector amounts to only 30 percent of GDP in Mexico, compared to peer countries such as Brazil and Malaysia, where the ratios reach 51and 145 percentrespectively. Indeed, the use of formal financial services is very low inMexico by most standards: in a survey of households in Mexico City, less than 25 percent of the adult population reported usingformal sector financial services.72Incontrast, almost half of the adult population in Brazil, and nearly 90 percent in the US, have access to financial services (Figure 5.3). These findings were corroborated in a survey o f households' use o f banking instruments in Mexico's three largest cities (Mexico City, Guadalajara, and M~nterrey).~~ Figure5.3 Use of FormalFinancial ServicesI s Low In Mexico, Compared T o the United States or Brazil Percentage of p' bulation with access to formal financial services (Mexico City, Brazil, US) .- 7i 100% 5 0 90% E0 80% 70% 2 g u) ': 60% 5 50% 40% .-c 30% 20% Q 10% sQ 0 0% - Mexico D.F Brazil USA Source: World Bank(2003). Unsurprisingly, exclusion is worse for the poor as access to financial services is strongly correlated with income. Inthe Mexico City survey, the medianuser of formal banking instrumentsbelonged to the 8th income decile. For informal arrangements, such as participating ina savings club, a cajapopular, or keepingcash at home, the medianuser belongedto the 5th or 6th income decile (Table 5-2). 71This section draws extensively on World Bank, 2003b. 72 The usual household data sources do not collect information on use of financial services. Unless specified otherwise, the information regarding urban poor household's access to financial instruments therefore draws on a survey of Mexico City prepared for "Broadening Access to Financial Services Among the Urban Population: Mexico City's Unbanked", World Bank, Latin America and the Caribbean Region, 2003. The reader should note that access to financial services may be larger inMexico City than elsewhere and so the survey probably overestimates access to financial services inurban areas ingeneral. 73 This survey covered a wide sample (45,000 households) in three cities (Mexico City, Guadalajara, and Monterrey). However, since its purpose was to collect information on people with access to banking services it isbiasedtowards higher income groups and may notbefully representative of the population as awhole. 88 Table 5.2. Use o f formal financial instruments increases with income inMexico City T o which income group does the median user o f each type o f financial instrument belong? Type o f financial instruments Income group o f median user (decile) InformalBanking Instruments Tandas 5th Savings at home 6th Cooperativa or CajaPopular 6th All informal 5th FormalBanking Instruments Savings, debit card or AFORE 7th account Checking account 9th Investmentaccount 9th All formal 8th Source: World Bank2003. What is more surprising is the extent to which the poor are "unbanked" inthe sense o f n o t even having access to a deposit account: over 90 percent o f independent workers, workers without pay and the economically inactive were without formal financial services. This cannot be blamed on an inability to save -those without access to financial services are about as likely to own a home as those with access. This is in itself evidence of a capacity for asset accumulation though the appropriate mechanisms may be lacking. - What alternatives existfor thepoor? The poor rely o n a variety o f alternatives t o formal banking services. Apart from investing in housing and physical assets, discussed earlier, the poor generally resort to cash savings held at home; loans to and from friends andrelatives; and to informal savings institutions. The latter take a variety o f forms: tandas, cajas de ahorro, clubes, etc., described indetail inBox 5.4, but appear to have a limitedreach: inthe Mexico City survey, only 30 percent o f those who did not have access to formal financial mechanisms reported holding some form of financial savings with informal institutions. Notably, these informal mechanisms were also usedby people who didhold bank accounts. The cajas inparticular seem to be almost exclusively usedby people inthe upper endofthe income distribution. These informal savings o r loans systems rely o n personal trust and confidence in other people and tend t o offer l o w returns. O f the various informal savings mechanisms used, only the cajas offer positive - and often quite high- returns to the savers, but again, the scarce evidence available suggests that poorer clients do not use, or have access to, cajas. Other mechanisms used offer no opportunities to accumulate financial savings inways that maintain the purchasing power o f those savings. F o r the poor, the most common source o f credit indicated in the survey was commercial stores, followed by informal institutions as described above, together with loans f r o m friends and family. The poor in Mexico City are highly unlikely to borrow from formal banks. Non-bank credit terms tend to be high, however: grocery stores offering credit for food purchases charge the equivalent o f 30 percent monthly interests rates, chain stores consumer 89 loans (typically for consumer durables) charge some 15 percent on a monthly basis. Note that the latter i s relevant even for the poor -four-fifth o f them own a television, about half own a fridge and around a third a washing machine (Ojcinade la Presidencia,2003 Impulso). Box 5.4. Informal savings institutionsinMexico Tandas are rotating savings pools that operate as both loans and savings instruments. A tanda might work as follows: a group o f 20 people agree to contribute MX$200 a week to a common pool for 20 weeks. Each week the proceeds are given to one member o f the pool (who doesn't contribute that week) so that those at the beginningo f the pool effectively receive a loan which they pay off equal monthly installments until the end o f the cycle. Those who receive money at the end o f the cycle have effectively been saving with each pool contribution until the final withdrawal. From anecdotal evidence it appears tandas are usually made up o f 5 to 20 "partners" each contributing US$10 to US$20 on a weekly or bi weekly basis. The prevalence o f tandas in low-income communities provides stark evidence that the poor have both the capacity and the willingness to use financial services, even with small amounts. Some quotations from tanda participants follow: "My tanda i s sacred. Icannot fail to make a payment because it hurtsthe others"; "I participate only if Iknow Ican hlfillmy obligation. I'd rather not eat than fail to make a payment"; "My savings are small but drop by drop they make a puddle. Inthe bank they want us to save large quantities and they ask for a lot o f papers. Iprefer my tanda". Clubes are managed by commercial stores and follow the same scheme as the tanda in which a group o f individuals make regular and equal payments, inthis case to the store and inthis case the "pot" is not cash, but a specific article for sale. The articles may vary from person to person, but the amounts paid are usually the same. Clubes select the weekly or bi-weekly winner indifferentways, butmost involve a socialmeetingand agame ofchance (door prize, bingo, etc.) Cajas de ahorros are similarly formed in groups o f 15-50 members associated by a common place o f work or through a church. Unlikethe tanda, a caja usually requires a full year's commitment after which members can withdraw their accumulated savings with interest, depending on the caja's earnings throughout the year. Cajas also make loans to members and to non-members (whenrecommendedby members). Since the cajas are not regulated there is no way to discover the range o f interest rates they use, active or passive. However, a sampling o f cajas in the DF turned up deposits ranging from US$lO-$20, active interest rates from 5-6 percent per month for members and 10 percent for non-members, and, in the most profitable cases, passive interest paid annually at rates up to 30 percent-40 percent. Generally, cajas have relatively low loan ceilings and are quite hardto find, however. Besides the cajas examined here, an unknown number o f cooperativas exist. The cooperativas tend to offer interest on savings at a somewhat lower rate than the cajas. Systematic information i s not available, however. Source: Adaptedfrom World Bank(2003). Even if informal savings mechanisms, friends and families did provide adequate alternatives for savings and credit, there are specific costs to not having access to formal banking services. The poor still need to make payments on goods and services and receive and transfer money. Without a bank account, these transactions become costly. For example, paying for public services such as water, gas, etc. through banks (where one i s not a customer) i s typically expensive, while paying in person at the public services company represents substantial 90 time lost in transit and waiting in line.74 Likewise, cashing checks, which are the predominant form of payment inMexico City, i s more costly, both interms o f time and money, than receiving direct deposits to an account.75 What obstaclesexistfor thepoor to accessfinancial sewices? Key reasons for low income people not to open a bank account include a lack of money, high minimum balances requirements and general distrust of banks. Focus group interviews also suggest that anticipation o f badtreatment and/or rejection plays an important role indeterringlow-income groups from approachingbanks. In addition, banks in Mexico lack the incentivesto pursue low-income clients. There i s clearly an untappedmarket for financial services to low income groups; yet, the formal banking sector i s generally reluctant to extend services to this part o f the market - because o f the low incentives to expand lendingservices ingeneral, andbecause o f specific concerns o fthe ability o f small and micro-businesses and low income individuals to accumulate savings andto repay loans. The commercial banking sector has also become more focused on the high-end clientele, "investor" profile rather than the typical "saver". Banks earnings are increasingly derived from services charges - possibly a reflection of limited competition rather than actual costs for the banks- and investment, the latter mostly from low risk government papers. There is, in short, little incentive to expandinto the low-income segment o f the market. Non-bank financial institutions, on the other hand, have increased in number and many have become regulated or at least legally recognized - but coverage in urban areas remains very limited. The government has taken some steps towards developing the so called "social interest banking" sector - cooperatives, credit unions and savings and loan associations. A development agency/second tier bank (BANSEFI) has been created, which offers technical assistance and funding to social interest banks (organizations covered by the popular savings and credit law). The objective i s to increase membership in the social interest banking organizations from the current estimate o f 3 millionto 4.5 millionby the end o f 2005. BANSEFI has also taken over the management o f over 550 former national postal savings offices o f the Patronato del Ahorro Nacional (PAHNAL) and, as o f 2003, has approximately 1.5 million clients. However, most o f BANSEFTSoffices are rural and so leave the urban low-income residents needs unmet. There are only 60 branches in metropolitan Mexico City. Overall, only 2 percent o f the social interest banks have offices in Mexico City and their presence in the most urbanized states o f Nuevo Lebn, Estado de Mexico and Jalisco i s also minimum. Box 5.5. Wholesale, small-scale banking can be profitable: the case of BancoAzteca Banco Azteca (BA), Mexico's sole commercial bank entirely focused on the low-income market, began operations in 2001. It now counts some 1.75 million accounts nationwide and deposits total over MX$240 million, around 80 percent o f the total so called "social interest banking" sector. In all, BA's phenomenal growth i s a strong indication o f pent-up demand for financial services among low-income households. 74 Close to half of the respondents in the survey reported spending over two hours in travel time to make services payments. 75 Some estimates based on the survey suggest that taking into account time and traveling costs, foregone interest rates, and unrealized profits, the cost of not having access to formal financial services could amount to a staggering 15percent of a low-income household's yearly income. 91 BA grew out of Elektra, a commercial outlet specializing in domestic appliances and furnishings-mostly electronic- and in selling on credit. BA (i) focuses on households between the 35th and 65th percentile o f income distribution (which includes the moderately poor) (ii) a holds strong urban presence (150 branches in metropolitan Mexico City compared to BANSEFI's 60.) (iii) onElektra's network of850 storesnationwideandexperience, includingthecredit draws information system (iv) offers both credit and savings products. Savings: BA offers a low-end savings product - Guardadito - which pays minimal annual interest rate (1 percent) but requires a minimum deposit and balance o f only MX$50 (approx US$5.00). The Guardadito accounts average a balance o f MX$1300. There i s also a savings instrument for clients who can maintain a minimum o f MX$5000 (US$ 500.00): time deposits paying annual interest o f between 4 and 7 percent depending on the time frame. Credit: Initially BA made loans only for commercial goods but it now offers personal loans (unsecured) up to MX$5500. The average loan size is MX$2500 and payments are collected weekly. Since the system builds on intense follow-up o f clients, operations costs are relatively high, but this is compensated for by high interest rates -between 6-10 percent per month - substantially higher than the conventional commercial bank. To qualify, Elektra requires official identification, proof o f income and proof o f residence from its clients. Credit officers visit each potential client at home where they assess the value o f the client's property and possessions. Informal workers (who make up an estimated one third o f BA's clients) are allowed to present receipts for goods purchased for business to demonstrate their incomes: e.g. carpentry tools, gasoline receipts, etc. Credit officers follow the repayment and make weekly visits to the person's home whenever there i s a late payment. Credit officers earn bonus commissions when credits pay off, but must pay for defaults out o f their wages. The practice o f close tracking o f clients pays off inveryhighrepaymentrates - 97 percent- and very low late payment rate (80 percent on time). Traditional regulations have been overruled on an ad-hoc basis to accommodate BA's specific low-end clientele. Regulations normally require that proof o f income or o f net worth be obtained before authorizing credit. Since this i s only applicable to formal sector employees, the National Commission for Banking Securities (CNBV) has accepted BA's socio-economic studies as an alternative, as well as BA's redefinition o f default or late payment based on monthly delays, rather than on delays by payment. Under present regulation, all Mexican banks must use (and finance) the country's only credit bureau, even though this bureau covers a "high-end" clientele which i s useless for a bank working with a "low-end" clientele. Instead, CNBVhas agreed to allow Banco Azteca to rely only on its own database for a non-extendable six-month period. Similarly, CNBV has granted a six-month waiver o f its requirement that each credit be approved personally by amember ofthe bank's credit board allowing instead for electronic signatures. Source: World Bank, 2003 The example of Banco Azteca shows that low-end banking can be profitable on a wide scale. The most prominent recent newcomer to financial services, Banco Azteca, i s evidence o f the potential success o f low-income focus. Banco Azteca began operations in 2001 and already counts some 1.75 million customers nationwide (Box 5.5). It also shows that the riskiness of small-scale, wholesale banking business can be reduced if resources are spent on credit ratings, background research and collection. An essential ingredient to Banco Azteca's success has been amendments to traditional banking regulations - regarding credentials neededto obtain credit, andto accommodate the low-income profile o f its clientele. 92 POLICY OPTIONSFORIMPROVINGTHE ASSET PORTFOLIOSOF THE POOR The poor need to accumulate assets - whether to protect themselves or to get ahead. But it is clearly harder for them than for the rest of society to do so. It is harder because they have fewer resources in general, but also because they usually lack access to social insurance, which helps protect a nest-egg and/or allows savers to pursue a higher returnhigher risk savings strategy. And it i s harder because good saving instrumentsare not usually available to them: they are typically excluded from formal financial institutions; alternatives to these (micro-finance institutions, informal financial arrangements) seldom offer good saving options; and low-income housing markets are probably neither very liquid, nor particularly buoyant. Improving the asset portfolios of Mexico's urban poor requires a three-pronged approach. The first includes policy interventions to increase the poor's income security as well as their access to social insurance. The second entails making low income housing a better asset without distorting the poor's `decision to buy or rent (which tends to be based on lifecycle considerations rather than income). Andthe third entails broadening the poor's access to financial assets, particularly savings instruments. Interventions to increasinghousehold resources and insuranceoptions A first and obvious area for public policyto increasethe ability of the poor to save is to increase the resources available to them. A good example here is Mexico's own Oportunidadesprogramwhich provides conditional cash transfers that promote school attendance and the use of preventive health care -in turn essential building blocks to increase household resources over the long run. Its impact in urban areas i s described in more details in chapter 2 (curiously, the major impact o f Oportunidadesin urban areas i s an improvement in housing, as households appear to use the income supplement to invest intheir dwellings [Gertler et al, 20031). There is also room for deepening the reach of public schemes aimed at reducing household vulnerability such as unemployment insurance, pensions and health insurance, which is currentlyvery limitedfor Mexico's poor.This is the focus ofthe companion report on Social Protection. One issue, however, that i s specific to the urban poor i s the complete lack of a workfare type o f programthat can rapidly be rolled out. The lack of workfare programs is a major gap for the urban poor for several reasons.Firstbecausehistorically the impact ofmacro-shockshas beenthrough lower realwages rather than ~nemployment.~~ It follows that even ifunemployment insurance were to bereformed and made available to the poor, it would not help the collapse in real wages such as happened between 1994 and 1996 that resulted in a tripling o f the number o f urban extreme poor. Second, as discussed in chapter 1, the primary response o f the urban poor, as opposed to the rural, to an income shock i s an added worker effect, whereas additional workers, usually the spouse enters the labor market. This type o f strategy i s obviously less effective intimes o f recession but can be made more so through workfare programs such as Argentina's Jefes y Jefas de Familia described inchapter 3. Note finally that there are efforts to develop micro-insurance programs under way, although no clear paradigm has yet emerged(see www.microinsurance.com for a discussion). 76This however may be changing as discussed inchapter 3. 93 Making housinga better asset Housing is likely to remain the most valuable asset held by the poor even if they are provided with good alternatives in terms of saving instruments. Evenin developed countries such as Belgium and the UK the only form in which the poor ever hold any kind of wealth i s through homeownership (Van den Bosch, 1998). Thus, improving the asset portfolio o f the poor necessarily entails making low-income housing a better asset. So how then to improve the low income housing market without distorting housing markets or the poor's investment choices? It includes making it more liquid, notably by developing housing finance schemes for the poor that allow for a usedhousing market to develop (such as the Costa Rican or Chilean ones discussed in Box 5.6). It may also require making it more secure though actions such as titling. Additionally, measures to improve neighborhood quality, whether by improving services or by reducing crime and violence situation of poor neighborhoods can make a particular neighborhood's housing market more buoyant, although the impact on housing market in general is more limited77.Thus, key areas for reform include developing resale markets for low income usedhousing, s well as improving tenure security. Housing for the poor is a major focus of the current government, as evidenced by the 2001-2006 National Housing Policy's efforts to target low/moderate-income households. The approach includes developing a unified system of housing subsidies complemented by credit and savings; expanding housing finance by working with financial institutions experienced in serving low/moderate income groups to finance low-cost housing solutions; and strengthening property rights. This is an important change in emphasis. Over the last decade, federal support to the housing sector was largely focused on the return o f private finance following the exodus o f commercial banks from mortgage lending at the time o f Tequila Crisis. While these efforts have borne fruit, they largely focused on middle income families and there was little systematic attention was paid to housing the poor (the Tu Casa Program, or its predecessor Vivah, until recently only provided starter houses for lower middle income families, and at a very small scale and with significant implementation problems).78 Box 5.6. The Costa Rican Direct Demand Subsidy Program The Costa Rican direct demand subsidy program, started in 1987, was modeled after a Chilean one. The program, which can be used for construction alone, land purchase and construction, or purchase or improvement o f an existing unit, consists o f the following elements: 77 Slum upgrading, or even just improving infrastructure services, including transport links in poor neighborhoods, will typically translate in both an increase in property values and in easier resale. This benefits homeowners in that particular neighborhood, although the impact on renters is less clear and the effect on the low-income housing market in general is limited. This is why we do not discuss this here in details. For a discussion of community based and municipal strategies to cope with crime and violence see Moser et a1(2004) or Van Bronkhorst (2003). 78The discussion of housing policy is mostly based on a review of low-income housing policies done jointly by the Mexicangovernment and the World Bank in2002 (WorldBank, 2002). 94 A subsidy voucher of an amount inversely related to household income with a maximum value o f about U S $4,000; A mortgage loan given by an "authorized entity", including government banks, NGOs, cooperative federations and savings and loans: these entities have the authority to choose beneficiaries, deliver the direct subsidy and extend a loan to complement the direct subsidy and the household's down payment. Households work with the authorized entities to determine the maximum value o f what they can afford to purchase, the loan amount, and the requireddown payment. The householdthen looks for a housing solution with this maximum price. The government-housing bank then buys the authorized entities' social housingportfolio at below marketrates. At its initiation, the program attempted to re-capture the subsidy from the household when the house was sold, butthese efforts were abandoned inthe early 1990s. In contrast to Chile and most other countries that have adopted direct demand subsidies (with the exception o f Colombia), the program of Costa Rica succeeded in attracting the private sector into the very low-income market. The main reason i s that a group o f sophisticated NGOs experienced in housing development - a rarity for developing countries- have become the main developers under the program rather than for-profit developers. At first, many for-profit developers usedthe direct subsidy program as well. Since 1994, for-profit developers have largely stopped usingthe program, mainly because o f increased political and economic risk.NGOs have stepped in to fill this gap. Some NGOs help households construct a unit on an existing lot by providing technical assistance. Others are "authorized entities" that assemble groups o f beneficiaries, extend the credit, and develop units through contracting for-profit construction firms. The program has proved stable until recently, delivering a significant number o f direct subsidies each year since its inception in 1987 through the mid 1990s. The total number delivered from 1988 through 1998 (93,049) represented 13 percent ofhouseholds inthe country. Authorized entities have experienced few arrears on these loans. Though the program has proved politically popular, fiscal constraints since the mid 1990s have caused a decline in resources available for subsidies program. Various stopgap measures -such as issuing of bonds for funding the direct demand subsidy-have proved problematic and complicated the operations o fthe program. Source: World Bank(2004) Institutionally, the administration has taken bold steps to achieve these national priorities. Untilthe last three years, the federal government lacked an effective vision for reform o f housing sector so that the federal agencies involved inhousing operated with neither an overall strategy nor coordination. N o agency existed with responsibility for sector policy. The current administration established a new organization National Commission for Housing Promotion (Comisibn Nacional de Fomento a la Vivienda (CONAFOVI), to lead and coordinate other agencies in the sector. In turn, CONAFOVI has developed an official housing policy based on widespread participation o f public and private sector leaders and analysis. Policies regarding housing subsidies, housing credit and household savings, property rights and low and moderate- income land development are discussed below. In recent years, the Government effectively used the Special Credit and Subsidies for Housing Program (Programa Especial de Criditoy Subsidios a la Vivienda,PROSAW to pull private mortgage lenders further down market. The government has relied on the Tu Casaprogram to support poverty objectives inlower income segments, though it impact has been limitedinrecent years due to implementation issues and insufficient local counterpart funds. The 95 cost o f providing serviced land i s inmany cases beyond the financial capacity o f the subnational government. The government i s now trying to improve the poverty effectiveness o f its housing programs by standardizing its previously fragmented, uncoordinatedprograms and transferring all on-budget federal housing subsidies to sit under the direction o f one institution. In addition in 2004, Tu Casa expanded the housing solutions to include lower cost options more affordable to poor families and municipalities with difficult providing matching funds. Inaddition to the small starter homes it traditionally supported, Tu Casa now provides subsidies for home improvement and expansion. In the area of household credit and savings, the government undertook important structural reforms supporting mortgage credits for finished housing. While these efforts do not directly help the poor, they have helped to meet housing demand o f the middle class and therefore stem filtering up o f low-income housing. Nevertheless, micro loans supporting low cost housing solutions - such as home improvement and sites and services- are more appropriate to the way the poor gradually construct and consolidate their homes. The government, through its Mortgage Federal Society (SociedadHQotecaria Federal, SHF), the public bank charged with fostering private housing finance, has committed to move beyond mortgage financing for middle- income households and explore micro finance for housing. In2005, SHF intendsto launch a pilot program o fproviding fundingto private lenders for micro finance. In addition to credit and subsidies, householdsavings is a factor in the poor being able to acquiring and improve their home. As discussed in earlier, the government created BANSEFI in 2001 to promote household savings and support financial institutions that focus on low/moderate income households ("popular financial institutions"). BANSEFIhas recently begun to partner with public and quasi public housing agencies to support housing savings programs. Unfortunately, as discussed before, BANSEFI remains more active in rural than urban areas although that might be changing as the government relies on BANSEFI to disburse the Oportunidadescash transfers. Fully enforceable individual freehold title can greatly increase the price of housing and making it unaffordable to the poor - indeedthis is why the poor often settle in informal housing. These price considerations underliethe fact that degrees o f security o f tenure can result from a range o f ownership characteristics - having a un-registeredtitle, having a registeredtitled, land zoned appropriate for use, etc., -rather than a single standard o f fully enforceable free hold title. About 40 percent o f Mexicans who say they own their property, in fact lack legal/planning approval, formal titles andremain unincorporatedinto the cadastre and property registry. The rigidity and high standards embeddedin formal mechanismsfor access to land for housing purposes and formal mechanisms for access to real property rights registries - for legal protection and recognition of transactions - and the complications of the ejido system (communally owned land dating from Mexican revolution) largely cause this problem. The state public property registries records take long periods to access and are sometimes unreliable and costs are high and vested interests within and outside government take advantage o f the system. State governments use property registries largely as mechanisms to collect revenues rather than provide a service. Typically, the fees from registries account for the second or thirdlargest component o f own-source revenue. Government efforts on property rights in housing sector have focused on modernizing the public property registries. Well aware o f the importance o f real property rights for many markets and for social stability, the government initiated a project to strengthen 96 the real property registries in three states, focusing on modernizing the practices and technical aspects o fthe registries. Through the Habitat program, the government has expanded support for the purchase of land reserves in recognition of the fact that low-income urban land supply is a major bottleneck for housing the poor. Most commonly under this scheme, sub-national governments purchase rural land, facilitate the provisiono f trunk infrastructure and the regulatory change from rural to urban land use and then sell to developers or households or provide the land as a subsidy to low income households. Though great progress has been made, additionalefforts in the following areas may be necessaryto fully respondto the housingneeds of the 40 percent of the populationbelow 3 minimumwages. Broad land policy for the poor is needed. Land reserves - the public provision o f urban land for low-income families- may currently be the only solution for low- income housing.79Support o f land reserves should, however, be seen as a short-term response linkedto a broader mediumterm program that addresses the bottlenecks in land markets and aims to build low-income land markets. Such a program would entail reviewing regulatory barriers, building local capacity to plan, administer permitsandprovide infrastructure and expanding and streamlining titling andregistry processes. This could include reducing standards to allow formal land to be affordable to the poor (e.g. lot sizes, setback requirements, etc) and allowing the private sector to develop housing with progressive infrastructure - in many states only the public sector i s allowed to develop housing with progressive provision o f infrastructure. It should also include continued efforts to provide titles to the poor, complemented by reforming property registry systems to reduce costs and increase capacity o f the system to register the new titled poor. The Peruvian experience may yield some important lessons (Box 5.7). Infact, the government is planning a titling program, the structure o f which will be developed following an in-depthstudy that i s to involve Peruvian experts. Box 5.7. Peru's UrbanLandTitlingProgram Peru has seen rapid urbanization, with the population living in cities increasing fiom 46 percent in 1960 to 73 percent in 2000. While the official policy for long centered on repressing illegal settlements, the government now recognizes that these developments are inevitable. In 1996, Peru launched the Informal Real Estate Property Formalization Program, with the purpose o f establishing legal titles for informal urban properties and thus secure poor families rights to their principal assets. The programs implementation depended critically on consensus building among the political entities, people in the informal settlements and civil society in general. By 2003, over 1.3 million plots had been titled through the program and women represented more than 50 percent o f the beneficiaries o f the formalization process. The program has turned out numerous positive results: 79Even the middle class's demand for urban land is being met through government provision of land reserves for INFONAVIT, SHF and FOVISSSTE housing programs. Were private developers to further enter into this business, they would likely first provide land to the more lucrative middle and upper income markets, rather that to the poor. 97 Two thirds o f newly titled homeowners perceived that the title increasedthe value o f the home and three quarters believedthat the title increase the security o f ownership. Titled households invested more in home improvements in the year following titling, housing quality improved (with more titled homes made o f durable materials) and so did access to services (notably water.) Finally, crowding was reduced as households enlarged their homes and increased the number o f rooms, which also had the additional benefit o f stimulating the rental market. As properly titled real estate could be used for collateral, access to formal credit increased from 7 to 42 percent among beneficiary households while recourse to informal credit decreased. Formal titling increased labor force participation due to the reduced need for constant presence in the house (to demonstrate ownership), for participating in community groups and for filing administrative claims for formalization. This resulted in substantial increases in family income as well as ina significant decline in child labor, for which adult labor was substituted. Source: Field(2002). Mosaueira (2003). Greater micro credit to accompany home improvement and expansion. The SHFpilot program for 2005 shows great promise as it fosters increased private sector lending, and appropriately reflects the recent moves o f the private sector into this market through consumer and materials loans. The historical absence o f private credit for low-income housing has been recognized in Mexico, and many states and municipalities responded by providing micro loans for housing. Credit from the public sector has a terrible track record in terms o f loan origination and recovery, fostering poor financial practice on the part o f beneficiaries and crowding out the private sector with subsidized rates. A challenge for SHF will be to foster expansion o f private micro credit inthe face o f continuing crowding out o f the private sector by state andmunicipal lending. Align subsidies programs to social policy goals and foster private finance to low income households. Government efforts to unifytheir subsidy programs have to date focused on unifying subsidy administration below one agency, the National Fundfor Housing Economy Support (Fondo Nacional de Apoyo Econbmico a la Vivienda, FONAEVI). The challenge o f aligning subsidy amounts and targeting toward low income housing goals remains, as well as refocusing fiscal resources from subsidies to support the development o f private mortgage finance down market - whose social impact i s limited as households below 3 minimum salaries cannot afford a finished mortgage financed house with a fiscally prudent level o f subsidies- to subsidies targeting social goals o f low income housing and that focuses on neighborhood upgrading, sites and services and home improvement. The expansion o f Tu Casa to include subsidies for home improvement and expansion i s a step in the right direction. Modernize the institutional andfinancial framework for property registries. Property registry reform to date has focused on the mechanics o f registries, using electronic systems, reengineering work flows, etc. While these efforts may to some degree improve efficiency the big issues related to registries as a major income source for states and the lack o f competitive and transparent mechanisms for the selection, recruitment, contracting andpromotion o fpublic employees remain. 98 Expanding access toformal financial sewices Mexico is characterized by a shallow financial market and neither poor nor non- poor make much use of financial services, especially not through the formal banking system. A full discussion o f the policy reforms that can help promote more access to financial services by the poor i s beyond the scope o f this chapter (for a discussion in the context o f Latin America see World Bank 2003a and 2003b). Briefly, however, here are some o f the approaches that have been advocated. Concerningthe formal banking sector, this includes: Approaches to improve thefinancial infrastructure forfinancial intermediation. This includes strengthening credit information registries, legal and regulatory framework for secured transactions, and regulatory framework to encourage product design. Approaches that encourage banks to off.. low cost financial products to poor households. This includes: a) encouraging the use of informationtechnology (PDAs, smart cards, and handheld computers); b) requiring or encouraging banks to offer "lifeline" accounts with low or no minimumbalance requirement; and c) reducing the physical and social distance o fbanks from the poor (e.g. adequate branching). Approaches to reduce the lack of familia@ between poor households and banks - through financial literacy programs by publishinginformation on the profitability o f reaching down and through programs to encourage large employers to pay through electronic transfers rather than by checks. Approaches to transfer direct income transfers to the poor through the formal banking sector. In addition, the regulatory framework is currently not designed to deal with low- end banking. For example, regulatory costs are lower over a small number o f high price transactions than for a large number o f low value number transactions. Butthe example o f Banco Azteca shows that regulations can be altered to accommodate low-end conditions - like wavering the need for formal proof o f income for credit (which automatically excludes employees in the informal sector). In sum, the urban poor do accumulate assets, but are constrained in their choices - because of their lack of resources, their risk aversion, but also the fact that good savings and insurance instruments adapted to their needs are not usually available to them. As a result, they probably over-invest in housing and durable goods, and under-invest in financial assets. Policy measures to make housing a more liquid asset and to increase access to financial services, particularly savings, are therefore essential to help the urban poor cope with poverty and vulnerability. 99 ANNEX Table 1-A: Overview of PovertyTrends for Extreme andModeratePovertywith Incomeand Consumption 1992 1994 1996 1998 2000 2002 Income National 22.4 21.4 37.1 34.0 24.2 20.3 Extreme Rural 35.6 37.0 52.4 52.4 42.4 34.8 Urban 13.3 10.1 26.5 21.2 12.6 11.4 National 52.5 55.8 69.6 63.7 53.7 51.7 Moderate Rural 65.0 72.3 80.8 75.1 69.3 67.5 Urban 43.8 43.7 61.9 55.7 43.7 42.0 Consumption National 24.8 22.0 36.1 33.4 24.2 21.0 Extreme Rural 41.8 38.7 52.9 52.0 42.4 37.2 Urban 13.0 9.8 24.5 20.5 12.5 11.1 National 51.2 50.4 64.0 60.6 53.7 51.3 Moderate Rural 69.3 68.7 76.5 75.8 69.3 66.1 Urban 38.7 37.0 55.3 50.1 43.7 42.2 Source: World Bank (2004a). Figurel.A: There is a strongrelationshipbetweenregionallocationandthe levelof urbanand ruralpovertyrates Ruralversus urbanpovertyrates,2002, for Mexico's regions 70i South Pac 0 SouthGulf&Cat I 0 0 Moderate Poverty Center 50 0 ~ 0 Extreme poverty North 0 0 SouthPac 6 I 0 South Gulf &Car 0 Center 2o 10 0 0 North Mexico, D.F. 0 0 x) 20 30 40 50 60 70 80 90 Ruralpovertyrates Source: Own calculations based on ENIGH,2002 100 TABLE 2-A: METROPOLITANZONES OF MEXICO Zone Numberof Municipalities Aguascalientes 2 Chihuahua 2 Mexico City 54 Coatzacoalcos 6 Colima 2 Cbrdoba 2 Cuautla 2 Cuernavaca 5 Guadalajara 7 Guaymas 2 Lebn 2 MQida 4 Monclova 4 Monterrey 9 Oaxaca 9 Orizaba 7 Pachuca 2 PozaRica' 2 Puebla 26 Querktaro 3 Saltillo 3 SanLuisPotosi 2 Tampico 4 Tijuana 2 Tlaxcala 6 Toluca 6 Torrebn 4 Veracruz 2 Xalapa 3 Zacatecas 2 Zamora 2 TOTAL 188 1: The metropolitan zone of Poza Rica includes also the locality of Poza Rica in the municipality of Papantla. Since we do not have locality-level data, we did not include this locality into our metropolitan zones. 101 Figure2.b: Publicmarginalityindex: Averagevalues of marginalityvs. rangeof marginalityfor metropolitanzones: 2000 35 ~ 30 ~ 25 ~ 20 ~ Monterrey G adala' ra 15- r 8 @ * 0 exico 10- * 5 ** ** ** 0 I * A , , , I Figure3.A: Privatemarginalityindex:Averagevalues of marginalityvs. rangeof marginalityfor metropolitanzones: 2000 I-3 5 - 30- 25 ~ 2 0 - Nexjco * 15- Mgnterreke x)- ** * GYadalajara * * 5 * * 0 A , Table 3-A: Dispersionin municipalaveragemarginality(columnsI,11,IV,V) and in municipalitygrowthrates(columns111,VI) 1 102 Private marginality dispersionin Public marginality dispersionin Municipali Municipality ty growth Municipality Municipality average rates average growth rates I I1 I11 IV V VI 1990 2000 1990-2000 1990 2000 1990-2000 Tijuana 0.0 0.0 0.0 Tijuana 0.0 0.0 0.0 Veracruz 1.1 1.3 2.5 Colima 0.4 1.5 1.9 Guaymas 1.3 1.o 0.2 San LuisPotosi 0.4 0.9 0.5 Pachuca 1.3 0.8 2.2 Zamora 1.1 1.3 0.2 Cuautla 2.7 2.4 0.3 Cuautla 4.2 3.5 0.7 Zacatecas 3.2 1.2 1.9 Guaymas 4.6 0.1 4.5 Zamora 3.4 1.2 2.2 Veracruz 4.8 2.8 7.6 San LuisPotosi 3.5 2.8 0.7 Aguascalientes 5.6 3.4 2.2 Colima 4.3 1.9 6.2 Cordoba 7.1 3.2 3.9 Cordoba 5.1 2.0 3.1 Zacatecas 8.8 4.8 4.1 Monclova 6.8 3.7 3.1 Pachuca 9.2 1.9 7.3 Aguascalientes 8.0 5.1 2.9 Leon 9.7 4.5 5.2 Leon 8.0 3.0 5.0 Monclova 12.2 15.7 3.5 Merida 9.4 8.8 3.4 Torreon 12.5 10.7 2.1 Chihuahua 10.6 14.0 3.3 Saltillo 13.3 17.8 8.9 Guadalajara 12.4 6.6 6.1 Cuernavaca 13.6 10.2 3.7 Torreon 12.6 10.3 2.3 Monterrey 13.7 18.2 12.2 Xalapa 13.0 10.5 8.1 Chihuahua 14.8 5.9 8.9 Queretaro 13.3 10.2 5.1 Merida 16.9 16.5 4.7 Monterrey 14.3 13.8 8.3 Guadalajara 17.8 15.5 8.5 Oaxaca 16.1 12.7 10.2 Xalapa 18.1 2.7 15.4 Toluca 16.4 11.5 6.1 Toluca 19.2 15.6 4.4 Cuernavaca 16.7 15.5 4.3 PozaRica 19.6 15.0 4.6 Saltillo 16.8 14.8 6.5 Tlaxcala 20.8 10.4 10.4 Tampico 16.8 13.0 5.9 Queretaro 23.2 11.8 11.3 PozaRica 17.2 11.8 5.4 Oaxaca 23.2 19.8 20.2 Coatzacoalcos 18.9 8.1 13.6 Mexico Df 28.9 12.7 24.9 Tlaxcala 19.5 17.3 3.0 Coatzacoalcos 29.3 27.8 15.4 Mexico DF 20.2 16.6 12.4 Tampico 29.9 30.4 20.3 Puebla 28.0 28.2 7.9 Orizaba 33.4 28.2 6.9 Orizaba 32.9 28.6 6.5 Puebla 48.0 42.8 12.1 Median 12.4 8.8 4.3 Median 13.6 10.4 5.2 Mean 11.4 9.0 4.8 Mean 15.0 11.5 7.6 1 Columns are sortedby ascendingcity average dispersionin 1990 103 Table 4-A: Mexico's urbaninequalityin an internationalperspective Distribution of household per capita income - Inequality indices URBAN RURAL Country Gini E(0) E(l) E(2) Gini E(0) E(l) E(2) Bolivia 2002 54.0 0.5 0.6 1.2 57.3 0.7 0.6 1.1 Brazil2001 57.7 0.6 0.7 1.6 53.1 0.5 0.6 1.7 Chile 2000 56.5 0.6 0.7 1.8 52.4 0.5 0.7 4.7 Colombia 1999 55.1 0.6 0.6 1.9 55.0 0.6 0.7 3.7 CostaRica2000 44.2 0.4 0.3 0.5 44.0 0.4 0.4 0.5 DominicanR 1997 48.0 0.4 0.5 0.9 47.5 0.4 0.5 1.5 Ecuador 1998 52.2 0.5 0.6 1.4 54.1 0.6 0.6 2.2 El Salvador2000 50.6 0.5 0.5 1.6 46.9 0.4 0.4 0.6 Guatemala2000 55.8 0.6 0.6 1.3 51.8 0.5 0.6 1.3 Jamaica 1999 54.9 0.6 0.7 2.2 46.8 0.4 0.4 1.0 Mexico2002 1/ 46.2 0.4 0.4 0.7 49.5 0.4 0.5 3.8 Nicaragua2001 56.7 0.6 0.8 4.2 52.2 0.5 0.6 3.0 Panama2000 52.2 0.5 0.5 0.9 54.4 0.6 0.6 1.3 Paraguay 1999 50.3 0.5 0.5 1.2 59.9 0.7 0.9 11.8 Peru2000 44.0 0.4 0.4 0.7 47.3 0.4 0.4 0.5 Venezuela 1998 46.3 0.4 0.4 0.8 45.4 0.4 0.4 0.6 Source: For Mexico, own calculations based on ENIGH 2002. Other countries: World Bank (2004a). 1. Based on income per capita Figure4.A: Mexico's urbanareas are less unequalthan rural areas, especially at the top of the incomedistribution 104 Urban-Rural differences ininequality among the poor (X-axis) and ininequality among the rich 2 0 Urban areas more unequal Urban areas more unequal than rural at both the top than rural at the top of the and bottom of the income income distribution, but distribution 185s at the bottom Nicaragua Jamaica 1 0 El Salvador Guatemala c- 0 05 B'"z'lb10 0 15 0 20 Panama Dominican Republic Ecuador- Urban areas 185s unequal Urban areas 185s unequal than rural at both top and than rural at the top of the bottom of the distribution Colombia income distribution, but ~2 0 Chile Memo ~3 0 4 0 Urban-Rural differences in GEp) Source: For Mexico, authors' calculations basedon ENIGH2002. Other Countries: Regional Urban Poverty Study (draft), Latin America and CaribbeanDepartment, WB. FigureA.5: Publicmarginalityindex: some convergence1990-2000 e Monterrey e 0 . e 0 0 e -2 e .b$xico Qty, ee e * e -4 Gradalajalrs -6 4 e e e -8 e -10 0 5 10 15 20 25 30 3: Public marginality index 1990 105 Figure6.A: Privatemarginalityindex: no convergence. 1990-2000 0 . .....:. -2I . ....... ) 5.0 10.0 15.0 20.0 25.0 30.0 35.0 40.1 -4 0 8 *. -6 . %xicoClty zg .. -8 e * -10 + gK -12 Gwdakjjga . 0 Monterrey -14 -16 -18 Private marginality index 1990 TableA.5: Correlatesof logincomechargesin urbanhouseholdincomeestimation results: Quantile Analysis: 1992-1995 Dependent variable L o g Income Change Presample Period Crisis Period 0bservations 41676 41676 41676 41676 41676 41676 0.2 0.5 0.8 0.2 0.5 0.8 a b C d e f Primary Inc. 0.2324 *** 0.0085 -0.0649 *** 0.2320 *** 0.0358 0.0729 ** Primary 0.1832 *** 0.0047 -0.0552 *** 0.2241 *** 0.0597 *** 0.0662 ** Secondary I n c 0.1084 *** 0.0024 -0.0326 ** 0.1282 * 0.0118 0.0387 Secondary 0.0788 ** 0.0081 -0.0156 0.1665 * 0.0226 0.0278 Young 0.041 1 0.0079 -0.0286 -0.1021 -0.0094 0.0195 0Id -0.1918 *** -0.0467 *** 0.0142 -0.0456 0.001 1 -0.0394 >1.3 Children 0.0278 0.0174 * 0.0039 0.01 87 -0.005 1 0.0124 Single Mothers -0.0466 0.0046 0.0078 -0.001 1 0.0302 0.0306 Single Women -0.1627 *** -0.0550 *** -0.0793 *** 0.2190 ** 0.0615 * 0.1164 *** Single M e n -0.1025 * -0.0305 -0.001 4 0.0309 0.0264 0.0137 Informal Self-Employed -0.2631 *** -0.0685 *** 0.1051 *** 0.0487 -0.0055 0.0181 Informal Salaried -0.0928 *** 0.0196 0.0771 *** 0.0712 -0.0081 -0.0001 N o remuneration -0.1569 *** 0.0595 *** 0.2424 *** -0.4475 *** -0.1244 *** -0.1064 *** Constant -0.4356 *** -0.0176 * 0.5289 *** -0.4056 *** -0.2799 *** -0.3082 *** 106 Table 6-A: Determinantsofincome Log of real income Coefficienl t-statistics Number of infants inthe household -0.317 (2.77) ** Number of children 1-15 inthe household -0.388 (25.49) ** Number of adults inthe household -0.088 (3.65) ** Number of infants inthe household squared 0.035 (0.33) Number of children 1-15 inthe household squared 0.032 (9.76) ** Number of adults inthe household squared 0.013 (4.36) ** Female headedhousehold 0.129 (3.16) ** Shareo f the household members who has migrated 0.125 (5.14) ** Heado fthe household i s Catholic 0.024 (0.96) Age o f the head 0.010 (2.38) * Age o f the head -0.000 (1.81) Years o f schooling o f the head 0.008 (1.10) Years o f schooling o f the head squared 0.003 (6.28) ** Years o f schooling differential between the headandthe highest 0.034 (10.22) ** educatedperson Headi s employed 0.578 (14.19) ** Headoccupation as white collar 0.062 (1.38) Headoccupation as low skilled incapital intensive act 0.110 (4.83) ** Headoccupation as low skilled inlabor intensive act 0.027 (0.76) Headoccupation as agricultural laborer -0.271 (2.54) Head i s indigenous -0.207 (5.45) *** Barrio distance from markets (average distance to work) 0.002 (0.54) Mode o f individual having movedinthe barrio inlast 5 years -0.368 (2.93) ** Modal age o fthe house i s 1-5 years old 0.021 (0.44) Modal age o fthe house i s >10 years old 0.017 (0.50) region== Border -0.047 (0.80) region== North 0.093 (2.46) region== Centre 0.155 (4.22) *** city size== between 50000 and 100000 0.365 (8.49) ** Municipality index o f access to water -0.006 (0.92) Municipality index o f access to sanitation -0.013 (2.09) Municipal level share o f employment inservices -1.233 (4.48) *** Municipal level share o femployment inconstruction 5.620 (4.84) ** Municipal level share o f employment inf-wshareOO -203.508 (5.45) ** % 15 or more are illiterate -0.019 (4.57) ** illiteracy growth between 1990 and 2000 0.315 (1.66) population in 1990 0.000 (3.91) ** Population growth between 1990and2000 1.054 (12.28) ** Constant 8.544 (49.66) ** Observations 13294 Robust t statistics inparentheses * Significant at 5%; ** significant at 1% 107 Table A-7: Simulated barrio average level of income expressed as a proportion of the povertylinein different scenarios [dl- [bl effects o f [dl-[al [bl [CI [dl [cl-[al effects o f City [a1 Equality Equality effects of differences in Baseline Absolute differences in equality inthe between inequalities characteristics barrio barrios inthe barrio o f barrio barrios inhabitants characteristics Aguascalientes, Ags 1.40 1.32 1.40 1.27 0.00 -0.05 -0.013 Tijuana, B.C. 1.87 1.32 1.84 1.37 -0.02 0.06 -0.50 La Paz, B.C.S. 1.78 1.32 1.77 1.49 -0.01 0.18 -0.29 Cd. del Carmen, Ca. 1.47 1.32 1.46 1.53 0.00 0.21 0.07 Cd. Acufia, Coah. 1.26 1.32 1.26 1.26 0.00 -0.06 0.00 Colima, Col. 1.15 1.32 1.16 1.35 0.00 0.04 0.20 San Cristbbal, Chi. 0.74 1.32 0.69 1.06 -0.05 -0.26 0.32 Cd. JuArez, Chi. 1.77 1.32 1.77 1.32 0.00 0.00 -0.45 Cd. Victoria, Dgo. 1.31 1.32 1.30 1.31 -0.01 0.00 0.00 Lebn, Gto. 1.37 1.32 1.36 1.16 -0.01 -0.16 -0.21 Acapulco, Gro. 1.12 1.32 1.12 1.36 0.00 0.04 0.24 Pachuca, Hgo. 1.11 1.32 1.10 1.36 0.00 0.04 0.25 Pto. Vallarta, Jal. 1.60 1.32 1.57 1.43 -0.02 0.12 -0.16 Toluca, Mex. 1.20 1.32 1.19 1.12 0.00 -0.19 -0.07 Uruapan, Mich 1.23 1.32 1.21 1.28 -02.02 -0.04 0.05 Cuernavaca, Mor. 1.20 1.32 1.18 1.65 -0.02 0.33 0.45 Tepic, Nay. 1.45 1.32 1.44 1.53 -0.01 0.22 0.09 Cd. Gral., Escobedo 1.60 1.32 1.59 1.34 -0.01 0.03 -0.26 Sta. Maria Atzompa 1.23 1.32 1.23 1.32 0.00 0.01 0.09 Cd. Tehuacan, Pue. 1.13 1.32 1.12 1.22 -0.01 -0.10 0.09 Querbtaro, Qro. 1.45 1.32 1.44 1.40 -0.01 0.08 -0.05 Canch, Qtroo. 1.59 1.32 1.57 1.29 -0.03 -0.02 -0.30 Cd. Valles, S.L.P. 0.93 1.32 0.93 1.33 0.00 0.01 0.40 Culiacin, Sin. 1.33 1.32 1.34 1.27 0.00 -0.05 -0.06 Villahermosa, Tab. 1.10 1.32 1.07 1.40 -0.03 0.08 0.29 Cd. Victoria, Tamp. 0.96 1.32 0.95 1.25 -0.01 -0.07 0.29 Chiautempan, Talx. 1.30 1.32 1.30 1.46 -0.01 0.15 0.16 Coatzacoalcos, Ver 1.16 1.32 1.14 1.42 -0.02 0.10 0.26 MQida, Yuc. 1.24 1.32 1.24 1.46 0.00 0.14 0.22 Note: [a] Baseline: households have real characteristics, barrios have real characteristics [b] Absolute equality: household have sample average characteristics, barrios have sample average characteristics [c] Equality inthe barrio: households have barrio average characteristics, barrios have real characteristics [d] Equality between barrios: households have real characteristics, barrio have sample average characteristics 108 TableA-8: Determinantsof differentindicatorsof social capital someone in has asked would like to would like to has attended1 i s a member of knows of the barrio for help participate in participate in organization an organization informal could someone in creatingnew providingnew meeting organizationsin provide the barrio infrastructurein services the barrio help barrio Owns home -0.021 -0.023 0.202 (3.82)* * 0.109 (3.82)* * 0.109 0.133 (7.38)* * 0.206 (0.73) (0.77) (5.36)** (4.65)** Share ofthe -0.184 -0.181 0.106 -0.084 -0.084 -0.041 -0.007 householdmembers who has migrated inthe last 5 years (5.63)** (5.42)** (2.36)* (2.63)** (2.63)** (1.29) (0.23) Head ofthe 0.023 0.021 0.063 -0.055 -0.055 -0.254 -0.002 household is Catholic (0.79) (0.71) (1.64) (1.89) (1.89) (9.19)** (0.09) Years of schooling of 0.004 0.005 0.002 0.013 0.013 0.014 -0.003 the head (1.57) (2.09)* (0.47) (5.02)** (5.02)** (5.65)** (1.40) Gender -0.046 -0.056 -0.113 0.158 0.158 0.004 -0.044 (1.92) (2.30)* (3.48)** (6.61)** (6.61)** (0.17) (1.89) age (3.83)* * -0.014 -0.018 0.024 0.058 (15.95)* * 0.058 0.040 0.008 (4.97)** (5.25)** (15.95)** (10.99)** (2.25)* Age squared 0.000 0.000 -0.000 -0.001 -0.001 -0.000 -0.000 (2.83)** (3.88)** (10.01)** (16.20)** (16.20)** (9.51)** (2.78)** Indigenous person -0.086 -0.070 -0.171 0.042 0.042 0.167 0.040 (1.99)* (1.59) (3.19)** (0.98) (0.98) (4.10)** (0.99) Income variance inthe -0.000 -0.000 0.000 0.000 0.000 0.000 -0.000 barrio (3.19)** (2.57)* (1.10) (6.19)** (6.19)* * (5.09)** (1.54) Ethnic variance inbarrio (6.82)* * -0.835 -0.845 -0.051 0.121 0.121 0.496 0.698 (6.76)** (0.29) (0.95) (0.95) (3.95)** (5.66)** Religious variance in 0.494 0.414 -0.296 1.990 1.990 1.610 1.262 barrio (2.77)** (2.28)* (1.22) (10.92)** (10.92)* * (9.12)** (7.13)** Region Border -0.203 -0.126 0.166 -0.441 (9.15)* * -0.441 -0.011 -0.326 (4.30)** (2.63)** (2.42)* (9.15)** (0.24) (6.96)** Region Centre -0.009 0.023 -0.097 -0.086 -0.086 0.169 -0.029 (0.26) (0.62) (1.98)* (2.31)* (2.31)* (4.72)** (0.81) Region South 0.088 0.116 -0.125 -0.228 -0.228 0.052 -0.603 (2.06)* (2.68)** (2.19)* (5.22)** (5.22)** (1.22) (14.30)** Large city 0.196 0.206 -0.031 -0.103 -0.103 0.039 -0.113 (4.03)** (4.14)** (0.48) (2.12)* (2.12)* (0.81) (2.45)* Fast growing city 0.034 0.028 -0.127 -0.100 -0.100 -0.204 -0.384 (1.25) (1.01) (3.37)** (3.64)** (3.64)** (7.83)** (14.50)** Modal age ofthe house -0.257 -0.262 0.186 0.151 0.151 -0.025 0.270 is 5-10 years old (8.44)** (8.46)** (4.04)** (4.51)** (4.51)** (0.78) (8.36)** Modal age ofthe house -0.238 -0.231 -0.099 -0.119 -0.119 -0.104 -0.245 is >10 years old (5.53)** (5.24)** (1.69) (2.70)** (2.70)** (2.34)* (5.62)** Barrio average ofhaving -0.948 0.795 0.795 0.956 1.914 someone to help (5.07)** (5.66)** (5.66)** (6.77)** (13.84)* * Constant 0.016 -0.010 1.575 -2.097 -2.097 -2.456 -0.622 (0.12) (0.07) (8.28)** (14.47)** (14.47)** (17.19)** (4.46)** Absolute value ofz statistics inparentheses, * significant at 5%; ** significant at 1% Table A-9: Probabilityofbeingemployed 109 employment coefficient z-statistic Gender -1.262 (50.88) ** Years 0.147 (40.10) ** Years squared -0.002 (40.12) ** Movedinthe last 5 years to the barrio 0.067 (1.15) Changes house inthe last year -0.040 (1.19) Catholic 0.014 (0.58) Indigenous 0.033 (0.83) Literacy -0.185 (2.81) ** Years of schooling 0.022 (2.17) Years of schooling squared 0.002 (3.10) *** Household headi s female 0.424 (10.63) ** Share o f the household members who has migrated inthe 0.164 (2.40) * last 5 years Numberofhouseholdmembers ofworking age 0.363 (41.53) ** Share o f unemployed inthe barrio -5.590 (7.24) ** Married -0.777 (29.07) ** Has asked for help someone inthe barrio 0.203 (4.41) ** Barrio average o f having someone to help 0.159 (1.17) Region Border 0.033 (0.95) Region Centre -0.092 (3.15) ** Region South -0.071 (2.13) * Large city 0.051 (1.35) Fast growing city 0.044 (2.10) Modal age o f the house i s 1-5 years old 0.114 (4.66) *** Interaction access to network andunemployment -1.756 (1.71) Interaction access to network and household members in -0.034 (2.09) * the Labor Force Constant -2.169 (15.31) ** Observations 31016 110 TableA-10: Barrio leveldeterminantsof numberof crimes Number of crimes reported Smaller city -83.4 (less than 100000 inhabitants) (3.41)"" Barrio average of having someone to help -231.6 (2.44)" Barrio average of would like to participate in creating -49.7 new infrastructureinbarrio (2.19)" Average income 0.061 (2.89)"" Average income squared -0.000 (2.32)" Income variance inthe barrio -0.013 (3.O 1)"* Population ofthe settlement -0.003 (3.00)" * Age ofthe settlement 21.4 (1.42) Public lightning -98.2 (2.86)" Policinginthe streets -240.4 (3.77)" * Frequency of patrols 198.6 (4.00)" * Constant 43.2 (0.29) Observations 31 R-squaredadjusted 0.493 111 Bibliography Alegria Olazabal, Tito and Gerard0 Ordbiiez Barba (2004). Legalizando la Ciudad : Asentamientos Informales y Procesosde Regularizacibn en Tijuana. - Mexico : Colegio de laFronteraNorte. (Processed). Alesina, Albert0 and Eliana La Ferrara (2002). "Who Trusts Others?". - In: Journal of Public Economics 85(2): 207-234. Alonso, William (1964). Location and Land Use : Toward a General Theory of Land Rent. - Cambridge :HarvardUniversity Press. Araujo, C. (2004). "Spatial Heterogeneity in Well-being across Mexican Municipalities". - Washington, D.C. : The World Bank, (Processed). Baker, Judy and Somik La11(2003). A Profile of Urban Poverty in Latin America. Background Paper for the Urban Poverty Report. -Washington, D.C. : The World Bank, (Processed). Bakker, Nico, ed. (2000). Growing Cities, Growing Food : Urban Agriculture and the Policy Agenda. Bonn Germany : German Foundation for International Development. - Beezley, William (2003). The Hound of Los Pinos and the Return of Oscar Lewis : Understanding Urban Poverty in Mexico. Paper Presented at the Workshop on Perspective on Urban Poverty in Latin America (Washington, D.C. September 17, 2004). -- Woodrow Wilson International Centre for Scholars. Bendini, M., 0. Gracia and Mauricio Santamaria (2004). Short Run Output Shocks and Labor Market Performance : the Case of Mexico. - Washington, D.C. : The World Bank, (mimeo). Blau, David M. and Philip K. Robins (1990). "Job Search Outcomes for the Employed and Unemployed". -In:JournalofPolitical economy98(3): 637-655. Boltvinik, J. and Martha Schteingart (1997). "La Magnitud de la Pobreza en las Colonias".-In: Pobreza, Condiciones de Vida T, Salud en la Ciudad de Mexico.-Mexico : El Colegio de Mexico, (1997). -- pp. 427-478. Browning, Martin and Tomas F. Crossley (1999). Shocks, Stocks and Socks : Consumption Smoothing andthe Replacement of Durables During an Unemployment Spell. - Canada : CanadianInternational Labour Network. Budar, Oscar and Rodrigo Garcia Verdu (2004). UnAnalisis Dinamico del Mercado Laboral en Mexico. - Mexico : Banco de Mexico, Direccibn General de Investigacibn Econbmica. - (Documentos de Investigacibn Mayo 2004), (Forthcoming). Calderbn-Madrid, h g e ly BelemTrejo (2002). The Impact of the Mexican Training Program for Unemployed Workers on Re-employment Dynamics and on Earnings. - Washington, D.C. :Inter-American Development Bank. 112 Cardoso, Adalbert, Peter Elias, and Valeria Per0 (2003). "Urban Regeneration and Spatial Discrimination: the Case of Favelas Residents in Rio." - In: Anais do XXXI Encontro Nacional de Economia [Proceedings of the 31th Brazilian Economics Meeting] f41, ANPEC. - Brazil: Associaqgo Nacional do Centros de Pos-graduaqgo em Economia [Brazilian Association of GraduatePrograms inEconomics] Carrasco, Tania, Kathryn Johns-Swartz, and Augusta Molnar (2003). "Pueblos Indigenas, Identidad, y Pobrezaen las Ciudades de Mexico, Cancun, y Coatzacoalcos-Minatitlan". - In:Exclusibn Social T, Estrategias de Vida de 10s Indigenas Urbanos en Peni, Mexico, T, Ecuador.-Washington, D.C. :The World Bank. CIESAS (2003). PobrezaUrbana en Mexico : Estudio Cualitativo sobreNecesidadesy Formas de Organizacibn Social de Poblaciones Bajo Condiciones de Pobreza en 31 Ciudades de Mexico. Mexico, (Processed). CONAPO (2002). indice de Marginacibn Urbana 2000. http://www. conapo.gob.mx/publicaciones/indice2000.htm Cortez, Willy W. (2001). Wage Convergence in Major Mexican Cities : 1987-2000.- Mexico : Universidad de Guadalajara, Departamentode Metodos Cuantitativos. Cragg, M. I.and M. Epelbaum (1996). "The Premium for Skills in LDCs : Evidence from Mexico". - In:JournalofDevelopment Economics 5l(1): 99-116. Datta, K. and G.A. Jones (2001). "Housing and Finance in Developing Countries : Invisible Issues on ResearchandPolicyAgendas". - In:Hhbitat International (25)Issue 3. Davila, E.R. (2001). Mexico the Evolutionand Reform ofthe Labor Market. -- Washington, D.C. :The Brookings Institution. Davidson, Russell and James G. MacKinnon (2000). "Bootstrap Tests: How Many Bootstraps?'. - In:Econometric Reviews 19(1): 55-68. D e Ferranti, David. [et al.] (2003). Closing the Gap inEducation and Technology. -- Washington, D.C. :The World Bank. DiPasquale, Denise and Edward L. Glaeser (1999). "Incentives and Social Capital : Are Homeowners Better Citizens?'. - In:JournalofUrbanEconomics 45(2): 354-384. Durlauf, StevenN.(2001). The Memberships Theory of Poverty : The Role of Group Affiliations in Determining Socioeconomic Outcomes. -- New York : Ruse11 Sage Foundation Cambridge and London : HarvardUniversity Press. Escobar Latapi, Agustin and Mercedes Gonzalez de la Rocha (2002). Evaluacibn Cualitativa del Programa de Desarrollo Oportunidades: Seguimiento de Impact0 2001-02 Comunidades de 2,500 a 50,000 habitantes". - Mexico : Centro de Investigaciones y Estudios Superioresen Antropologia Social. (2004). "Evaluacibn Cualitativa del Programa Oportunidades en Zonas Urbanas : 2003". - In: Resultados de la Evaluacibn Externa del Programa de Desarrollo Humano 113 Oportunidades. - Mexico : Instituto Nacional de Salud Publica : Oportunidades : Centro de Investigacionesy Estudios Superioresen Antropologia Social. Fajnzylber, Pablo, Daniel Lederman Norman Loayza (2000). "Crime and Victimization: An Economic Perspective". - In: Journal of the Latin American and Caribbean Economic Association l(1): 219-278. (2002). "Inequality and Violent Crime". - In:Journal of Law andEconomics 45 (1): 1- 40. (2003). William F. Maloney and Gabriel Montes. LDC Micro Firm Dynamics : How Similar are They to Those in the Industrialized World? Evidence from Mexico. - Washington, D.C. : The World Bank, (mimeo). Fay, Marianne and Anna Wellenstein (2005a). "Keeping a Roof Over the One's Head -Shelter, Land and Vulnerability to Natural Disaster". In: The Urban Poor in Latin America. - Washington, D.C.: The World Bank, (Processed). and Lorena Cohan and Karla McEvoy (2005b). "Public Social Safety Nets and the Urban Poor". - In: The Urban Poor in Latin America. - Washington, D.C.: The World Bank, (Processed). and CaterinaRuggeri Laderchi (2005~)."Relying on Oneself : Assets ofthe Poor". - In: The Urban Poor inLatin America. - Washington, D.C. :The World Bank, (Processed). (2005d). "Urban Poverty in Latin America". - In:The Urban Poor inLatin America. - Washington, D.C. : The World Bank, (Processed). Feenstra, R.C., and G.H. Hanson G.H. (1997). "Foreign Direct Investment and Relative Wages : Evidence from Mexico's Maquiladoras". - In: Journal of International Economics (42): 371-393. Field, Erica (2002). Entitled to Work : Urban Property Rights and Labor Supply in Peru. - Princeton, N.J. : Princeton University, (Processed). Filmer, Deon and Lant H.Pritchett (2001). "Estimating Wealth Effects without Expenditure Data - Or Tears : An Application to Educational Enrollments in States of India". - In: Demography 38(1) February 115-132. Galasso, Emanuela and Martin Ravallion (2004). Social Protection in a Crisis : Argentina's Plan Jefes y Jefas.- Washington, D.C. : The World Bank. - (Policy ResearchWorking Paper 3165). Gilbert, Alan andAnn Varley (1991). Landlord and Tenant : Housing the Poor inUrban Mexico. - New York :Routledge. Gill, Indermit S., Truman Packard and Juan Yermo (2004). Keeping the Promise of Old Age Income Security in Latin America : A Regional Study of Social Security Reforms. - Washington, D.C. : The World Bank. 114 Goetz, Edward G. (2003). Clearing the Way : Deconcentrating the Poor in Urban America. - Washington, D.C. :Urban Institute Press. Gough, Katherine (1998). "House for Sale?The Self-help Housing Market inPereira Colombia". -In: Housing Studies (13)2: 149-160. Gould, W. W. (1992). "Quantile Regression with Bootstrapped Standard Errors". - In: m a Technical Bulletin9: 19-21. (1997). "Interquantile and Simultaneous-Quantile Regression.". - In: Stata Technical Bulletin38: 14-22.Reprintedin StataTechnical BulletinReprints7: 167-176. Grootaert, Christiaan [et al.] (2003). Integrated Questionnaire for the Measurement of Social Capital (SC-IQ). - Washington, D.C. : The World Bank, Social Capital Thematic Group, (Processed). Gutierrez, Juan Pablo, Stefan0 Bertozzi and Paul Gertler (2003). Evaluacibn de la Identificacibn de Familias Beneficiarias en el Medio Urban0 : Evaluacibn de Resultados de Impact0 del Programa de Desarrollo Humano Oportunidades.- Mexico : Instituto Nacional de Salud Publica. Heckman, J. and Carmen Pages (2000). "The Cost of Job Security Regularization : Evidence from Latin American Labor Markets". -In:NBERWorking Papers No. 7773. Hernandez Laos, E.; N. Garro and I.Llamas (1998). "Productividad y Mercado de Trabajo en Mexico". -In: CEM on Mexico: Enhancing Factor Productivitv Growth. - Washington, D.C. : World Bank. (Background Paper). - Inter-American Development Bank (2004). Good Jobs Wanted : Labor Markets inLatin America. - Washington D.C. : Inter-American Development Bank. Kaplan, D., G. Martinez Gonzalez and R. Robertson(2003). Worker- and Job-flows inMexico. - Mexico : Instituto Tecnolbgico Autbnomo de Mexico : Instituto Mexican0 del Seguro Social :Macalester Collage, (Unpublished). Katz, Lawrence F., and Murphy, Kevin M. (1992). "Changes in Relative Wages, 1963-87: Supply andDemandFactors". - In:Quarterly JournalofEconomics 107: 35-78. Kennedy, Peter E. (1981). "Estimation with Corrrectly Interpreted Dummy Variables in Semilogarithmic Equations". - In:American Economic Review 71(4): 801. Knack, Stephen and Philip Keefer (1997). "Does Social Capital Have an Economic Payoff! : A Cross-Country Investigation". - In:Quarterlv JournalofEconomics 112(4): 1251-1288. Kochar, Anjini (2000). Designing Household Survey Questionnaires for Developing Countries : Lessons from Fifteen Years of the Living StandardsMeasurement Study. - Washington, D.C. : The World Bank. Vol. 2, pp. 183-209. - Kugler, A. (2000). The Impact of Firing Costs on Turnover and Unemployment : Evidence from the Colombian Labor Market Reform. - Washington, D.C. : Inter-American Development Bank, ResearchNetwork. (Working Paper No. R-393). - 115 Lewis, Oscar (1959). Five Families : Mexican Case Studies in the Culture of Poverty. - New York :Basic Books. Loayza, Norma, Pablo Fajnzylber and Cesar Calderbn (2005). Economic Growth in Latin America and the Caribbean : Stylized Facts, Explanations, and Forecasts. - Washington, D.C. : The World Bank. (Report No.321SO). - Maloney, William F. (1999). Does Informality Imply Segmentation in Urban Labor Markets? : Evidence from Sectoral Transitions in Mexico. -In:The World Bank Economic Review 13(2): 275-302. [et al.] (2003). Notes on Income and Consumption Shocks in Mexico. - Washington, D.C. : The World Bank, LatinAmerica and CaribbeanDepartment, (mimeo). Matin, I.; Hulme and S. Rutherford (1999). Financial Services for the Poor and the Poorest : D. Deepening Understanding to Improve Provision, Finance and Development Research Programme. -University Manchester. (Working Paper Series PaperNo. 9) - McKenzie, David J. (2003a). "How do Households Cope with Aggregate Shocks? : Evidence from the Mexican Peso Crisis". - In:World Development 3l(7): 1179-1199. (2003b) "Measuring inequality with asset indicators".- In:BREADWorking PaperNo. 042. (2003~).The Consumer Response to the Mexican Peso Crisis. - Department of Economics, StanfordUniversity, (mimeo). Mincer, Jacob (1974). "Education, Experience, and Earnings". New York: Columbia University Press. Monkkonen, P. (2004). "Measurements in a Highly Polluted Asian Mega City: Observations of Aerosol Number Size Distribution, Modal Parameters and Nucleation Events". - In: Atmospheric Chemistry andPhysics Discussions, 4: 5407-5431 Montes, Gabriel and Mauricio Santamaria (2004). Poverty and Labor Markets in Urban Mexico. - Washington D.C. : The World Bank, (Processed). Moser, Caroline; Alisa Winton and Annalise Moser (2004). "Violence, Fear and Insecurity and the Urban Poor in Latin America". - In: The Urban Poor in Latin America. -- Washington, D.C. : The World Bank, (Processed). Mosqueira, Edgardo (2003). PresentationMade at Conference inMexico City, May 2003. Munshi,Kaivan (2003). Networks inthe Modern Economy :Mexican Migrants inthe U S Labor Market. - Pennsylvania, PA : University of Pennsylvania, Department of Economics, (Processed). Muth,R.(1969). Cities andHousing. Chicago :University ofChicagoPress. - Narayan, Deepa [et al.] (2000). Voices of the Poor : Crying out for Change.- Oxford : Oxford University Press :The World Bank. 116 National Academy (1990). Inner-city Poverty in the Unite States. - Washington, D.C. : The National Academies Press. National ResearchCouncil of the National Academies (2003). Cities Transformed :Demographic Change and its Implications inthe Developing World. -Washington, D.C. :The National Academies Press. New York Times. (2004). Debate Stirs Over Tiny Loans for World's Poorest.-(April 29, Section A, page 1). Ortiz, Alexandra andAlain Bertaud (2001). "Land Markets andUrban Management : The Role of Planning Tools". In: The Challenge of urban Government : Policies and Practices. - Washington, D.C. :The World Bank Institute. - pp. 239-252. Pages, Carmen, and Claudio E. Montenegro (1999). Job Security and the Age-Composition of Employment : Evidence from Chile. -Washington, D.C. : . Inter-American Development Bank. (Working Paper No. 398). - Perlman, Janice E. (2003). Marginality from Myth to Reality in the Favelas of Rio de Janeiro, 1969-2002. http://wbln00 18.worldbank.org/LAC/lacinfoclient.nsf/49a0 102c9b95cf028525664b006al 7a4/e249d2b285efe58185256f9a0050cbbd/$FILE/From%2OM~h%20to%2OReality%2Oi n%20Rios%20favelas%20- 1969-2002.pdf Presidenciade la Republica, Oficina de la Presidenciapara las Politicas Publicas (2003). "Public Policies for the Alleviation of Moderate Poverty and the Promotion of Access to Formal Economic Activities : IMPULSO : a new strategy ofthe Fox Administration". - Mexico : Presidencia, (Powerpoint Presentation). Putnam, R. D., R. Leonardi and R. Y. Nanetti (1993). Making Democracy Work : Civic Traditions inModern Italy. Princeton, N.J. : Princeton University Press. - Rakodi, Carole (1995). "Rental Tenure in the Cities of Developing Countries". - In: Urban Studies 32(4-5). Ravallion, Martin (1997). "Can High-Inequality Developing Countries Escape Absolute Poverty?'. - In:Economics Letters, 56: 51-57. (1999a). "Appraising Workfare". - In: The World Bank Research Observer 14(1) February :31-48. and Quentin Wodon (1999b). Poor Areas or Only Poor People?. - Washington, D.C. : The World Bank. (Policy ResearchWorking Paper Series No.1798), (Processed). - and Jyotsna Jyotsna. (1999~).Spatial Poverty Traps?. Washington, D.C. : The World - Bank. (Policy ResearchWorking Paper Series No. 1862), (Processed). - (2000). "Monitoring Targeting Performancewhen Decentralized Allocations to the Poor are Unobserved". In:The World Bank Economic Review (14)2 May: 331-345. - 117 Retsinas, Nicolas P. and Erick S. Belsky, E.S (2002). "Examining the Unexamined Goal". - In: Low Income Homeownership : Examining the Unexamined Goal. . -Washington, D.C. : Brookings Institution Press. -Chapter 1. Revenga, Ana (1997). "Employment and Wage Effects of Trade Liberalization : The Case of Mexican Manufacturing". -In:Journal ofLabor EconomicsNo. 15. Robertson, R. (2000). "Wage Shocks and North-American Labor Market Integration". - In: American Economic Review (90): 742-764. Ruggeri Laderchi, Caterina (2004). "Working One's Way Up : The Urban Poor and the Labor Market". -- In: The Urban Poor in Latin America. -- Washington, D.C. : The World Bank, (Processed). Sampson, Robert J., Jeffrey D. Morenoff and Thomas Gannon-Rowley (2002). "Assessing Neighborhood Effects : Social Processes andNew Directions inResearch" .- In:Annual Review of Sociology. Vol. 28, pp. 443-478. - Schady, N.and C. Sanchez-Paramo(2003). Off and Running?Technology, Trade, and the Rising Demand for Skilled Workers inLatin America. - Washington, D.C. : The World Bank. - (World Bank ResearchDepartmentWorking Paper No. 3015). Scott, John (2003). "Decentralization of Social Spending in Mexico". - Mexico : Centro de Investigacibn y Docencia Econbmicas (CIDE). Secretaria de Desarrollo Social (SEDESOL) (2000). Programa Nacional de Desarrollo Social : 2001-2006. Mexico : SEDESOL. - (2003a). Acuerdo por el que se Modifican las Reglas de Operacibn del Programa Hhbitat para el ejercicio fiscal 2004. Publicadas el 25 de Marzo de 2003. Mexico : Diario Oficial de la Federacibn, Primera Seccibn, Martes 23 de 2004. -- pp. 16-67. (2003b). ProgramaInstitucional Oportunidades:2002-2006. Mexico. (2004). Acuerdo por e l que se Emiteny Publican las Reglas de Operacibn del Programa de Ahorro, Subsidio y Credit0 para la Vivienda Progresiva "Tu Casa", para el ejercicio fiscal 2004. -- Diario Oficial de la Federacibn, Segunda Seccibn, Viernes 7 de Mayo de 2004. -- pp. 196-205. Siaens, Corinne and Quentin T. Wodon (2003). Latest Estimates of National Urban and Rural Poverty inLatin America. - Washington, D.C. : The World Bank, (Processed). Skoufias, Emmanuel (2004). Households Vulnerability and its determinants in rural and urban Mexico : Background Paper for the Mexico Poverty Report. - Washington, D.C. : The World Bank, (Processed). Soloaga, Isidro and Mario Torres (2003). Agricultural Growth and Poverty Reduction. --Paper Prepared for the "Role of Agriculture" International Conference, October 2003 in Rome FAO. http://www.fao.org/es/ESA/Roa/pdf/3~Poverty/Poverty~Mexico.pdf 118 Subbarao, Kalanidi (2003). Systemic Shocks and Social Protection : Role and Effectiveness of Public Works Programs. - Washington, D.C. : The World Bank. - (Social Protection Discussion Paper Series ;No. 0302). Szekely, Miguel (1998). The Economics of Poverty, Inequality and Wealth Accumulation in Mexico. -- New York : St. Martin's Press. The Urban Poor in Latin America (2005). Ed. Marianne Fay - Washington, D.C. : The World Bank. - 266 p. (Directions inDevelopment). - Van Bronkhorst, Bernice (2003). A Resource Guide for Municipalities : Community Based Crime and Violence Prevention in Latin America. -Washington, D.C. : The World Bank. http://wbln00 18.worldbank.or,g/LAC/lacinfoclient.nsf/d29684951174975c85256735007f ef12/638a8cac15f058f285256ea9006bO77e/$FILE/ResourceGuide~urbanviolence.pdf Van den Bosch, Karel (1998). "Poverty and Assets in Belgium". -In: Review of Income and Wealth ;Series 44, No. 2, June 1998. Wahba, Jackline and Yves Zenou (2003). Density, Social Networks and Job Search Methods : Theory and Application to Egypt. - Goswell Road, Loandon : Center for Economic Policy Research.- (CEPR Discussion PapersNo. 3967). Woolcock, Michael (2004). "Calling on friends and Relatives : Social Capital".- In:The Urban Poor inLatin America. -- Washington, D.C. : The World Bank, (Processed). World Bank (2000). Mexico Earnings Inequality After Mexico's Economic and Educational. - Washington, D.C. :The World Bank. (Report No. 19945-ME). - (2002a). Mexico Low-Income Housing: Issues and Options. -- Washington, D.C. : The World Bank. (Report No.22534-ME). - (2002b). Urban Service Delivery and the Poor : The Case of Three Central American Cities. Washington, D.C. : The World Bank. (Report No. 22590). - - (2003a). Brazil : Access to Financial Services. - Washington, D.C. : The World Bank. (Report No. 27773-BR). - (2003b). Broadening Access to Financial Services Among the Urban Population : Mexico City's Unbaked. -Washington, D.C. : The World Bank, (Processed). (2004a). Poverty in Mexico : An Assessment of Conditions, Trends and Government Strategy. -Washington, D.C. :The World Bank. (Report No. 28612-ME). - (2004b). Mexico Public Expenditure Review. - Washington, D.C. : The World Bank. - (Report No. 27894-ME). (2004~).The Urban Poor in Latin America. - Washington, D.C. : The World Bank. - (Report No. 30465). 119 (2005a). A Study of Rural Poverty in Mexico. - Washington, D.C. : The World Bank. - (Report No. 32929-MX). (2005b). Mexico: An Overview of Social Protection. Washington, D.C. : The World - Bank. (Report No. 32929-MX). - (2005~).Mexico Infrastructure Public Expenditure Review.- Washington, D.C. : The World Bank, Washington, (Processed). 120