Report No. 32929 - MX Mexico Income Generation and Social Protection for the Poor (In Four Volumes) Volume III: An Overview of Social Protection 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 Table Of Contents Introduction 1 Household Vulnerability inMexico 7 Mexico's Social Protection System: Overview and Assessment 28 Towards an Integrated Social Protection Strategy inMexico: A Discussion of 54 Issues and Options Appendix 69 Bibliography 76 ... 111 1.INTRODUCTION This report provides a strategic overview of Mexico's federal social protection system, which i s comprised o f both social insurance and social assistance programs. It assesses its performance to date regarding income risk management for vulnerable groups, and identifies options for stepping up that performance. In doing so it responds to an increasing consensus in Mexico regarding the need for a major social protection policy reform, even if the direction o f those reforms i s still in flux. The report i s designed to provide a first stage diagnostic o f major issues facing the social protection system, as an input to ongoing discussion and debate in Mexico, and as a platform for further technical work on the specifics o f reform. The report focuses on the role of federal government in the design and implementation of social protection policies, while recognizing that sub-national governments are playing an important role in the provision o f these services, as will be explored in the next phase o f the World Bank's Mexico Programmatic Poverty work. In doing so, it draws upon the first phase o f the Mexico Programmatic Poverty Work as well as the urban, rural and vulnerability studies beingconductedinparallel with this analysis. The report is organized into four chapters. Chapter 1presents a briefreview of social protection concepts, and establishes a framework for determiningthe optimal role for government in the provision of social protection (risk management) tools, as distinct from market-provided risk management tools and those undertaken informally by and between households. Chapter 2 discusses the major sources o f income vulnerability in Mexico, identifying the major risks that Mexican families faces, and explores the extent to which households have access to private risk management mechanisms for addressing these risks. Chapter 3 provides an overview and assessment o f the social protection system inMexico, and Chapter 4 discusses issues and options. WHAT I S SOCIALPROTECTION? Social protection is a set of tools to manage income risk, ranging from formal mechanisms such as old-age pensions, health insurance and transfer programs to informal mechanisms such as remittances. Social protection has an important role to play in addressing not only short-term vulnerability, preventing households from temporarily falling into poverty, but a longer-term role in reducing risks that perpetuate chronic poverty. It is based on the idea that individuals, households, and communities are exposed to multiple income risks from different sources, both natural and man-made, individual and covariant (World Bank, 200l), that can lead to welfare losses unless public or private social protection mechanisms are available to assist households inmanaging risk. WHAT I S THE ROLE OF THE GOVERNMENTINPROVIDINGSOCIAL PROTECTION? There are two basic rationale for governments to provide social protection: (i)to address market failures that leave families or individuals without sufficient risk-pooling mechanisms to cope with shocks, and (ii) achieve redistributivegoals, including the protection to 1 o f vulnerable groups such as the illor disabled, and via the provision o f asset-building strategies, such as human capital development. The failure o f private insurance markets to provide viable risk-management mechanisms provide a clear rationale for policy intervention by the government in those markets, in the form of public (or publicly-regulated) risk-pooling programs such as unemployment and health insurance. Income redistribution via social assistance programs (income transfers and asset-building programs to poor households) form the second core o f social protection policy, and the extent to which governments take on this second role o f social protection policy i s largely a matter o f choice andpolitical will. Social protection policies can be grouped into two major categories: social insurance and social assistance. Social insurance mechanisms are those where participants pay into the system and withdraw benefits when eligible, effectively pooling risk across all program participants. Social assistance programs are comprised o f government transfers to targeted populations, funded by general tax revenues. Yet in practice the boundary between these two types of programs can break down. Social insurance programs may be partially funded from general tax revenues, such that a component o f the benefit becomes in effect a subsidy. Social assistanceprograms, on the other hand, ifnot rationed by means o f a budget constraint but instead available to anyone in the general population displaying conditions o f eligibility, simply become risk-pooling mechanisms across the entire general population; contributions are paid via the tax system, and benefits drawn out according to need (World Bank, 2004). What are the circumstances under which the government should intervene to provide social protection, and what is the optimal type of intervention? Building on World Bank, 2004, which draws on the comprehensive insurance framework (Erlich and Becker, 1972), we establish a basic typology o f income risks, based on the two factors determiningthe cost o f insuringagainst a risk, which are: (i) frequency ofthe risk, from rare to constant, and (ii) the size o f the loss resulting from the risk, from small to catastrophic. From this typology we establish four main categories o f risk, as illustrated in Figure 1: small and rare (dog bite), small and constant (minor illness), catastrophic and rare ('job loss), catastrophic and constant (chronic poverty). Then we match these four major categories o f risk with the consequent ideal social protection response, and the role for the government in providing that response. The role for the government i s minimal inthe bottomtwo quadrants o f the figure -where informal mechanisms by and betweenhouseholds can be largely successful in managing risk- and becomes larger as the size o f the potential loss moves up the figure to catastrophic losses, where the role for the government in social protection provision i s greatest. Similarly, the type of intervention that i s ideal differs from left to right -rare losses (to the right o f the figure) are most efficiently dealt with via risk-pooling (insurance, either provided by the market or the government) whereas frequent and predictable income losses, which cannot be cost-effectively insuredagainst, are most effectively addressedvia re-distributivesocial assistancetransfers (either from the government, or betweenhouseholds). 2 Figure1.1 SocialProtectionResponses and Role for Government, by Size and Frequency of Income Risks Cata rophic Catastrophic and Rare i.e. unexpected job loss - Catastrophic and Frequent i.e. income volatility transient poverty associatedwith informal employment, or permanent income loss due to disability or old age - and low asset base with which to manage income shocks [deal SP response: risk-pooling Ideal SP response: re-distributive transfers, and incentives for savingshuman capital investment Goal: address market failure (where insurance Goal: income smoothing, prevention o fnegative coping, markets fail to provide) andincreasedhumancapital Size Potential Role for Government: HIGHprovide Potential Role for Government: HIGHprovide social of social insurance schemes (funded risk-pooling assistance funded from general tax revenues (risk pooling loss mechanisms) where market fails to provide, i.e. across entire population) andwhere age-appropriate, workfare or unemployment insurance. incentives to invest inhuman capital. INSURANCE ASSISTANCE Small and Rare i.e. dog bite Small and Constant i.e. common cold Ideal SP response: private riskmanagement, Le. Ideal SP response: private riskmanagement, Le. self- coping insurance (saving) Potential Role for Government: MINIMAL - Potential Role for Government: MINIMAL - regulation except incentives for positive coping behavior of capitalmarkets to facilitate access to savings mechanisms Small Rare Frequent I Frequency ofloss I Appropriate social protection policy design takes into account both the potential negative effect of crowding out private transfers, and the potential positive effects of replacing detrimental private coping and mitigation strategies with formal mechanisms. Public social protection -be it in the form of risk pooling or in the form o f targeted redistribution- interacts with informal risk mitigating and coping mechanisms, including private transfers. One key question for policymakers is whether or not a public intervention will simply displace private risk management mechanisms, such as market insurance, or informal, inter-household transfer systems. In insurance, the appropriate role for government i s in regulating private insurance markets, and stepping in to provide public risk-pooling programs where markets fail (such as in the case o f unemployment insurance). In targeted redistribution efforts, programs need to take into account the existence o f private transfer networks, including monetary transfers as well as the exchange o f commodities (food, clothing etc.) and assistance such as help with child care. If public transfers crowd out private transfers, government intervention contributes less to recipients level and safety o f income and consumption than initially thought.' Put differently, private safety 1 This is particularly true for the level and fluctuation of consumption. Informal risk coping strategies, including private transfers, are insurance mechanisms used to protect consumption from adverse shocks to income. As a result, we would expect income to fluctuate more than consumption. Since welfare ultimately 3 nets might fill the gaps in the absence o f publicly funded programs, and may at times alleviate poverty more efficiently than government intervention (Cox and Jimenez, 1990). However, a second set o f questions relate to `negative' coping mechanisms which increase a household's vulnerability to risk and poverty in the future. Damaging behavioral responses to shocks adopted by the poor in order to mitigate (ex-ante) and cope (ex-post) with risk include, among others: the sale o f productive assets, adverse incorporation under the wings of a patron, safer but low-return production techniques, and reduced investments in physical and human capital. Skoufias (2004) stresses the need to better understand the nature o f private insurance strategies, pointing to the fact that a displacement o f informal means o f consumption insurance i s not necessarily a bad consequence o f government intervention. Ideally, optimal policy design takes into account both the potential negative effect o f public transfers crowding out private transfers, and the potential positive effects o f replacing detrimental private coping and mitigation strategies with formal mechanisms. The poor typically have relatively weak capacity for self-insurance, and limited ability to pool risks over and above extended families or other informal mechanisms. These informal mechanisms tend to be costly, and to break down in the context o f covariate risks. Government action can be crucial inthese areas. Two main features of the role for publicly-providedsocial protectionemerge from this typology. First, the role for government in the provision of social protection services or programs varies from minimal to high depending on the size o f the potential income loss. Inthe bottom two quadrants o f the graph, the size o f the income loss, where the risk should occur, i s low. Public resources are much less effectively spent to mitigate small risks in the bottom two quadrants, where private risk management mechanisms optimally predominate. Conversely, the potentialrole for government grows with the size o f the potential loss. Inthe top two quadrants o f the figure, where the size o f the potential loss i s high, public resources can very effectively be spent to address these types o f risks, as long as interventions are designed to complement rather than replace riskmanagement tools that markets or households provide for themselves. Second, the optimal type of intervention - risk-pooling (insurance) to address market failure vs. targeted redistribution(assistance) - depends on the frequency of the risk.Risksthat are rare can be insuredagainst, andtherefore are most appropriately managedvia risk-pooling mechanisms among those exposed to the risk. Risks that are frequent cannot be effectively managedvia risk-pooling, and are more suitedto explicit social assistancevia targeted income transfers. Thus the figure also illustrates the dichotomy between the two major roles for social protection: insurance (where markets fail to provide risk-pooling among households faced with relatively rare risks, that nevertheless induce high costs should they occur) and social assistance via redistribution (where households face frequent -or permanent- risks to income coupled with having a low asset base with which to manage risk, such that even small shocks to income can be catastrophic). For example, for populations vulnerable to `catastrophic and rare losses' (such as job loss), shown in the upper left quadrant o f the figure, the optimal role for the government i s to provide funded risk-pooling mechanisms (e.g., workfare). The top right quadrant illustrates populations facing frequent income losses, and for whom the impact o f the shock can be catastrophic due to an insufficient asset base to effectively manage those risks. The derives from consumption and since consumption accounts for households' private efforts at protection, it is generally the preferred measure. It is, however, worth noting that risk mitigating strategies that successfully lower variability of consumption by lowering variability of income lead to reduced income inthe future and a permanent fall inmedian consumption levels. Focusing on variability of consumption should therefore be accompanied by a view to changes in the long run average level of consumption (see Skoufias 2004 for a review of the main arguments from economic theory). It is, however, worth noting that some risk mitigating strategies successfully lower variability of consumption. 4 population inthis quadrant often faces chronic poverty inthe absence o f public interventions, and includes, for example, the poor self-employed with a permanently high degree o f variability in earnings, and the poor who move into old age who experience repeated shocks as they continue to engage inthe labor market but suffer decreasedproductivity/ability to work. For this type o f risk, the role for the government i s again high, and the optimal public intervention i s targeted, re- distributive transfers (such as conditional cash transfers, workfare, and old age minimum pensions). Some o f these programs, such as conditional cash transfers, can be tied to human capital development where these are age-appropriate, in order to reduce recidivism and facilitate the exit from chronic poverty (see Box 1.1) Box 1.1: Social protection and chronic poverty Socialprotection has a particular role to play in addressingchronic poverty, notably by moving beyond risk-pooling to the use of targeted transfers. The combination o f material poverty, extreme capability deprivation and various forms o f social exclusion experienced by the chronic poor group calls for a multi-dimensional response. Formal mechanisms for risk mitigation and coping are central to successful social protection strategies not only because they help prevent individuals falling into chronic poverty, but also because they facilitate exit from chronic poverty while preventingrecidivism. Access to formal risk-pooling mechanisms matter to the chronically poor. The cost o f realized risks i s greater for this group due to lower endowments o f physical and human assets which weaken the ability to manage shocks. Evidence also shows that the chronic poor experience higher frequency o f income fluctuations, longer recovery periods, and short-range upward mobility which leaves them particularly vulnerable to recidivism. In the absence o f public policy, behavioral responses adopted to mitigate and cope with risk and vulnerability can play a crucial role in maintaining the poor inpoverty: "If the poor are more exposed to shocks than other groups, ifthey have fewer buffers to protect themselves against the realization o f these shocks, and if to achieve short term security they are forced to adopt low return livelihood strategies which further increase their vulnerability over time, thenrisk and Vulnerability will contribute to trapping individuals and households inpoverty." Sources: Yaqub (2002), Baulch (2002), Krishna (2003) and Barrientos, Hulme and Sheperd, (2001) Governments have a particular role to play in the provision of social protection services, in coordination with other public policies such as sectoral (health and education) investmentsand labor markets. A well designed set o f social protection policies is one component o f an effective poverty reduction strategy. Social protection policies can assist households in managing income risk, while education and health investmentsare criticalto raising incomes over the long term. Social protection policies also depend on a well-functioning labor market, and the design o f social protection programs must take into account labor market incentives or disincentives. Social insurance programs are funded via contributions and the level and nature o f these contributions have an important effect on incentives for formality in employment. Further, social assistance programs can create disincentives to work if benefit levels are set too high or if there i s no clear exit strategy inherent in the program design. Finally, assistance programs work best ifthey create strong positive links to employment. These issues will be discussed more fully inChapter 4. In sum, there is an important role for public provision of social protection where exposure to risk i s high and the potential income losses are large -and where private risk management tools fail, are insufficient, or produce negative externalities that further embed a 5 household in poverty. The principles o f social protection as laid out in this chapter can be summarized as follows: the role for government increases the larger the size o f the income risk, and the type o f government intervention that will be most successkl varies with the frequency o f the risk. Pooling risk across a large number o f households (insurance) i s most effective where risks occur infrequently but where the income costs are large, should the risk occur. Insurance becomes less effective the more frequent the risk, carving out a particular role for social assistance (income transfers) particularly for the poorest households who have limited assets for risk management (the chronic poor). Private risk management tools work best where the losses associated with risks are low and the role for government intervention increases with the size o f the potential income loss. Chapter 2 builds on this framework to identifymajor sources o f income risk to which households are exposed in Mexico, as well as the evidence on private risk management tools available for managing those risks. 6 2. HOUSEHOLDVULNERABILITY INMEXICO Individuals and households face a diverse set o f income risks. Social protection comprises a series o f risk management strategies including informal mechanisms (e.g., savings and remittances) and formal social protection instruments (e.g., social security and social assistance) to manage those risks. Drawing on available data and existing risk and vulnerability assessments, a mixed picture emerges. Mexico has made tremendous progress inreaching certain risk groups, particularly the young, poor and rural population. However, several key vulnerable groups can be identified for whom the frequency o f risk and severity o f loss compels a re- examination o f government policy. Among these are the elderly living in poverty and the low- income population which faces very high costs associated with health care. In addition to these idiosyncratic (individual) risks, Mexicans are also periodically exposed to aggregate shocks (such as economic shocks or natural disasters), yet the evidence suggests that the overall variability o f consumption in the face o f these risks i s low. Adjustment to macroeconomic shocks has largely taken place via falling wages, such that income losses as a result o f economic downturns have been diffused across the working population, as opposed to resulting in unemployment focused on certain key sectors, although some evidence points to possible changes in this adjustment mechanism in the future. Private risk management i s an important component o f household responses to the above risks, but its limits are also apparent. The chronic poor, for whom private riskmanagement tools are fewest, emerge as aparticularlyvulnerable group. A household's level of vulnerability, Le., the likelihood of experiencing a loss of welfare in the future, is determined by: (i)the characteristics of risks it faces, and (ii)the household's ability to manage risks via access to private or public mechanisms. Vulnerability can be definedas the probability o f experiencing a loss inthe future as measured by some benchmark o f welfare, such as income or consumption. In this report we define vulnerability as the probability o f falling into, or fwther into poverty. Inother words, we will focus on the probability o f becoming poor or poorer. The degree o f vulnerability depends on the characteristics o f risks (the size o f the potential loss) and on the households' ability to manage risk. Mexican households face a range of risks and employ a variety o f risk management strategies, both public (i.e., social protection policies such as insurance or assistance) and private (i.e., own savings, family remittances). When the combination o f private and public insurance and coping mechanisms i s effective, risk ceases to be a concern as shocks do not result in adverse welfare effects. Income shocks are idiosyncratic(individual) or covariant(economy-wide) and differ in terms of frequency, size of loss, and the mechanisms by which they are transmitted. Population-wide, the most common idiosyncratic risks affecting one's capacity to generate sufficient income are low-human capital development during childhood, health or other shocks to primary wage earners duringworking age, and poverty in old age due to an inability to work and insufficient retirement savings. Vulnerability to the different types o f idiosyncratic risks changes as an individual moves through the stages o f the lifecycle, as will be shown below. The major sources o f covariant risk are macroeconomic shocks that raise the risk o f income losses economy- wide and natural disasters or weather-related risks that affect the living standards and income prospects o f a large segment o f households in one geographic area at once. 7 IDIOSYNCRATIC RISKAND VULNERABILITY INMEXICO This analysis begins by assessing idiosyncratic risk among the Mexican population through the lens o f the lifecycle, consistent with the government o f Mexico's Contigo strategy.2 The analysis is based on data from the Encuesta Nacional de Ingreso y Gasto de 10s Hogares (ENIGH) from 1996 and 2002, from which leading risk indicators betweenthe rural and urban poor populations using the official poverty lines have been calculated, as outlined in Appendix tables 1-4.3 Differentsources o f exposure to idiosyncratic risk can be identified across three basic stages o f the lifecycle: children and youth, working age adults, and the elderly. Social protection policy responses also differ across these age groups. One major source o f idiosyncratic risk - health shocks- occurs across the lifecycle, and i s also discussed below. Infants, children andyouth are often among the most vulnerable in a population because o f the high concentration o f youth among the poor and because o f the long-lasting effects o f shocks to the young. The young are most vulnerable to risks that affect the development o f their human capital and thus future earnings potential, including malnutrition. little education andhealth shocks. The working agepopulation are most vulnerable to shocks to their capacity to generate income, either idiosyncratic (i.e., health shock that impinges upon the ability to work or unemployment) or aggregate (i.e., economic crises), that cannot be coped with and therefore translate to consumption losses. Policy responses here include both income generation (by supporting training, access to credit and financial services and channels for accumulating assets) and income security (through market insurance mechanisms, notably in health and unemployment). For the income security function, the role for publicly-provided social protection i s clear because these are market failure-related shocks with severe consequences, such as a serious illness, where risk pooling mechanisms are needed. The role for social protection i s not clear-cut for income generation, as will be discussed below. The elderly are most vulnerable to idiosyncratic shocks to income as they become unable to work or generate sufficient earnings due to lost productivity in old age. There i s a need to provide income security (mainly through pensions and health insurance). Health i s a risk that spans all age-groups and for which the poor show highvulnerability, as reflected in high levels o f out-of-pocket expenditure and catastrophic health shocks among the poor. Infants, children andyouth: Long-term shocks to human capital Considerableadvances have been made in Mexico over the past decade with respect to the human capital developmentof infants, children and youth. As outlined in World Bank (2004a), Mexico has now achieved close to universal primary education, with net primary schools enrollment among 6-11 year-olds rising from 88 percent in 1980 to 93.5 percent in 2000, and a See Chapter 3, Figure3.14 for further details on the Contigo Strategy. This work serves as (i)a complement to the vulnerability analysis carried out as part of the Mexico Poverty Programmatic work, based on the analysis of rural panel data from the Progresa evaluation from 1998-1999 and urban panel data from the ENEU applied in 1994 and in2001 (Skoufias 2004); and (ii) update of the an past World Bank social protection review (Arriagada and Hall, 2001), using the same methodology to provide a consistent view of the evolution of risks and social protection over time. 8 rapid expansion o f lower secondary enrollment. Similar gains have been made in health with considerable reductions in infant mortality (50.9 per 1,000 live births in 1980 to 20.5 in 2003), near universal vaccination coverage (97.7 percent) and decreases inmaternal mortality (from 15.6 percent in 1980 to 7.3 percent in2000). Table 2.1 Schoolattendance and Age-for-grade 1996 and 2002 Urban Rural Extreme poor All Extreme poor All SchoolAttendance Ages 6-11 1996 95.5 97.5 94.4 94.9 2002 94.8 98.0 96.1 97.0 Ages 12-14 1996 80.6 88.7 71.4 74.0 2002 88.5 92.0 83.8 86.5 Ages 15-17 2o02 1996 47.7 64.2 25.2 30.0 49.3 66.3 44.6 53.4 Agefor grade Ages 6-11 1996 1.04 1.01 1.10 1.os 2002 1.05 1.01 1.05 1.03 Ages 12-14 1996 1.14 1.07 1.30 1.24 2002 1.14 1.06 1.14 1.11 Source: WB staff calculations, ENIGH2002. Despite important progress in terms of meeting the needs of infants and youth, historical patterns of exclusion prevail beyond lower secondary school. Although smaller gains have also been achieved inupper secondary and tertiary, difference betweenincome groups and urban and rural areas remain considerable. As shown in table 2.1, attendance i s low and regressive at upper secondary levels, with marked urban and rural disparities. Upper secondary enrolment (among 15-17 year-olds) i s 49.3 percent among the urban poor, compared to a general rate o f 66.3 percent; in rural areas, enrolment drops to 44.6 percent among the poor and 53.4 percent overall. University enrolment i s quite limited and markedly regressive (see appendix). It i s important to note that although wban-rural gaps seem to persist, overall progress in rural areas was again matchedby near stagnation o f indicators inurban areas. This mighthave changed since 2002, inparticular inthe light o f the decision in2000 to extend Oportunidadesto urban areas. Sources of income risk during working age In assessing the major sources of income risk among the working-age poor, two distinct groups can be identified: a younger, more urbanized, skilled population and the older, less-skilled population concentrated more heavily inrural areas. The first group has higher levels o f human capital, havingbenefitedfrom the advances made inhealth and education, notably over the past 15-20 years. By contrast, there i s a lower-skilled but economically active population that constitutes 50 and 23 percent o f the poor rural and urban population respectively, with no education or incomplete primary. This population i s generally older, able bodied, but has limited productive capacity. Table 2.2 shows that the percentage o f low-skilled workers i s significantly smaller for the age group 25-40 than what i s found inthe age group 41-64. 9 Table 2.2 N o educationor incompleteprimary (low skills) by age group, 1996 and 2002 Urban Rural Extreme poor All Extreme poor All Ages 1996 36.7 16.7 60.8 49.4 25-40 2002 29.0 9.5 55.4 35.1 Ages 1996 69.9 41.6 90.2 85.4 41-64 2002 62.7 29.0 85.4 70.2 Source: WB staff calculations, ENIGH2002. Each of these populations requires distinct public policy interventions; the role for social protection is largest among the older working poor, soon to become elderly. The younger, more skilled population i s best positioned to benefit from interventions that fall largely outside the scope o f social protection, such as training (though international experience here also points to the limits o f training programs in generating higher earnings), more education and improved access to financial markets. The role for social protection becomes clearer when addressing short-term income risk arising from unemployment -however, unemployment i s not a major risk at present (see below). Among the older working age population, with very low skill level and declining productivity as they approach old age, there i s more scope for social protection intervention, particularly in the form o f a means and age-tested targeted minimum pensionbenefit as this population moves into retirement age. Unemployment is low and relatively stable in Mexico; however, the urban poor show somewhat higher unemploymentthan other segments of the population. Inthe period 1987-2003, with the important exception o f the Tequila Crisis in 1995, unemployment remained below 5 percent despite constant increases in labor participation (Montes, Santamaria and Bendini, 2004). Unemployment rates in Mexico are very low by regional standards, including when compared to most developed economies, such as Colombia and Chile, which experienced similar or stronger episodes o f growth than Mexico. Moreover, unemployment duration has also been low, with most skilled and unskilledworkers finding ajob within a six-month period. Table 2.3 Unemploymentand part-timeemployment, 1996 and 2002 Urban Rural Extreme poor All Extreme poor All Unemployment 1996 Male 4.6 3.2 1.5 1.5 Female 2.5 1.8 0.9 0.6 2002 Male 5.7 2.6 0.5 1.2 Female 1.7 1.9 0.1 0.4 Part-time,job* 1996 20.3 18.5 33.3 30.2 2002 27.7 17.9 42.1 33.5 * Share o f employed population working part-time. Source: WB staff calculations, ENIGH2002. 10 A growing share of the workingpoor are employedinpart timejobs. The percentage o f the population employed inpart-time jobs increased significantly more for the extreme poor in both rural and urban areas. In fact, in urban areas, the percentage o f the workforce in part time jobs declined slightly, while the percentage o f the extreme poor working in this status increased from 20.3 percent in 1996 to 27.7 percent. Inrural areas the share o f part-time employed had an overall increase, but it was significantly more pronounced among the extreme poor. This trend can be explained both by economic cycleslstructural change (market liberalization) and by the increased female participation in the workforce. An increase in part time jobs among the poor does not necessarily imply an adverse phenomenon. Itmay be that the poor are beginningto find part timejobs. However, the situation poses a challenge to the social security system, as part time employment rarely allows the worker to contribute sufficiently to social security systems. With respect to unemployment insurance, Mexico's historic wage flexibility has proven to be a relativelyefficientsafety net in past crises in Mexico. Inthe 1995 crisis, GDP per capita fell by 5 percent. The labor market adjusted through a 25 percent fall in wages but relatively little unemployment. This hadthe effect o f spreading losses relatively evenly across the population rather than concentrating the losses on a particular group (World Bank, 2004). Gonzalez-Anaya (1999) finds that this adjustment mechanism was observed in many Latin American countries during the nineties. However, this adjustment through prices (wages) a less through quantities (employment) might be changing, and with lower inflation rates unemployment might be again a more likely risk (see discussion in companion report Urban PoverQ in Mexico, WorldBank 2005). In addition, informal coping mechanisms that involve increased labor force participation of spouses effectively protect the incomes of households when faced with unemployment shocks to heads of households, and does not translate into increased school non- attendance or reduced educational performance (year-to-grade) among children. In a recent study o f the determinants o f household vulnerability in urban Mexico, Skoufias (2004) finds that it i s only during the Tequila crisis (i.e., an aggregate shock) that the loss o f a job by the household head i s associated with an increase in income risk. Estimates o f the determinants o f labor force participation show that increased labor market participation o f women (added worker effect) i s a common informal risk management strategy o f urban households. Combined with the estimated small impact o f unemployment shocks on income, this finding suggests that strategies that adjust the labor market participation o f spouses have been largely successkl coping mechanisms. Box 2.1: InformalHouseholdRiskManagement StrategiesinMexico Households engage in a number o f informal risk management strategies that need to be understood in order to craft effective formal, public social protection policies. A recent study suggests that families in urban Mexico employ both household self-insuring techniques as well as risk pooling techniques at the local level in the face of shocks as means for smoothing consumption. Spouse added worker strategiesare a substantial form o f informal self-insurance. There i s evidence o f the "added worker effect," where non-working wives enter the workforce when confronted with the loss o f a spouse's income. The research found substantial reductions in household work when a woman entered the labor market, suggesting that there may be hidden costs to this type o f riskmanagement such as less time with children and losses to social capital. Child added worker-strategies are limited. Putting children in the labor force can be seen as transferring future family income to the present and hence another self-insurance strategy with potentially highprivate and social costs. However, only certain groups ofpoor girls appear to drop out o f school when a father or mother loses his or her job. The largest effect on workforce 11 entry appears when a mother opens a micro-enterprise. As found in studies elsewhere, child labor inMexico appearsto bepro-cyclical. Extended families are probably efficient insurers and may offer protection not easily observed in the data. The sociological and anthropological literatures suggest that extended families provide possibilities o f pooling risks, while maintaining the correct "self-protection" incentives at the household level. Migration is often resorted to as a form of income diversification and remittances play an important role in risk management in Mexico. Incomes o f foreign workers are uncorrelated with their home economy and exchange rate devaluation increases the value o f transfers in peso terms. However, the cost o f this self-insurance strategy tends to be high in terms o f lower education attainment. Informallabor markets are not complete safety nets that offer readily available jobs. Most o f those enteringunemployment enter from the informal sector and the duration o f unemployment i s only 22 percent less than formal sector workers. This makes sense given that micro-enterprises everywhere have high failure rates. Informal capital markets in the form o f rotating credit associations, reciprocal loans, guelaguetza, or relations o f compadrazgo may offer consumption smoothing and risk pooling possibilities. Informal risk management strategies o f rural households depend on two additional activities: diversificationinto non-farm occupations and subsistence farming. Diversification of income sources: Diversifying income through non-farm economic activity has been an important way for Mexican rural households to concurrently increase income and mitigate risk. In reaction to the Tequila Crisis in 1994-95, rural households in Mexico increasedtheir involvement innon-farm occupations, particularly low return ones which are easier to access. Subsistence economy: Subsistence economy i s commonly understood as the agricultural production o f food crops carried out by farmers in one or several small plots o f land for self- consumption, using family labor. Subsistence farming i s rarely practiced alone but rather as a supplement to other production such as coffee or other cash crops, and it is not exclusive to poor farmers. The importance o f the subsistence economy has been declining rapidly but it still is the number one safety net inruralareas Source: "Income Risk, Household Coping Strategies and Income Security Policy in Mexico" 2001, World Bank report #22093-ME based on analysis o f the National Urban Unemployment Survey (ENEU), and "A Study o f Rural Poverty inMexico", World Bank Report (Forthcoming). The ability of households to effectively protect incomes against unemployment shocks via added-worker effect depends on: (i) demographic composition o f the household; the (ii)macroeconomic conditions (ability to mitigate the effects of employment shocks improves when the economy is growing, and falls during economic crises); and (iii) o f gender-based level segmentation o f labor markets. As referred to the above, Skoufias (2004) shows that increased labor market participation o f women (added worker effect) i s a common informal risk management strategy o f urban households. This implies that protection against the risk to urban incomes from unemployment depends on household composition such as the existence o f a non- working spouse. The inability o f households headed by single women to employ added-worker risk management strategies might indicate higher exposure to risk, however this seems to be a rushed conclusion. Maloney et al. (2003) find that during the 1994-95 tequila crisis, female- headed households did not do any worse than the median household in terms o f exposure to income risk, when corrected for income volatility in normal times. Unemployment rates in Mexico are presentlylow and inthis context private coping mechanisms function well. However, 12 the scope for the impact o f future crises to be absorbed via downwards wage adjustments (as opposed to rising unemployment) may be limited. This i s further discussed in section on economic downturns and crisis. TheElderly Old-age poverty is an important riskfacing individuals inMexico, and poverty rates are higher among the elderly than for the population as a whole. Recent data from international comparisons indicate that there i s a gap between the national poverty rate and the 65+ poverty rate, with poverty among the elderly being more pronounced. This gap in poverty rates i s not unique to Mexico among countries in the region; table 2.4 shows that a higher incidence o f poverty among the elderly than for any other age group i s found in six out o f the eight Latin American countries included. The regional comparison, however, also shows that the rate o f poverty among the elderly in Mexico i s high given the country's level o f development. This result is not surprising as Mexico has concentrated its poverty alleviation efforts and programs on children and young adults, a priority that is, from a social return point o f view, an understandable focus, due to Mexico's thin tax base and scarce public resources. Still, the vulnerability o f the elderly demands a closer look at the social protection mechanisms offered to this population. Table 2.4 Poverty Rates among the Elderly inLatin America, 1998 EntirePopulation 65 and older Bolivia 30.5% 47.5% Brazil 24.6% 18.5% Chile 20.8% 23.9% Colombia 24.0% 32.9% Costa Rica 21.7% 29.1% Guatemala 19.1% 27.1% ElSalvador 27.4% 38.0% Mexico 22.1% 37.6% Note: Comparison basedon total income that includes both monetary andnon-monetary income. Source: Gill, Packard andYermo (2004) Access to public pension schemes among the poor is virtually non-existent intoday's Mexico. Only 6.7 percent o f the elderly extreme poor in urban areas and less than 1 percent in rural areas receive a pension (Table 2.5). An immediate result o f lacking income security at old- age i s that a large share o f the elderly poor continue to work, particularly inrural areas. Coverage o f pensions systems and other forms o f income support among Mexico's elderly i s discussed in more detail in Chapter 3, yet it i s worth noting that a series o f social assistance programs and transfers reach the elderly poor, including Procampo, Oportunidades (in 600,000 beneficiary households, the support i s received by an elderly person), Liconsa, Programa para Adultos Mayores (covers 98 percent o f elderly in Mexico City), Acuerdo Nacional para el Campo and Programa Alimentario (see appendix). With the exception o f Mexico City's universal pension scheme, most o f these programs target the rural poor. However, despite these programs, poverty remains prevalent among the elderly inMexico. 13 Table 2.5 Pensioncoverageand labor force participation amongthe elderly, 1996 and 2002 (shares) Urban Rural Extreme poor All Extreme poor All Receives pension 1996 7.9 21.2 0.9 5.5 2002 6.7 22.1 0.8 5.3 W o r k s and does not receive a pension 2002 29.8 24.9 57.9 49.7 Source: WB staff calculations, ENIGH2002. Informal risk management strategies rely on extended families, private transfers (remittances), and accumulation of assets (savings or investments). It i s worth noting that the comparisons in Table 2.4 are based on total income poverty. Accumulated assets tendto increase with age, andthe elderlywould tendto bebetter offrelative to other groups, ifthese are included. However, the value o f assets held by the elderly depends critically on two institutions that are often lacking or malfunctioning in Latin America: (i) legal institutions that protect property rights, and (ii) financial institutions that allow households to convert illiquid assets into income for consumption in old age. In addition, it i s important to better understandthe existing informal risk management and private transfers (inter-generational) inplace, and to understandhow these are affected by socio-demographic changes. Private strategies to cope with income-riskin old-age may have adverse long-term effects. One example i s how the need to protect against old-age poverty affect land inheritance patterns in rural areas. The main perceived risks faced by old land-owning men, whose possibilities o f migration or participation inthe labor market are diminishedby age, are to be left alone, and to be unable to earn enough income to survive. Common responses to this situation observed in ejidos and comunidades are: (i) owners cling to their land resisting any pressures old to pass it on in life; (ii) least one o f the sons or daughters stay in the household to look after at parents and help tilling the land, often with the promise that they will inherit it; and (iii) parents keep the family bound such that transfers keep coming in from distant children. As such, the resistance o f Mexican small land holders o f advanced age to pass on their lands to the young generation i s part o f a broader survival cum risk management strategy identified in field studies (see A Study of Rural Poverty in Mexico, World Bank, 2005). However, it also results in reduced agricultural productivity, youth exodus, and demographic imbalances. Current demographicchanges indicate that addressingold-age poverty may become an increasing challenge, due to the aging of the Mexican population. A recent study o f demographic trends in Mexico found that the percentage o f the population above 60 years o f age rose from 6.6 in 1989 to 8.57 in 200L4 As shown in Figure 2.1, the increase in average age of household heads i s particularly pronouncedinrural areas. Higherlife expectancy i s increasing the number of elderlyrelative to young, which puts pressure on private riskmanagement strategies of extended families and inter-generational transfers and formal pension systems alike. An aging population, together with urbanization and changing socio-cultural norms, also affects household composition and the effectiveness o f the extended families as a safety net in old-age. Over the past decade the average number o f members per household has decreased, as single-member households or two-member households without children become more common. Given that women live longer than men, many o f the one-member households are composed o f women above 60 years o f age (Arriagada, 2001), while the growth in two-member families i s driven Ariza and de Oliveira (2004) 14 mainly by an increasing number o f older couples whose kids have left the household, the so- called empty-nesters (Ariza and Oliveira, 2004). From 1989 to 2002, the share o f single-member households in urban areas rose from 4.6 to 6.5 percent, while the share o f two-member households with no kids rose from 6.3 to 8.3 p e r ~ e n t . ~ Figure2.1. Age ofHouseholdHeads, 1992-2002 50 1 48 ~ 46 ~ 44 ~ 42 ~ 40 ~ I I I I 1992 1996 2000 2002 +Urban poorest 10% -M- Ruralpoorest 10% -A- National average Source: ENIGH2002. Idiosyncratic health shocks, a major source of risk across all agegroups Exposure to health shocks twinned with insufficient capacity to manage those shocks is a major risk associatedwith poverty at all stages of the lifecycle. The low-income population in Mexico faces very high costs associated with health care. More than 5 million Mexican citizens face catastrophic health expenditure, resulting in at least 2 million o f them falling into poverty.6 Inaddition, the distribution o f out-of-pocket expenditure i s regressive as the poor show much higher levels o f out-of-pocket expenditure than the rich. This translates into a higher frequency of catastrophic health shocks and suggests that available risk-pooling mechanisms do not reach the poor and/or do not offer efficient protection. When the survey Lo que Dicen 10s Pobres asked the poor in Mexico what they feared most in the coming 10 years, only unemployment and the death o f a family member ranked higher than i l l n e ~ s . ~ Out-of-pocket expenditure and catastrophichealth shocks. Estimates show that about 50 percent o f total health expenditure in Mexico i s private, almost all o f which i s out-of-pocket. The share of health expenditures absorbed privately byhouseholds is large based on international comparisons -for reference, the share of private expenditure o f total i s 25 percent in Colombia and 3 percent inthe UnitedKingdom. Further,these costs imply a particular financing burden on the poorest households. Based on Mexico's Ministry of Health definition o f catastrophic health shocks (30 percent o f disposable consumption, after spending on food, education and housing), the extreme poor are more likely to experience such shocks than the rest o f the population (Table 2.6). About 47 percent o f the poorest, urban and rural, experienced catastrophic shocks to income in2002, comparedto 26.3 percentofrural and21.7 ofurbannon-poor. 5Ibid. "Seguro Popular and the Reform of the Mexican Health System", World Bank, processed. Catastrophic health shock defined as more than 50 percent of household income net of basic nutritional consumption Secreteria de Desarollo Social, Lo que Dicen los Pobres, October 2003. 15 Table 2.6 Percentageofpopulationthat experiencedcatastrophichealthshocks, 2002 Rural Urban Extreme poor 46.8 47.2 Moderate poor 36.5 35.0 Non-poor 26.3 21.7 Source: Poverty in Mexico, World Bank (2004). Findings on catastrophic health shocks reflect inequities in healthexpenditure and healthinsurance coverage. There are significant inequalities in public health expenditure, and the poor are in general uninsured. Consequently, they frequently pay out-of-pocket for health services. Households in the first decile spend around 11 percent of income on health care while the richest spend less than 4 percent, pointing to severe problems o f efficiency and equity o f public health subsidies.* The link between public social insurance and vulnerability to catastrophic health shocks becomes even clearer when we look at households who fell into poverty as a result o f illness and related out-of-pocket expenditure. A recent study found that while only 9 percent o f insured households fall below the poverty line after catastrophic health care expenditures, 40 percent o f uninsuredhouseholds are impoverished when suffering a health shock.' In 2002, 73 percent o f households impoverished from health expenses were not insured. There i s also an urban-rural gap, as 60 percent o fruralhouseholds fall below the poverty line as a result o f catastrophic health care expenditures, while 17 percent o f urban households face the same problem. Box 2.2: Vulnerability to Health-RelatedRisk in Rural Mexico and the Impact of Oportunidades Oportunidades cash transfers combined with free access to health clinics, played an important role in protecting income from falling when household heads experience short- termillness. Inarecent empirical analysis ofvulnerability inruralMexico, Skoufias (2004) found that in communities not covered by Oportunidades, households with an ill household head experienced an average income growth rate that was between 20.4 percent and 21.7 percent lower than for households where the head did not get ill.This implies that households are unable to protect their income fiom short-term illness." In communities covered by Oportunidades, the growth rate o f income did not vary much as household heads experienced short spells o f illness. More serious illnesses (in terms o f duration in days and inability to leave the bed) turned out to have a smaller negative effect on income (not significantly different fiom zero). There were also no apparent differences between those covered and not covered by Oportunidades. Consumption appears to be well insured in terms of short-term illnesses even without the access to Oportunidades.Consumption appears to be insuredeven fiom more severe illness shocks, as estimates using serious illness shocks instead o f light illness shocks did not reverse the results.'' One possible explanation for this i s that long illness i s not really a shock but rather a permanent state. The study concludes that vulnerability could be more effectively tackled with an insurance-type of program that ensures that household welfare (consumption or income) does not fall below a socially acceptable norm. Skoufias also argues that the absence o f strong effects o f health shocks on consumption does not imply that households will not get any welfare World Bank 2003. "Universal Health Insurance Coverage in Mexico: In Search of Alternatives", Mimeo, Washington D C World Bank 9Ibid Defined as being confined to bed for a maximum of 3 days out of the last 30 days l1 Defined as being confined to bed for more than 6 days over last 60 days. 16 benefitfrom health insurance. Improvedhealth increases welfare directly, butthis effect cannot be measured directly. Source: Emmanuel Skoufias (2004) Variationsin healthstatus across incomes, populationgroups and regions may also be seen as a reflectionof unequal healthexpenditure and limitedaccess to social insurance programs. Mexico's many achievements in the health sector over the past decades (expanded access to basic services, increased spending) have led to significant improvements in the health status o f the population. Infant mortality has fallen sharply while immunization among children has become nearly universal. Yet, health status remains unequal, indicating continued inequities in access to social services. For instance, children of women living in extreme poverty are 2.5 times more likely to die before the age o f one than the children o f women who are not poor (World Bank, 2003). Children mortality rates from infectious diseases are over three times the national average in Chiapas, and 2.5 times in Oaxaca. The three southern states have the lowest rates of life expectancy and relatively high rates o f infant and child mortality. Finally, the indigenous population also score lower on health status indicators. For instance, mortality levels are higher in the states with a high concentration of indigenous population. As will be shown in Chapter 3, differences in health status across incomes, regions and populations reflect unequal access to and quality o fhealth and social services. DisabiIity-free Iife expectancy Life expectancy at birth 2000 at birth, 1998 Male Female Male Female National 71.5 76.5 61. I 66.9 Chiapas 69.6 75.0 56.7 62.3 Oaxaca 69.9 75.2 56.6 63.3 Guerrero I 69.9 75.31 55.2 65.71 Source: Mexico PlanNacional de Salud 2001-2006. The lower health status o f poor and indigenous households, and households in poorer regions, may be in part due to the lower share o f these populations that i s covered by health insurance. However, another factor could be the generally much lower access to various basic infrastructure services, such as pipe water, sanitation, and electricity, experienced by these populations. In managing idiosyncratic risks to income, significant progress has been made in Mexico among the very young, where major changes inthe acquisition of human capitalare occurring as a result o f public policy. Gaps persist in managing old-age poverty and catastrophic health shocks. There have been recent, major improvements in human capital development among the younger population, particularly for the rural population largely due to the notable expansion in basic health and primary education, as well as to the Oportunidades transfer program. At the same time, beyond lower secondary school historical patterns o f disparity are still reflected in gaps between urban and rural areas and poor and non-poor populations in terms o f access to lower and upper secondary school and university. Among the working age, unemployment i s low and the duration o f unemployment spells short in Mexico, while informal lz Ruwan Jayasuriya and Quentin Wodon, 2003 "Development Targets and Efficiency in Improving Education and Health Outcomes in Mexico's Southern States" in Development Strategy for the Mexican Southern States, Washington, D.C.: World Bank 17 coping mechanisms seem to protect households against idiosyncratic unemployment shocks. However, structural changes inthe way labor markets adjust might signal an increased frequency o f such shocks, signaling an emerging vulnerability which will be discussed in the next section. Old age poverty i s a significant problem in Mexico, which has the second highest poverty rate among the elderly in the region; the problem i s also likely to grow given demographic changes signaling the aging o f poor (rural) households in particular, and o f the Mexican population in general. Health shocks or illness are also an acute risk for the poor in Mexico, regardless o f age. Highvulnerability is reflectedinthe frequency of catastrophic health shocks andlinkedto limited access to health insurance. AGGREGATE SHOCKS, RISKAND VULNERABILITY While the above analysis focused on individual (idiosyncratic) risk, this section looks at the role o f two economy-wide, aggregate shocks in the vulnerability o f Mexican households: (i) economic downturns and crises, and (ii) disasters. natural Economic downturns and crises In terms of poverty reduction, the positive growth effects of the entire past decade were erased by the devastating effect of the 1994-95 Tequila Crisis. As shown in the report Poverg in Mexico, poverty rates in Mexico are only now returning to pre-crisis levels. As documented by Maloney, Cunningham and Bosch (2003), the Tequila Crisis that eruptedin late 1994 and the ensuing economic collapse came with dramatic social consequences. In 1995, prices rose 35 percent and output fell 6.2 percent. As wages remained fixed in nominal terms, real wages declined by 25-35 percent. Unemployment, while low by global standards, almost doubled from 3.9 percent to 7.4 percent.13Welfare outcomes were dismal: householdincomes declined by roughly 30 percent, extreme poverty more than doubled between 1994 and 1996 (going from 10.1 to 26.5 percent), while moderate poverty increased from 43 to 62 percent.14In the period since 1995, Mexico has not experienced a major economic crisis, and successful macroeconomic policies have stabilized fundamentals such as exchange and interest rates. However, increased trade liberalization and exposure to international competition are also occurring, signaling at once major trade opportunities as well as exposure to future macroeconomic shocks with differential effects across the Mexican population and economy. Shocks to income from economic crisis are transmittedthrough different channels and affect rural and urban populations differently.An economic crisis can affect household income through a range o f channels: (i) via labor markets in the form o f lower labor demand, increased unemployment, decreasedprobability o f finding new employment and a decrease inthe level o f earnings o f those still employed; (ii) changes inrelative prices; (iii) backs inthe level cut o f public transfers; and (iv) changes in the value o f and returns to assets (Skoufias 2003). As pointed out by Poverty in Mexico, macroeconomic stability and avoidance o f an economic crisis since 1995 have been a major positive development interms o f reducing the risks facing Mexican households. However, the severity o f shocks to income experienced in the 1994-95 crisis warrants a critical review o f vulnerability to this type o f shocks and the risk management mechanisms available. The urban population in Mexico is particularly vulnerable to macroeconomic instabilityand labor marketadjustments, as most of their incomederives from labor.At the Maloney, Cunningham and B o s h (2003). l4Montes, Santamaria and Bendini. 18 same time, risk coping strategies depend heavily on the labor market. Both rural and urban poverty increased dramatically as a result o f the macroeconomic crisis in 1994-95, however the effect was particularly strong in urban areas.15 For the urban population in Mexico, shocks to income from a domestic macroeconomic crisis are first and foremost transmitted via the labor markets inthe form o f increased unemployment and decreasedreal wages. The incomes o f urban poor depend heavily on the labor market as 92 percent o f the income o f the poorest 2 deciles derives from labor, compared to a rural average o f 81 percent and a regional urban average o f 74 percent. The urban poor also rely more on labor market strategies as informal coping mechanisms, via the adjustment o f the labor supply o f household members or added-worker strategies, thus illustrating the dual role of labor markets as both a source o f income risk and a means o f ex-post income protection. The dependence o f the rural poor on labor income i s lower, which impliesthat even if wages and employment are affected in similar ways as inurban areas, the income shocks transmitted via labor market developments are smaller. The rural poor are particularly exposed to potentialpolicy reversals and budget cuts affecting public transfers that occur during economic downturns; in addition, macroeconomic shocks place strain on the private risk management mechanisms that protect both urban and rural populations innormal times. Income transfers (both private and public) represent a greater share o f rural incomes. As a result, rural households are more prone to income shocks arising from changes in the macroeconomic environment that play out in terms of cuts in public expenditures on social programs (especially in countries with pro-cyclicality in public expenditures, such as Mexico) as well as changes inprivate transfer flows. Economic crises have been shown to depress the flow o f private transfers, particularly those that originate domestically. Glewwe and Hall (1998) found that in Peru, transfer networks assisting the poor in relatively stable periods do not protect them during a major shock, where the domestic transfer network shrinksbothinterms ofthe number ofrecipient households as well as the size ofinter-household transfer flows. This finding i s replicated inMexico, where the crisis affected rural households not only via lowered realwages, but also interms o f a reduction o fprivate transfers (Poverg in Rural Mexico). As in Peru, transfers did however provide assistance when originating from abroad, which points to the potential prominent role o f international remittances (and by extension, migration). During the Tequila crisis, transfers from friends and relatives outside o f Mexico (largely in the US), i.e., sheltered from the impact o f the crisis, increased, mitigatingthe shock to some extent. Despite the dismal effect of the 1994-95 crisis on welfare, poor households did not experience more variabiZi@ in income compared to normal times.16 Surprisingly, relative income variability did not increase for households generally considered to be among the more vulnerable groups, such as the self-employed and informal workers, single-parent families, young and old workers, with two important exceptions: households where the headhadno income at the beginning of the period became decidedly worse off during the crisis, while households headed by less-educated heads actually did better than during normal times.17 Finally, a comparison of rural and urban households shows that rural households experienced lesser shocks on average - probably due to the greater reliance o f rural households on self-consumption. These findings are likely explained by two factors: first, the fact that the bulk of the shock was absorbed via downward pressure in wages, which difksed the income effect o f the shock across a wide range o f households, and second, that households employed a range o f informal strategies for coping with shocks (see Box 2.1). l5World Bank Mexico Labor Market Study (forthcoming). World Bank, 2004. "Urban Poverty inMexico" 17WorldBank. Mexico Poverty Assessment PhaseI1- Poverty inUrban Mexico, forthcoming. 19 The most common private risk management strategies during the Tequila crisis were out-migration to the U S and increased income diversification. After 1995, migration to the United States increased alongside increased diversification o f rural incomes. Inboth rural and urban areas, however, poorer households may lack savings and other means to smooth consumption, resulting in higher consumption shocks than those households better able to mitigate (ex-ante) and cope (ex-post) with shocks to income. In Mexico, macroeconomic shocks have not typically resulted in higher unemployment; however, this may change and wage flexibility may cease to be a safety net. To date, macroeconomic crises in Mexico tend to have resulted in labor market adjustment via a lowering o f real wages as opposed to unemployment - the result o f fixed nominal wages combined with inflation. Mexico's historic wage flexibility has proven to be a relatively efficient safety net in crises in Mexico. Inthe 1995 Tequila crisis, GDP per capita fell by 5 percent. The labor market adjusted through a 25 percent fall in wages but relatively little unemployment. This hadthe effect o f spreading losses relatively evenly across the population rather than concentrating the losses on a particular group (Maloney, Cunningham, and Bosch, 2003). Recently, however, weaker labor market conditions combined with low total factor productivity suggest that the low unemployment that has characterized Mexico untilthe late 1990s will be more difficult to sustain in future crises (see Box 2.2). If this is the case, future macroeconomic shocks could be associated with more long-term unemployment, reducing the effectiveness o f private risk management strategies of both urban (added-worker effect), and rural (migration to urban areas to findjobs) households. These trends may indicate an increased need for formal mechanisms such as workfare programs or unemployment insurance that mitigate covariate employment shocks. Sectoral crises, such as the collapse in real agriculture prices experienced in the 1990s and the even more concentrated coffee crisis also represent an important source of risk to income. As shown in Rural Poverg in Mexico, the fall in real agricultural prices experienced by Mexican farmers in the 1990s resulted in a positive reaction through increased output and yields. Price incentives, however, remain low. This i s largely the result o f the openness o f the economy, which i s increasing under NAFTA, and the conditions of unequal competition faced by Mexican farmers in most crops vis-a-vis their northern neighbors, given their poorer endowments coupled with extensive agricultural subsidy programs in the United States and Canada. While Mexico also operates a comparatively small agricultural subsidy program, Procampo, inpractice it i s functioning more as a rural safety net (with moderately pro- poor targeting, though a sizeable portion o f benefits also flows to farmers in higher income deciles). Impact evaluations indicate positive consumption effects on small farmers as well as some income multiplier effects on medium and large-scale farmers; however, it has been less successful in achieving its stated goals o f supporting domestic producers o f basic staples in adjusting to international competition under NAFTA, and helping farmers switch to more competitive crops (Rural Poverty in Mexico); it i s also set to expire in 2008 (more on Procampo inBox 3.3). Box 2.3: Has risk and vulnerability to economic shocks changed since the Tequila crises? Numerous factors, including greater openness to trade, the fall in inflation, the slow growth in labor productivity and the weakening o f labor unions may have led to more frequent labor market adjustments through unemploymentthan what was seen during the 1994-95 crisis. Greater openness to trade and increased competition. One concern that has been raised with regard to trade liberalization is that increased competition leads to greater product 20 demand elasticity which again leads to greater own wage elasticity. As a result, shocks are more directly translated to the labor market and workers face higher aggregate risk.LessonsJi-om Nafta found little indication o f higher unemployment or increased volatility o f the labor market in Mexico following NAFTA. Looking at individual income risk invarious manufacturing sectors in Mexico (Krebs, Krishna and Maloney, forthcoming), however, find that tariff reform has a significant short run effect on income risk.They also findthat tariff reductions increase the cost o f recessions substantially.18 Lowered inflation. A general increase in the propensity to be fired may also have emerged because o f the success in fighting inflation. Price stability makes it more difficult to respond to aggregate shocks by reducing real wages by holding nominal wages fixed. The ability o f wages to adapt to changes in economic activity and absorb shocks in times o f recession seems to have declined since the late 1990s when inflation hit single-digits. The structural relationship (Okun) between output and wages seem to have suffereda break around 1999, as every decline in output has since translated into smaller declines inwages. While Mexico experienced a decline in its growth rate between 2000 and 2001, mean real remunerations kept increasing through 2002. Labor productivity. Despite slow labor productivity growth, Mexico was able to remain competitive in the U S market until 1998 thanks to its low dollar-denominated wages. However, since the late 199Os,real wages have been steadily rising in spite o f declining GDP and low labor productivity. Duringthe last recession (2001-02),wages and unemployment increased in spite o f a large GDP fall, disproportionately affecting export-oriented firms. If wages continue to rise and factor productivity is not improved, it will become increasingly difficult for Mexican exporters to remain competitive abroad, which can afflict the labor market as a whole, and particularly the poor, who have greatly benefitedfrom export growth after 1996. Weaker labor unions.Historically, labor codes negotiations inMexico have emphasized employment stability. This i s reflected both in the legal framework, which heavily protects employment and mandates generous severance pay and in the importance given to corporatist bargaining mechanisms such as the P A C T 0 However, the process o f privatizing and deregulating state-owned enterprises, and in general exposing firms to greater external competition, may have altered the bargainingpower o f workers. At the same time, various agreements and policies made by the government during Mexico's inflationary period of the late 1980s and early 1990s, urged firms to exercise restraint inraisingwages and prices. With increased competition, lost wage flexibility, continued low TFP growth and weaker bargaining power of workers, the low unemployment that characterized Mexico untilthe late 1990s will be more difficult to sustain. Alternatively, it may trigger an increase in informality, which could also hurt opportunities for the poor. It is important to keep in mindthat macroeconomic stability (lowered inflation) and less rigid labor markets are both key in reaping the full benefits of increased trade liberalization and economic integration. Given the disastrous impact o f the Tequila crises, macroeconomic stability i s probably the single most important risk and vulnerability reducingpolicy o f the post-crisis period. Source: Lederman, Maloney and Serven (World Bank 2004), Montes, Santamaria and Bendini (World Bank 2004), and Krebs, Krishna and Maloney (Forthcoming). Natural disasters and weather related risks Public and private risk management strategies seem largely effective in managing natural disasters and weather-related shocks. In a recent study of vulnerability in Mexico, Notice that the study also finds that tariff reductions lead to decreased income risk during economic booms. Inaddition, the welfare analysis focuses exclusively on the linkbetween trade reform and individual income risk, and does not account for expected positive effects from economic growth and a more efficient resource allocation. 21 Skoufias (2004) finds that for all rural households, covariate risks, mostly related to weather shocks, significantly affect household incomes and consumption, although households carry out income smoothing practices that partiallyprotect their incomes from such risks. Systemic shocks, however, are shown to be o f secondary importance with respect to idiosyncratic ones. Also, panel data shows that systemic shocks related to weather and other natural disasters can have very different impacts on households, and that shocks affecting income do not necessarily lead to consumption changes. Successful practices o f consumption smoothing make consumption more protected than income. Most agricultural insurance in Mexico i s oriented to middle and large commercial farmers, and as such, crop and livestock insurance i s not relevant for very poor farmers. Among production-related shocks analyzed in Skoufias (2004), plagas (pest) is the one with a more significant impact on income. Households who experienced a problem o f pest and diseases had an average income growth rate 16 to 17 percent below other households. Again, the impact on consumption was much smaller, which was reduced by only 3 percent. Box 2.4: Fondo de Desastres Naturales (FONDEN) and other risk management mechanisms The Mexican government allocates budgetary funds for disaster relief and reconstruction efforts by placing them in FONDEN, thus providing for the repair o f uninsured infrastructure, immediate assistance to restore the productivity o f the farm population, and relief to low-income victims o f disasters via: (i)support to rebuild homes; (ii) compensation for crop and livestock losses; (iii) temporary income and employment support over and above that traditionally offered byPET; and (iv) reconstruction oflocal infrastructure. The World Bank study Poverty in Rural Mexico assesses FONDEN as a very useful instrumentto absorb some of the income impact of large covariate shocks, but it compensates for part only o f the losses and depends on a number o f procedures and discretionary actions such as the declaration of emergency that limits its impact. More recently, FONDENhas started to adopt objective rules for declaring catastrophic events. This removes an ad hoc dimension in the declaration o f catastrophes and reduces the political interference inFONDEN'soperations. The intention is not to compete with private insurance, and operating in parallel to FONDEN is a group of mutual insurance funds amongst farmer organizations. These farmer organizations are called fondos de aseguramiento, formed to provide mutual crop insurance to their members. Analysis of historic reinsurance payouts reveals that weather events, similar to those covered under FONDEN, are the primary source o f systemic payouts by thefondos. Source: Poverty in Rural Mexico, World Bank (2004) and "Agricultural Markets and Risks : Management o fthe latter not the former", Varangis, Larson and Anderson (2002). As seen inurban areas, ruralhouseholds also resort to a variety of riskmitigating and coping strategies as a way of self-insurance. These include the accumulation o f assets, income diversification, sending women and children to work, withdrawing children from school, input and crop choices, precautionary savings, migration, marriage, income transfers among friends and relatives, and other informal risk sharing arrangements such as share cropping or input sharing (Skoufias, 2004). The subsistence economy, Le., the production for self-consumption o f food crops and small animals in small plots with family labor together with the access to collectively owned natural resources, i s a very important safety net in rural areas. Much o f the food security o f small peasant farmers i s linked to this type o f production (Box 2.1 discusses private risk management strategies, including subsistence farming in more detail). Poverty in Rural Mexico found that poorer households (two bottom quartiles) experienced less shocks than richer ones, 22 probably reflecting a higher risk-aversion leading to less risky investments and behavior among the poor. Low yields due to production-related shocks were significantly smaller in the bottom two quartiles, which the report attributes to the planting o f lower risk crops and the use o f low risktechnology bythis group. Yet, natural disasters and other weather related phenomena are an important source of income riskprimarily to rural populations,and a key determinant of inefficiencies incrop production.The highincidence ofrealized shocks among farmers inruralMexico can be seen as a reflection o f the risk o f agricultural production. Using data from a sample o f 666 crop farmers o f different parts o f Mexico in 2002, the Poverty in Rural Mexico report shows that weather-related shocks are very frequent among farmers -as many as 44 percent reported having experienced one in that year. The most common shock reported was drought followed by excessive rain or hurricane. Finally, efficiency analysis showed that shocks o f natural origin (rains, frosts, droughts, hail, pests, etc.) are a major determinant o f inefficiency in crop production. While coping with natural disasters once they occur falls clearly within the domain of social protectionpolicy, rural experts place a relatively heavy focus on policies to reduce and mitigate the effects of natural disaster and weather-related risks,which fall beyond the scope of social protection.PoverQ in Rural Mexico emphasizes the importance o f promoting the use o f technologies less vulnerable to prevalent risks in particular regions through appropriate research and extension. This includes promotion o f crop varieties more resistant to water stress or to pests, or crops maturing at a good time according to local weather patterns. Pest control and sanitary measures ingeneral constitute additional measures to reduce natural shocks. Access to financial and insurance services could also play a crucial role in self- insurance strategies toward natural disasters and weather-related shocks. Again -this falls beyond the scope of social protection. Formal crop insurance is not seen as particularly useful for the rural poor, whose main income i s not from independent farming and for whom the insurance i s too expensive. Parametric insurance systems linked to weather parameters offer an interesting alternative. Finally, a well-developed financial system and rural financial services could play a crucial role in self-insurance and risk management strategies among the rural poor, mostly by facilitating savings andpersonal loans. Box 2.5: Remittances as a source of private risk management for Mexican households The increasingimportance and magnitude of remittances Remittances from Mexican workers abroad reached a record 13.3 billion dollars in 2003. It exceeded foreign direct investment as a source o f foreign income and amounted to some 2 percent o fthe country's GDP in2003. Although migration is not a recent phenomenon, much o f the capital flows from remittances seem to be. In general, about half o f the recipients included in a study o f remittances in Mexico said they hadreceivedremittances for 3 years or less.'' Remittances played a role in the recent growth of rural incomes, especially poor incomes l9Pew Hispanic Center & MIF. "Receptores de Remesas en Mexico", Mexico City: IADB. 23 The reductioninextreme poverty inthe year 2000-02 was associated with rapidwage increase and for rural areas sine gains fiom remittances and transfers. The highestproportion o fhouseholds receiving remittances is found among the poorest 20 percent o f the population. In 2002, 11.2 percent o f households inthe poorest quintile received remittances compared to a national average o f 1.2. For the poorest 20 percent o f rural households this number rises to one-fifth. National Urban Rural 1st Quintile 11.2 4.0 19.5 2nd Quintile 2.5 0.8 6.8 3rd Quintile 1.o 0.7 4.2 4th Quintile 0.9 0.4 1.7 5th Quintile 0.1 0.0 1.o Total 1.2 0.5 3.5 Note: Quintiles calculated using current income per capita net o f transfers (remittances, Oportunidades, Procampo and other). Source: World Bank (2004) - Bank staff calculations usingENIGH. While `coverage' of remittances is progressiveand pro-rural, it is also limited. Only a minority of the extreme poor have access to remittances 86.5 percent o f the poorest households do not receive remittances. Usingpost-transfer income, the share o f households receiving remittances is lower for the poorest 20 percent o f the population than for any other income group. The different conclusions arrived at from using post and pre-transfer income simply reflect the fact that transfers (including remittances) are an important driver o f income growth among the households fortunate enough to have access to these transfers (see table below). Percentage of households receivingremittances by income deciles (post transfers), 2002 I II III IV V VI VII VIII IX x 3.2 2.1 11.2 7.6 3.0 8.1 6.3 5.9 4.9 4.7 Deciles calculated usingtotal current income per capita (post-transfer). Source: ENIGH2002. Sources: World Bank Poverty in Mexico (2004) and Pew Hispanic Center & MIF, 2003. "Receptores de Remesas en Mexico", Mexico Citv: IADB CHRONICPOVERTY AND VULNERABILITY Chronic poverty is the inter-generational transmission of extreme poverty, often accompanied by exclusion. Its distinguishing feature is extended duration, as distinct from those who move temporarily into (and out of) poverty in the face o f risk. Evidence suggests that for the chronic poor, transitory fluctuations are more frequent. Inaddition, poverty exits are short- range and the levels o f recidivismhigh (Yaqub, 2002). As shown in Box 2.6, chronic poverty in Mexico corresponds to the 20 percent o f the population that lives in extreme poverty. 24 Box 2.6: Evidence of Chronic Poverty in Mexico, I s there a `Nucleo Dum' of Poverty? International evidence suggests that chronic poverty i s characterizedby: (i) economic low mobility but often more frequent fluctuations o f income, (ii) low endowments o f physical and human capital, and (iii)low return on assets which i s often linked to social exclusion or discrimination. This describes well the situation o f the poorest segments o f the population in Mexico. The characteristics o f the extreme poor in Mexico includes: patterns o f lower schooling levels, higher dependency ratios, and variable job quality coupled with some evidence o f labor market discrimination are apparent. Extreme poverty is more concentrated among rural and indigenous populations, for whom poverty rates have been relatively stagnant over time and who face important aspects o f social exclusion and discriminationinthe labor market and elsewhere. A recent analysis of household in Mexico over the period 1984-2000, finds that there is very little mobility o f households across the income distribution, and that this lack o f mobility is particularly pronounced for households in the bottom quintile (poorest 20 percent) o f the distribution. Initial ranking inthe income distribution is the single most powerful factor associated with mobility.20Twinned with the fact that extreme poverty in Mexico has hovered around 20 percent o f the population over this same period, with little downward movement in the rate, these findings suggest the existence of a `nucleoduro'o f chronic poverty concentratedinthe bottom 20 percent o fthe income distribution. Are things improving? Substantial progress has been made in addressing human capital development issues among the young, and particularly among the extreme poor, notably through the introduction of the Progresa/Oportunidadesprogram. However, thejury is still out regarding these relatively new programs' long-term impact on the intergenerational transmission o f poverty and ultimately, reduced income inequality. Furthermore, major federal social protection programs, notably Oportunidades,do not reach all o f the extreme poor, nor do private remittances reach the extreme poor to a significant degree. Recent initiatives have been introduced that are aimed at closing this gap, such as the Programa Alimentario (providing cash or in-kind food benefits), though it i s too soon to be able to assess their impact. In sum, while Oportunidadesrepresents a major advance and new, promising initiatives are being introduced, Mexico does not yet have in place a complete strategy for adressing chronic poverty, particularly with respect to income vuherabilitv. The chronic poor are those households least equipped to manage income risk as a result o f their low asset base and limited access to private and public risk management tools. They are therefore among the most vulnerable to income shocks, whether idiosyncratic or aggregate. They are also the most likely to engage in negative risk coping strategies, which ultimately serve to perpetuate their poverty. Long-term costs o f informal practices o f risk coping (and ultimately survival) may cause permanently lowered incomes and chronic poverty. Damaging behavioral responses adopted by the poor in order to mitigate (ex-ante) and cope (ex- post) with risk include among others: adverse incorporation under the wings o f a patron, safer but lower production techniques, reduced investmentsin physical and human capital. The immediate cost o f planting a traditional, safer crop, rather than a more profitable but riskier one, i s obvious and constrain upward mobility. Other costs, however, are less obvious as some risk management strategies reinforce not only poverty but also the structural determinants thereof. For instance, the use o f informal insurance networks offered by patrons promotes clientelism and patronage both in 2ODuran Fernandez, Roberto. 2003. "La movilidad a1 interior de la distribucion del ingreso en Mexico durante el periodo 1984-2000",Gaceta de Economia, 9, Numero 17. 25 labor markets as well as in other spheres o f society.21This, again, can affect both the capacity o f the poor to invest and save, as well as the accountability o f public service provision. Formal risk pooling mechanisms offer better, less damaging alternatives to these practices. While extreme poverty rates have been reduced significantly in terms of national average, certain sub-groups, particularly rural and indigenous households, show little progress. Preliminary evidence indicates that the rate o f change o f extreme poverty rates differs greatly among different sub-groups o f the Mexican population. A recent study o f indigenous poverty in Mexico, for example, finds striking differences inthe pace with which indigenous and rural households move out o f poverty vis-a-vis the rest o f the Mexican population (World Bank, forthcoming). As shown in Figure 2.2, the national extreme poverty rate declined by 9 percent over a 10-year period (1992-2002), and urban poverty rates fell by 15 percent. At the same time, there was very little change in rural and indigenous poverty rates (3 percent), such that the vast majority o f indigenous and rural households remained poor. The incidence o f poverty i s also much higher inboth rural and indigenous municipalities than non-indigenous municipalities. And while there i s a significant overlap between indigenous and rural municipalities, indigenous municipalities also have a significantly higher incidence of extreme poverty than rural municipalities on average. This difference i s important because it suggests that indigenous people are poor for different or additional reasons than the simple fact that they mainly live in the rural sector. Figure2.2. Rate of Change inExtreme Poverty RatesAcross Different Population Groups inMexico: 1992-2002 100, 20 c 60 ~ 1 9 9 2 ~ 1 4 0 - - 3% 2002 n - 9% 20 - 20% ~ - 15% 0 , Indigenous Non-indigenous Rural Urban Total Source: World Bank. 2004. Indigenous People, Poverty and Human Development in Latin America, 1994- 2004 (forthcoming). Based on ENIGHsurvey data, 1992, 2002. Extreme deprivation (living on less than a dollar a day), has been practically eliminated among the non-indigenous, yet affects 30 percent of the indigenous population. Another way to look at this same question i s to note that by international standards, Mexico has all but eliminated extreme deprivation for the non-indigenous population (defined as households living on less than US$1 per day, usingvalues adjusted for Purchasing Power Parity (PPP). Over the past decade (1992-2002), the non-indigenous extreme poverty rate according to this measure dropped from 6.1 to 1.3 percent. However in 2002, 30 percent o f the population in indigenous municipalities was still among the extreme poor according to this international measure. In sum, indigenous people make upjust 12 percent o f Mexico's national population, yet are vastly over- represented among the extreme poor, and may comprise an important component o f chronic poor 21See J.J. Morduch, 1999. "Issues on risk and poverty", presented at the Stiglitz Summer Research Workshopon Poverty, the World Bank, Washington DC, July6-8,1999. 26 households, for which the chances o f moving out o f poverty are vastly lower than for the rest of the population. Oportunidades provides major benefits to the extreme poor; however, these households still need access to risk management tools. Oportunidades provides access to health services and income from cash transfers, which provide major poverty reduction benefits for poor households. The program was not design as a risk management tool, and therefore not surprisingly, Skoufias (2004) finds that inthe villages that receive Oportunidades,recipients still resort to private coping mechanisms that inthe long runpreventthe accumulation o f assets and an exit out o f poverty. For certain types o f shocks, households in Oportunidadesvillages are in fact significantly more likely to sell animals or other possessions than households in control villages (non-recipients). SUMMARY OFFINDINGSONRISKAND VULNERABILITY INMEXICO Mexican households have been more vulnerable to idiosyncratic than to aggregate risk. Two immediate priority areas for government intervention in addressing exposure to idiosyncratic risk emerge: insurance against catastrophic health expenditures and against the risk of poverty in old age. In addition, unemployment risk maybe more important during aggregate shocks in the future. For a large proportion o f the population, in particular among the poor, vulnerability to the first two risks i s high. This i s reflected in the gap between poverty rates for the population at large and the elderly, and in the high costs o f health care for the poor, the frequency of catastrophic health shocks, and the large proportion o fpeople who may fall into poverty as a consequence o f catastrophic health shocks. At the same time, as will be discussed inthe next chapter, public social protection programs inthese areas are still limited and incidence as well as expenditure patterns are regressive, notably in the provision o f old-age income insurance. While the poor employ a wide range o f private strategies (extended families, intra-generational transfers, among others) to insure against health and old-age risks, these are limited, maybe hampered in the future due to demographic changes and at times carry adverse long-term effects. While these are the risks that are most immediately apparent from current patterns o f vulnerability in Mexico, an emerging risk may be that o f unemployment and catastrophic income loss among poor segments o f the population in the face o f future aggregate shocks; while the risk o f unemployment has been low inthe past, labor market conditions may be changing such that shocks may no longer be easily absorbed via wage adjustments but through changes in employment. While private strategies to cope with unemployment are largely effective ingood economic times, the opportunity to do so is reduced during aggregate economic shocks, indicating the need for mechanisms to deal with covariate employment shocks. Finally, the chronic poor are particularly exposedto income risk due to their low asset base and limitedaccess to both public and private risk management tools, perhaps explaining the `stickiness' o f Mexico's extreme poverty rate over time, and highlighting a particularly vulnerable group as an important target for public policy intervention. 27 3. MEXICO'S SOCIAL PROTECTIONSYSTEM: OVERVIEWAND ASSESSMENT Mexico's social protection system has shown progress in terms of long-term financial sustainability and on improving welfare among thepoor. However, the system is stillfragmented withproblems of equity and eflciency. As a result, a large share of thepopulation -particularly low-income and those employed in the informal sector- experience low or non-existentprotection in the face of income risks. The characteristics of the Mexican social protection system are broadly similar to thoseprevailing in the rest of Latin America, but Mexico also spends less than countries at similar income levels on the social sectors in general. However, the share of public social expenditures as apercentage of public expenditures is not low by international standards reflecting a thin tax base and low tax revenue. A fiscal reform is needed to continue improving and expanding poverty reduction programs as well as increasing the coverage of the social protection system. The social security systemfor private workers was reformed in 1997, leading to a morefinancially sound system. Coverage of the system has increased, although necessary further increases are still a challenge for the country. The system is regressive in terms of coverage reflecting the inequality of Mexican income distribution. Thesocial security systemfor public sector workers however is highly regressive and shows huge problems of financial sustainability which generate a hugefiscal burden to current and future taxpayers. Regarding social assistance, while absorbing a relatively small share of the social protection budget22(7 percent), the Oportunidades program (originally Progresa) represents an innovation of great value to poverty reduction in Mexico. At the same time, due to its design, itfaces limits as a risk management tool for the poor. Other major initiatives target important risk groups, such as Procampo (a program of agricultural subsidies designed to aid rural farming households in competing under NAFTA), and Seguro Popular (a new health risk management system offering subsidized health insurance to poor households). However these programs face challenges of implementation, financing and institutional coordination as Mexico faces the challenge of providing adequate coverage to the majority of itspopulation. AN OVERVIEW OF THE MEXICAN SOCIAL PROTECTIONSYSTEM Mexico's well-established social protectionsystem is comprised of a formal sector based social insurance component providing both private and public sector workers with health benefits and retirement pensions, a nationalhealth system for the uninsured, and a wide array of social assistanceprograms. Elements ofthe social protection systemhave shown progress in improving coverage and welfare among the poor. However, important challenges remain to increase the coverage and efficiency o f the system. The government's commitment to social protection dates back to 1943 and the creation of the formal social insurance program for private sector workers ran by the Instituto Mexican0 del Seguro Social (IMSS), and a nationalhealth system, the Sistema de 22Social protection defined as social security expenditure and poverty-targeted social programs. Social security includes total expenditures in IMSS and ISSSTE, which includes workers', employers' and the federal contributions. 28 Salud. Originally, IMSS was a partially-knded defined benefits system but in reality operated from the start as a pay-as-you-go (PAYG) system. In 1997, the system was radically reformed and system o f private administration o f individual pension accounts was introduced. Public sector workers are covered by the Instituto de Seguridady Sewicios Sociales de 10s Trabajadores del Estado (ISSSTE), which manages federal employees' accounts. Inaddition, the state oil company (PEMEX), the armed forces as well as other public enterprises and state governments have developed individual insurance systems. IMSS remains the most important in terms o f the number of affiliates with 14 million rights holders in the formal private sector, followed by ISSSTEwhich covers 2.3 million government workers.23The figure below provides an aggregate picture, showing coverage by deciles lumping all types of workers. As observed, despite the progress, the overall coverage o f major social security programs across income groups i s still regressive. Similar patterns are observed for a host o f programs: housing credit, paid vacations, pensions, childcare, life insurance, to name a few (Tinajero 2003). Social insurance mechanisms for the population excluded from formal sector benefits, whose workers and their families make up the majority of the population, and who tend to be poor, are weak. This segment of the population i s served by the nationalhealth system and an array o f social assistanceprograms. Figure3.1. Social security (IMSS, ISSSTE, PEMEX) coverage across income deciles, 2002 60% L 50% t 40% L 30% t 20% L 10% L 0% t I II 111 IV v VI VI1 Vlll IX X Source: ENIGH2002. Social assistance policy has evolved over time, closely tied to national economic development objectives, with explicit poverty-targeted programs being introduced only after the 1980s debt crisis. The largest social assistance program, both in terms of coverage and expenditures, is Oportunidades which provides cash transfers to poor households conditional on children's school attendance and basic health care, has proven to be a successful transfer mechanism tied to human capital accumulation objectives. Oportunidadesi s the most important social assistanceprogram inMexico, with an annual budget in 2002 of about 17 billion pesos (approximately 1.6 percent of total spending) and coverage of over 4 million households, equivalent to about 20 percent o f the Mexican population. The program has been instrumental in Mexico's tremendous progress in education coverage, particularly at the primary level. The most direct impact o f Oportunidades, which has been 23Mexico Public ExpenditureReview, World Bank. 2004. 29 implementedprimarily inrural areas to date (but is currently being expanded to urban areas) and targets the extreme poor, has been in narrowing the gap in access to education across poverty levels and between rural and urban areas. The figure below shows Oportunidades coverage as highly progressive, covering almost 60 percent of households in the poorest decile in 2002. Administrative data suggest that the actual number o f beneficiaries o f Oportunidadeswas higher than that indicated in survey data. Estimates that attempt to correct the survey data based on Oportunidades' own administrative data suggest that 20% o f the population in the decile 1 and 40% of the population in decile 2 were uncovered. 24 Figure3.2. OportunidadesCoverageAcross IncomeGroups, 2002 60% 7 I II 111 IV v VI VI1 Vlll IX X Source: ENIGH2002. Despite the advances of Oportunidades, a significant portion of the extreme poor may remain uncovered. Major social assistance interventions such as Oportunidades are extremely well-targeted to the poor, yet still leave a significant portion o f the extreme poor (households inthe two poorest deciles) uncovered. According to the ENIGHhousehold survey, in 2002, almost 40 percent o f the population inthe poorest decile and 46 percent o f those in decile 2 - over 8 million Mexicans-were not affiliated to any public social insurance program, nor did they receive benefits from the major social assistanceprogram, Oportunidades(Table 3.1). These may be taken as upper bound estimates for the number o f uncovered, given that as noted above, administrative data for the program suggest even higher coverage rates for Oportunidadesamong the bottomtwo deciles 24Revised estimates of Oportunidades coverage as calculated by SEDESOL and provided to World Bank staff, 4.22.05. 30 Table 3.1. Householdcoverageof major socialsecurity programs and Oportunidadesin the poorest2 deciles, 2002 Programs Decile 1 Decile 2 Covered by IMSS, ISSSTE, PEMEX, etc. 3.94% 18.21% 57.11 Receives Oportunidades % 35.20% 38.85 Neither of the Above % 46.17% Total 100% 100% Neither o f the Above -but covered by another pension plan 0.06% 0.15% Neither o f the Above -but receives medical benefits through work 0.05% 0.27% Source: ENIGH2002. Own calculations. T o what degree do the extreme poor - who have limited access to socialinsurance- have access to some form of social assistance?Mexico's national household surveys only allow a limited answer to this question, since incidence figures are not collected for all social assistance programs. Further, some programs, such as Seguro Popular, have been recently implemented such that incidence data are not yet available. However, one can begin to pose the question by crossing data on access to social insurance with data on coverage o f Mexico's single largest social assistance program, Oportunidades,for which incidence data are collected in the national household survey. Interestingly,upper bound estimates indicate that up to one third o f households inthe poorest decile, and almost one half of households inthe next decile, may not covered by either social insurance or Oportunidades (Table 3.1). Further,very few o f these households have access to another pension plan, or report receiving medical benefits through work (less than 1 percent in all cases). Further, very few o f them report having access to the second largest rural assistanceprogram, Procampo (see Box 3.1). Box 3.1: Who are the 8 millionextreme poor uncoveredby Oportunidades? Over eight million extreme poor people (close to 40 percent o f the extreme poor) in Mexico live inhouseholds that do not have access to the country's formal social insurance system, nor do they receive benefits from the flagship poverty reduction program, Oportunidades.What are the definingcharacteristics o f this population? First, a majority (63 percent) live in rural areas. Given that the Oportunidades program is focused on rural areas, why are these households excluded from coverage? Most probably these are households that live in villages so small and isolated that Oportunidades does not operate there, given that it requires that a school and health post be in close proximity to beneficiary communities. These households also appear to be out o f reach o f the other major rural program, Procampo - less than 2 percent report receiving benefits from this program (see Box 3.3, below). They also have slightly older household heads, and more elderly inhabitants; one in five household members are over age 65, compared to one in eight among covered households. These findings suggest that isolated rural households, particularly those with elderly members, may be a particularly vulnerable group currently uncovered by the country's maior social protection initiatives. 31 SOCIALPROTECTIONINMEXICO -OVERVIEW OF EXPENDITURE TRENDS Mexico's expenditures in social programs, both social assistance programs aimed at reducing poverty as well as in social insurance are low given the level o f development o f the country. However, the share o f social expenditures as a proportion o f public expenditures i s comparable to countries such as Argentina, Chile, and Brazil because total public expenditures are low. This reflects a major challenge for the country, as public resources needed to continue improving and expanding poverty reduction programs and increase social insurance coverage, in particular among vulnerable groups, i s currently too limited. This implies that a fiscal reform which can expandthis limitedfiscal space i s critical. Mexico's SocialSpendingRelative toInternational Comparisons According to international comparisons developed by ECLAC, per capita social spending in Chile and Brazil is more than twice as high as Mexico's, while Argentina and Uruguay spend more than three times as much.25 This is despite steep cumulative declines in GDP between 1999 and 2001 in both the latter countries. Social spending in Mexico also falls short o f the regional average. When disaggregating the figures, Mexico i s below the regional average for spending on both health and social security. Expenditure on education i s well above the regional average while expenditure on social assistance i s just above the average. In both categories, however, spending inMexico i s still only a half or even a third o f what countries such as Argentina and Chile spend(Table 3.2). Mexico's share of social expenditures as a proportion of public expenditures i s above 60%, similar to Argentina, Chile and Brazil. However, public expenditure as a percentage of GDP is lower than other middle-income countries in Latin America, so the challenge i s not to increase the share o f social expenditures but to generate an expansion o f the fiscal space. While countries like Chile and Brazil devote 16 and 19 percent o f GDP respectively to social spending, Mexico's allocation to the social sector remains at about 10 percent (see Table 3.2). Differences betweencountries inhow expenditure i s defined and recorded account for some but far form all of the difference. The critical issue is the direct link between the percentage of GDP allocated to public and social sector allocation, and the state's ability to collect taxes. The low figures shown for Mexico o f per capita social expenditures and for the share o f social expenditures to GDP are a direct reflection o f a thin tax base and low tax revenue. The fiscal priority o f social sector spending i s as high in Mexico as in other upper-middleincome countries (as measured by percentage o f total public spending allocated to the social sector). 25As noted by ECLAC, statistical series on total public expenditure and social expenditure in the region differ in terms of methodology and coverage, inparticular due to different ways of defining and recording social spending in national accounts. It should be noted that the figures do not include social spending funded by sub-national governments with own revenues (but do include that portion that is funded via transfers from the federal government). As the degree of decentralized state and local funding varies by country, public social spending figures are underestimated and therefore not fully comparable. Despite this caveat, the difference between social spending inMexico and comparable middle-income countries-both in dollar terms per capita and as a percentage of GDP-is too large to be fully explained by locally financed social spending. State expenditures financed via own revenues are still relatively small in Mexico, particularly in the poorest states. For example, in Mexico's three southern states of Oaxaca, Guerrero and Chiapas, state expenditures funded by own revenues represent just 4-6 percent of total spending, the rest of whichis financed via federal transfers (World Bank. MexicoSouthern StatesDevelopment Strategy, 2003). 32 Figure3.3. Per Capita Public SocialSpending, 2000-2001 (in 1997 USD) Argentina I Uruguay I Chile I Brazil I Panama I Costa Rica I Mexico I Venezuela I Colombia I Peru I Bolivia I Dominican Rep. I Paraguay I Ecuador I Guatemala I ElSalvador I Honduras I Nicaragua I LAaverage 0 200 400 600 800 1000 1200 1400 1600 1800 Note: Social Assistance i s the residual social spending once education, health and social security are accountedfor. Source: ECLAC, Social Panoramaof LatinAmerica 2002-2003. Table 3.2. Budeetarv Pressure and Fiscal Prioritv inLatin America. 2001-2002 Total public Percentageof total public spendingallocatedto socialsector spending as a (fiscal priority of socialexpenditure) percentage of GDP (Budgetary Under 40% Between40% and Over 60 % pressure)* 60% Over 30% Nicaragua (13.2) CostaRica (18.2) Argentina (21.6) Colombia (13.6) Brazil (18.8) Panama (25.5) Between30% and Honduras (10.0) Bolivia (17.9) Chile (16.0) 50% Venezuela (11.3) Under 20 % ElSalvador (4.2) Guatemala (6.2) Mexico (9.8) Dominican Rep. (7.6) Paraguay (8.5) Peru (8.0) Note: Figures inparenthesisshow the percentageof GDP allocatedto social expenditure. Source: ECLAC, Social Panoramaof LatinAmerica 2002-2003. 33 Mexico's Social Protection as a Share of Federal SocialSpending The following section presents information on social protection expenditures as reported in the Anexo Estadistico del Cuarto Informe de Gobierno, 2004. It follows the functional programmable expenditure categories as presented in the PEF (Presupuesto de Egresos de la Federacibn), as follows: `Social spending' reflects the budget category desarrollo social (social development) used in the PEF, and includes, as seen in Figure 3.4, 6 major sub-categories: education, social security, health, regional and urban development, food and social assistance, and labor. ``Social security" is a sub-category o f social spending, and includes all federal government outlays in pensions (which are funded by employee and employer contributions as well as transfers from the federal government); ``health" i s a sub-category o f social spending, which includes expenditures undertaken by the Ministry of Health (SSA), Ram0 33 (FASSA) as well as health services provided by IMSS andISSSTE; andhealth insurance benefits; ``poverg reduction" i s a cross-cutting category including all targeted poverty reduction programs; these programs are classified under both `social spending' and `economic sector spending'. Figure3.4. Composition of Federal Social Spending in Mexico, 2002 Regional and Urban Development 11.4% Food and Social Assistance 2.3% Labor 0 4% Social Security 249% Education 40.4% I Health 20.7% Source: Based on Anexo del Cuarto Informe de Gobiemo 2004. Social spending in Mexico is dominated by three expenditure categories: education (40 percent), social security (25 percent), and health (21 percent). As shown in the figure below, social spending i s dividedbetweenfive main categories: food and social assistance; social security; health (including SSA, and health benefits provided under I M S S and ISSSTE); education; rural and urban development; and labor. Education, health, and social security together account for about 86 percent of total social spending. An additional 11 percent i s spent on regional and urban development, leaving roughly 3 percent to labor and food/ social assistance interventions. Table 3.3 provides a breakdown of social spending, and separates out programs that also fall under the crosscutting category Poverty Reduction (Superacibn de la Pobreza). Using this category to capture social assistance spending, an estimate o f total social protection spending 34 would include all targeted poverty reduction expenditures plus social security, or roughly 33 percent o f total social expenditures (noting that some poverty reduction expenditures are classified under the economic as opposed to social sector), and 3.4 percent o f GDP. Total social spending equals about 10 percent of GDP and close to 60 percent of total programmable spending. Table 3.4 identifies all the major programs under Poverty Reduction, including those classifiedunder social spending and included inTable 3.3 as well as those under economic sector spending (such as agriculture and fishery development, for example) which increases the total spending on targetedpoverty reduction programs from 50.5 billion pesos to over 70 billion pesos - or 0.8 percent o f GDP. Social security expenditures are 2.6 percent o f GDP. Table 3.3. Federal Social SpendinginMexico by sector, 2002 In MX pesos Spending per Percentage of Percentage of Percentage of (Million) capitaa programmable social spending GDP spendingb Food and Social Assistance 14,582.8 141.5 1.4 2.3 0.2 Social Security 159,980.9 1,552.6 14.8 24.9 2.6 Health 132,878.7 1,289.6 12.3 20.7 2.1 Education 259,588.9 2,519.3 24.1 40.4 4.1 Labor 2,486.0 24.1 0.2 0.4 0.0 Regional and Urban Development 73,112.4 709.6 6.8 11.4 1.2 Total social spending 642,629.7 6,236.7 59.6 100.0 10.3 Poverfyreduction 50,564.0 813.6 4.7 NA 0.8 "Calculated using population estimates for 2002 from the Secretaria de Gobemacion (Consejo Nacional de Poblacion projections basedon the XI1Censo General de Poblacion y Vivienda del 2000). bTotal N e t Public Expenditure in 2002 was 1,484,256 million pesos, of which 73 percent (1,078,860 million pesos) was programmable. "Federal spending only. Source: Own calculations basedon the Anexo del Cuarto Informe de Gobierno 2004. Table 3.4. Federal Spending on SelectedPoverty Reduction Programs, 2002 In MX pesos Spending per Spending per Percentageof (Million) capita beneficiarya Progra"able (in pesos) (in pesos) spending Totalb 77753.5 754.6 NA 7.2% Oportunidades 17,003.8 165.0 4,010.3 1.6% Programa de Desayono Escolares 1,752.1 17.0 166.2 0.2% Programa de Abasto Social de Leche' Programa de Abasto Rural 568.0 5.5 1,705.0 0.1% Programa de Empleo Temporal (PET) 3,867.4 37.5 4,147.0 0.4% Programa de Opciones Productivas 520.9 5.1 894.3 0.0% Subsidios para Vivienda 628.9 6.1 29,103.6 0.1% Other 53,412.4 518.4 NA 5.0% "This refers to spendingper family inthe case of OportunidadesandSubsidiospara Vivienda,per producer in the case of Programa de Opciones Productivas,perjob in the case of PET, and per individual for the rest. bIncludesfederal, state andmunicipal spending. "In2000 and2002, LICONSA was entirelyself-financed and didnot receive any federal transfers. In2003, federal expenditure on this program was 267.6 million MX pesos. Source: Anexo del Cuarto Informe de Gobierno 2004. The largest single program classified under the targeted poverty reduction category i s Oportunidades, absorbing one fifth of the total budget. In addition to Oportunidades, the 35 largest poverty reduction programs include Programas Compensatorios del CONAFE ,Programa de Empleo Temporal (PET), Programa de Desayunos Escolares, Programa Opciones Productivas and Subsidios para Vivienda. "Other" represents expenditures on multiple small programs for which individual budgets are not available in the Anexo del Cuarto Informe del Gobierno. Figure3.5. Trends in SocialSpending and SocialProtection Spending, 1995 -2002 MillionlwTPesos 300,000 7 Education 250,000 ~ 200.000 Social Security 150,000 ~ 100,000 ~ 50,000 ~ 1995 1996 1997 1998 1999 2000 2001 2002 Source: Own calculations based on the Anexo del Cuarto Informe de Gobierno 2004 Social spending has grown rapidly in recent years, driven mainly by growing social security obligations in the unreformed ISSSTE and transition costs in IMSS; targeted poverty reduction expenditures have also increased, though less rapidly. Duringthe 1990s, social spending overall recovered from the severe budget cuts o f the 1980s. As seen inFigure3.5, social protection expenditure (represented by spending on social assistance, social security and poverty reduction) was among the fastest growing expenditure categories, with social security spending exhibiting a relatively dramatic increase of 14.9 percentper year over the decade. During the same period, poverty reduction expenditure also increased, yet at a slower rate. While the ratio o f targeted spending to social security spending was about nine tenths in 1995, it had fallen to about one-half in 2002. A look at preliminary figures for 2003 and 2004 reveals that total spending on poverty reduction has increased by nearly 30 percent inthe past two years (with a total o f 100,848.6 millionMX pesos budgetedfor 2004). SOCIAL INSURANCEPOLICY INMEXICO -OVERVIEW AND MAJOR CHALLENGES Mexico's social insurance system shares many of the characteristics and challenges of many Latin American economies. On one side a formal social security system provides retirement and health benefits to formal sector employees through two different subsystems, one for private sector workers and one for public sector workers; on the other, there i s a challenge o f providing adequate and sustainable coverage to people in the informal sector whose workers and their families make up a large share of the population. IMSS was original characterized as a partially funded definedbenefits system but in reality operated from the start as a pay-as-you-go 36 (PAYG) system.26 Since the 1940s, additional insurance schemes have been added, which provide often generous benefits to federal employees, the most important o f which i s the Instituto de Seguridady Sewicios Sociales de 10s Trabajadores del Estado (ISSSTE) which manages federal employees' accounts. In addition, the armed forces, the state oil company (PEMEX) and other state companies have developed individual social insurance systems. In 1997, a social security reform limited to IMSS was carried out, which moved toward a defined contribution system with individual accounts and private account management.The pushfor reform came mainly inresponse to the three most severe problems o f the old system: financial disequilibrium, inadequate pensions and high levels o f evasion resulting inlow coverage.27The 1995Ley deSeguroSocialpavedthe way for social security reformwhich began in 1997, while a second set o f laws passed in 1996, Ley de 10s Sistemasde Ahorro para el Retiro, allowed for privatized management. The reform moved IMSS from a PAYG (pay-as-you- go)/defined benefit system towards a defined contribution system.28The reform eliminated the old PAYG scheme; providing I M S S affiliates -at the time o f the reform- with a choice between benefits under the old system or accumulated balances under the new system; introducing a guaranteed minimumpension for low-income workers whose accumulated savings fall short o f a post-retirement income at that level; and eliminating cross subsidies among IMSS insurance branches.29Finally, the introduction o f individual retirement funds or Sociedades de Inversion Especializadas en Fondos para el Retiro (SIEFORES) managed by specialized companies or Administradores de Fondos para el Retiro (AFORES) represented a huge change to both the financing and provision of services. The results of this reform are discussed in further detail below. ISSSTE has not been reformed and represents an important contingent liability. Problems o f the social security system's financial sustainability need to be addressed. In2003 the social security scheme for public employees had a deficit o f 21.2 billion MX pesos, set to increase to 35 billion by 2006. Today's present value o f ISSSTE's implicit future liability equals 45 percent o f GDP in 2002. For every peso o f income, 2.5 pesos i s needed in subsidy, a precarious andunsustainable financial stance (Ministry o f Finance, 2004). Socialsecurity andpensions I M S S incorporates two redistributive measures: (i)a flat account subsidy to all affiliates (Cuota Social), and (ii)a minimum pension to poor affiliates. The Cuota Social i s a subsidy offered to all affiliates, with substantial poverty prevention effects for the poorer segments o f the affiliates. The subsidy does not vary with income and i s calculated as a percentage o f the minimum wage. As all affiliates receive the same amount, the Cuota Social share o f total pension contributions i s greater the smaller the affiliates income. Gill et al. (2004) find that the Cuota Social contributions represent more than 25 percent of total contributions of the 68.5 percent o f the Mexican workforce earning 3 minimumwages or less. For these workers, 26 Ina PAYG system contributions from current workers are used to pay benefits to current retirees. In a fully funded system, assets inthe pension fund reserve equal the liabilities (includingfuture) inthe pension system. 27 Grandolini, Gloria and Luis Cerda, 1998. "The 1997 Mexican Pension Reform: Genesis and Design Features", Washington DC: World Bank. 28Some observers have noted that the 1997reform did not switch to a new system but inreality introduced a different PAYG system. See Espinosa-Vega, Marco and Tapen Sinha, 2000. "A primer and Assessment of Social Security Reform inMexico", Federal Reserve Bank of Atlanta Economic Review, First quarter 2000. 29Grandolini and Cerda (1998). 37 the Cuota Social covers commission charges and allows eventual benefits to be greater than contributions, and can be seen as an incentive for low-income workers to join the system. In addition, Mexico has adopted the Chilean model o f ensuringa minimumpension for contributing affiliates. The government finances the pension directly when the balance o f the individual account i s exhausted. Affiliates with at least 25 years o f contribution and whose accumulated savings fall below a givenminimumannuity are eligible for the minimumpension. The degree of redistribution depends on whether or not coverage expands. The design o f the new system is meant to be both redistributive and to provide the correct incentives for workers to contribute, thus increasing coverage. According to the Mexico Public Expenditure Review (World Bank, 2004), as long as coverage expands, federal transfers to the scheme will be reasonably targeted. At any point in time, redistribution only takes place within the covered population. Figure3.6. DistributionofAdditionalPublicExpenditureon Pensions, 1996-2002 35% , 30% 25% 20% n 15% 50h lo%I 0% 1 2 3 4 5 6 7 8 9 10 Source: Mexico Public Expenditure Review,World Bank 2004. The distribution of increased federal expenditure on pensions, in particular given the subsidies that go to publicsector workers, in the past decadehas been highlyregressive. Federal transfers to finance pensions o f formal private and public sector workers increased greatly over the last decade.30Figure 3.6 shows the share o f the total increase in transfers that accrued to each income decile over the period 1996-2002. For IMSS, increases were relatedto the transitional costs o f the 1997 IMSS reform and they will diminish in time, while in the case o f ISSSTEthe increase in federal transfers was driven by a need to finance a growing gap between current contributions and pensions payments. Federal transfers to I M S S increased by 147 percent from 1997 to 2002 (from 21,021 to 51,986 million pesos). Inthe same period the ISSSTE deficit grewby 336 percent (from 3,377 to 14,728 millionpesos). 3oMexico Public ExpenditureReview (World Bank, 2004). 38 ms Figure3.7. Distributionofpensionersand benefitsreceived,2003 Pensioners Benefits Luz y Fuerza 30% employees lMSS 23% Source: Mexico Public Expenditure Review,World Bank 2004. Finally, significant inequalities exist between the pension systems for public and private employees and within the public system. A focus on IMSS and ISSSTEfails to reveal the full extent o f existing inequalities within formal pension schemes in Mexico. In addition to ISSSTE, a range o f systems cover specific groups o f public employees. These groups, which include the armed forces (SEDENA, SECMAR), public workers employed by local governments, the principal state companies (PEMEX, CFE, Luz y Fuerza), and I M S S employees, have been able to negotiate exceptionally generous contractual obligations from the federal government, resulting in inequalities between public and private sector workers, and inequality within the public sector. For instance, while PEMEX, Luz y Fuerza and IMSS employees represent only 8 percent o f all pensioners, they absorb more than a thirdo f the benefits. Regardless of positive changes in the incentives to participate, increasing coverage of social security systems for privateworkers is still an important challenge to the system. The objectives of the wave of reforms in Latin America in the nineties, were, among others, to make the system financially sound and increase coverage through providing better incentives to workers and firms. As with many o f the reforms implementedin Latin America, observers still debate the impact o f the 1997 social security reform for private sector workers in Mexico (Kaplan, 2004), particularly with respect to formal employment, but overall coverage ratios remain low, especially among the poor. In Mexico, there was a rapid surge in affiliation during the 4 years following the reforms, however, the pace o f new affiliations stagnated with the deceleration o f growth. In any case, systemic changes that lead to a better link between contributions and benefits are a move in the right direction. Uniform contributions to health insurance are a good example o f how linking contributions to benefits increases the incentives to participate inthe formal labor market and social insurance schemes. Underthe old definedbenefit system, contributions varied in line with the wages o f the affiliate, while benefits were the same for all. In the new defined contributions system, the contribution i s uniform and based on the minimum wage, removing the perception of an extra tax on anyone earning above minimum wages relative to those who earned only minimumwage but received the same benefit. Estimates suggest that the strengthened link between contributions and expected benefits had a positive impact in Mexico resulting in an immediate surge in formal employment (Montes and Santamaria, 2004). Another view attributes the surge in formal employment to improved 39 r e g i ~ t r i e s .A~ ~considerable increase in federal contributions, from 0.6 to 13.9 percent o f the minimumwage, was also designedto act as an aggressive stimulus to formalization (Montes and Santamaria, 2004). In 2002, about 40 percent of the Mexican population was not covered by any pension system. Also, according to the World Bank's assessment o f the government poverty reduction strategy (2004), half the population falls out o f the formal health insurance system, while one tenth is without access to healthcare. Furthermore, as informal employment is linked to lower income, social insurance coverage i s not only limited but also highly regressive. As a result, the overall social insurance system i s dualistic, i.e., divided in terms o f the covered and uncovered population, which corresponds to the formal and informal populations Further, there are some threats to the financial sustainability of IMSS. According to a recent evaluation of the system publishedinthe Tercer Informe de la Situacibn Financieray 10s Riesgos del Instituto 2003-2004, I M S S needs to resolve three critical challenges in order to perform its functions ina sustainable and efficient manner.32 First, there is an ongoing dispute between the institute and its employees over the particularly advantageous retirement benefits enjoyed by these workers. Average age o f retirement for IMSS employees i s 53 years compared to 65 for other IMSS affiliates, yet employees o f the institute retire with a pension that i s on average 30 percent higher than their last salary, while other I M S S affiliates receive a pension upon retirement that i s on average lower than their last salary. Consequently, the pension o f an ex-IMSS employee i s on average 8.3 times the size o f that o f any other affiliated worker - in spite of having worked on average 12 years less. This generous pension scheme i s largely paid for by contributions o f other I M S S affiliates and not by those o f the I M S S employees themselves. The calculations in the above mentioned report show that only 6 percent o f the cost i s covered by I M S S employees' contributions, while 77 percent i s paid for by other affiliates, representing 19 percent o f total worker and employee contributions. Urgent reform i s needed to redress this disequilibrium and to avoid even graver damage to the financial viability o f the institute. The report projects a doubling o f the number o f retiredIMSS employees inless than a decade. Second, the demographics of the Mexican population result in a relative decrease in the number of working, contributing affiliates relative to the number of retired, benefiting affiliates. Increased life expectancy will increase the weight o f medical expenses for pensioners, inparticular those relatedto chronic illnesses. Contrary to the first challenge which is particular to the Mexican system, the problem o f demographic change poses challenges similar to other OECD countries' social security systems. Third, IMSS is far from fulfilling its social obligation of securing all Mexican non public sector workers. Due to years o f backlog in formal job creation and the impact o f the economic downturn since 2001, only 29 percent o f the economically active population was covered by IMSS in 2003. With no change in limited coverage, 11 million workers will retire with no pension in the following 25 years. The need to increase the coverage ratio runs against the urgentneed to increase I M S S revenue as higher contribution rates would further decrease the incentives for informal workers and employers or self-employed to register formally. In fact, decreased contribution rates mightbe envisaged as one way to provide the needed incentives for 31Comments received duringthe June 10-11,2004 Poverty Assessment workshop inMexico City. 32IMSS, "Informe a1Ejecutivo Federal y a1Congreso de la Union sobre la Situacion Financiera y 10s Riesgos del Instituto Mexican0 del Seguro Social" httD://www.imss.g;ob.mx/IMSS/IMSS INF/2004/fina04.htm 40 workers and employers to register formally. The problem o f low coverage i s further complicated by dynamic changes inthe labor market asjob creation over the past years has taken placemainly in service and commerce sectors with smaller business structures and a high incident of micro- enterprises and self-employed. It i s a known fact that the probability o f formal registration increases with size. In Mexico, affiliation i s optional for the self-employed. Finally, the proportion o f temporary as opposed to permanently employed workers i s increasing, creating problems o f insufficient time o f contribution. Female participation inthe labor force i s increasing - a population for which temporary work i s more widespread. Agricultural day-workers are in a particularly fragile position due to seasonal changes in labor demand, while the problem i s exacerbated among migrant workers that dividetheir time betweenMexico andthe UnitedStates. Some state governments have introduced non-contributory transfers to the old-age population in an effort to address the existing gap in old-age insurance coverage; the elderly are also covered inpart by nationalprograms such as Oportunidades andProcampo. In2001, Mexico City Government (MCG) launched a program providing transfers to the elderly (Programa de Apoyo Alimentario y Sewicios Me'dicosy Medicamentos Gratuitospara 10s Adultos Mayores de 70 aAos). The program was initially poverty targeted as only the elderly living in poor sections were eligible beneficiaries o fthe pension scheme. In2003, coverage was extended to all residents at the age o f 70. Other states operate transfer programs for the elderlypoor, includingthe Canasta Alimentaria Basica (Michoacan), the Programa de Atenci6n a1Adulto Mayor (Nuevo Leon), and the Pensi6n para Adultos Mayores (Guerrero). National programs that are targeted to the poor also reach the elderly; Oportunidades reaches close to 400,000 families headed by someone over the age o f 70, while more than 650,000 Procampo beneficiaries were over age 70 in 2004. In sum, multiple initiatives exist which reach the elderly poor either by design or accident, and government entities at multiple levels are stepping into attempt to address the issue. There i s still the need o f a coherent policy for addressing the issue o f old-age poverty in an efficient, fiscally sustainable manner. Thenational health system and health insurance Overall public expenditure on health services is mildly regressive; however, it becomes neutral if private contributionsto IMSS are deducted. The mid-regressivepattern in total public expenditures in health results from regressive expenditure on services for the insured (a result o f limited and regressive coverage) coupled with highly progressive expenditure on services for the As shown in Figure 3.8, federal expenditure on health services can be dividedinto fundingo f institutions servicing the uninsured(Ministryo f Health SSA) and the - insured (ISSSTE and IMSS). When looking at the distribution o f expenditure across income deciles, public financing o f health services for the uninsured i s highly progressive, while the public financing o f health services for the insuredpopulation inthe public system i s regressive in absolute terms, butnot relative to income. 33World Bank. Mexico Public ExpenditureReview. 2004. 41 Figure3.8. Federal health expenditure, insuredvs. uninsured, 2002 Insured - ISSSTE 10.6% Uninsured - SSA 33.4% Insured- IMSS 56.0% Source: Mexico Public Expenditure Review,World Bank 2004. The expansion o f public expenditure o n health services in the period between 1996 and 2002 was accompanied by increased progressivity and more pro-rural spending. Overall, federal health expenditure rose by 40 percent in the time period 1996-2002. Duringthe same period, the share of federal expenditure targeting the uninsured (SSA) increased from 21.5 percent of total expenditure in 1996 to 33.4 percent in the 2002. This illustrates h o w the expansion in public expenditure between 1996 and 2002 favored the uninsured, narrowing the gap between the two parts o f Mexico's dualistic health system. As such, between 1996- 2002, public spending on health became more pro-poor as the poorest 2 deciles accounted for 15 percent of spending in 2002 compared to 8 percent in 1996. It also became more pro-rural-up to 28 percent from 20 percent-reflecting both increased spending on SSA services inrural areas and an expansion of I M S S coverage inrural areas between 2000 and2002. 42 Figure3.9. Distribution across income deciles of marginal benefitsfrom growth in public health expenditure, 1996-2002 1 OTotal 0Uninsured - SSA .Insured - IMSS I U Source: Mexico Public Expenditure Review,World Bank 2004. However, per capita health spending varies widely geographically, with poorer states receivingless than richer ones. The state that receives the highestper capita expenditure on health (Baja California) receives more than three times more than the state that receives the least (Michoacan). Geographical differences become even starker when distinguishing between the insuredand uninsured:the insuredpopulation inBaja California receives more than five times as much inper capita health expenditure as the uninsuredpopulation inP ~ e b l a . ~ ~ Finally, Mexico's national health system is characterized by a parallel and strong presenceof private health providers that serve both the coveredand uncoveredpopulations. As a result, the national health system consists of a fragmented public system and a significant private sector operating largely in parallel with very little coordination (Cercone 2000). In short, the system i s built on three pillars: one servingthe insured, another serving the uninsured, and a third pillar- the private sector -serving both. First, there are the mandatory social insurance programs for formal sector employees, employers and government, funded by contributions and ranby IMSS, ISSTEetc. Second, there i s the social assistanceprogram for the uninsuredfinanced by general revenue and provided by the public sector (Le., SSA). Third, there are private sector health providers, usedby both insuredand uninsuredand financed by out-of-pocket expenditures. Following the decentralization o f health services in the mid-1990s, states are responsible for health services provided by the SSA to the uninsured.States also runtheir own health institutions. T o address the problem of limited health insurance coverage among the poor and health shocks, a voluntary insurance scheme - the Seguro Popular (SP)-was launched by the Ministry of Health in 2001. The Seguro Popular program was first piloted in 2001 and by early 2004 became known as the nationalprogram Sistemade ProteccibnSocial en Salud (SPSS); as o f May 2005 the program provides coverage to 2,000,000 families. Although based on 34Mexico Public ExpenditureReview. World Bank. 2004. 43 traditional insurance design features, the Seguro Popular effectively operates more as a targeted transfer mechanism built around the package o f basic health services due to a heavily subsidized package o f benefits coupled with a progressive contribution structure. In addition to the self- targeting mechanism o f minimum benefits, the Seguro Popular uses an income/assets survey to evaluate capacity to pay. Although the program i s funded and administered by the Ministry o f Health, the actual provision o f health services included in the package i s devolved to the states. As of April this year, 700,000 families have signed up for the Seguro Popular, and the government estimates that by the end o f the year 1.5 million households will have signed up. It should be noted, however, that the program presents new fiscal liabilities that mustbe considered, especially since the number o f services offered by the program continues to grow. A second voluntary insurance scheme, the Seguro de Saludpara la Familia (SSF), compliments the SP. The SSF operates within existing social security institutions and targets informal and self- employed workers able to pay a premium contribution. SSF coverage i s still very low about 350,000 beneficiaries and has not increased much inthe last year, about 2 percent. Box 3.2: Decentralization of social spending and service provision in the 1990s - Health and Education In the 1990s increasing decentralization led to a consequent change in the composition of social spending between federal and sub-national levels of government. Decentralization o f public expenditure was rapid, and massive in terms o f the magnitude o f decentralized expenditure. Decentralized spending represented 42 percent o f total social spending in2000, comparedto 29 percent in 1994.35 Rapid decentralization of resources, coupled with continued centralization of taxation and lacking incentives for states to increase their own revenue has left sub-national governments highly dependent on federal transfers. The disconnection between expenditure and revenue has emergedas one o fthe key challenges o f fiscal federalism inMexico. Two waves of decentralization with significant transfers of social spending and basic service provision to local governments: Late 1980s - Early 1990s: Tax authority was not devolved, however responsibilities were delegated to states and supported by transfers. Continued federal control over spending priorities was ensured by earmarked (and conditional) transfers. Starting inthe late 198Os,the provision o f basic health services was gradually transferred to states, while responsibility for basic education devolved more rapidly following the 1992 Acuerdo Nacional para la Modernizacidn de laEducacidnBasica Late 1990s: Continued decentralization, driven by increased political competition and demands from state governments ran by opposition parties. Decentralization relied on transfers, however earmarking and linkages to expandedresponsibilities were greatly relaxed. Major reforms included the creation in 1998 o f a new budget line-Ram0 33-which increased thepredictability ofhealth and education transfers to the states. With the exception of the new Seguro Popular program, none of these decentralization reforms affected social protection policy. Both social assistance transfers and social insurance remained centralized. Seguro Popular, a program which involves heavily subsidized health benefits for the poor and therefore some risk pooling across households has been largely devolved to the states, following the massive decentralization o f health services that occurred inthe 1980s and 1990s. Source: Giugale and Webb (2000), Haggard and Webb (2004). 35Rodriguez, Evelyn, 2003. "Some Notes on Changing Social Policy: Mexico's Experience", presented at World Bank Safety Net Primer Launch, December 1,2003. 44 SOCIAL ASSISTANCE POLICY INMEXICO-OVERVIEW AND MAJOR CHALLENGES The 1980s and 1990s brought about great changes infederal socialpolicyinMexico, as the focus of social assistance shifted towards poverty alleviation. While programs in the 1980s focused on `compensating' the poor for their situation, the 1990s saw the emergence o f more dynamic programs geared towards the development o f human capital and productive capacities in the poor. This trend culminated in the launch o f Progresa in 1997. The program combined innovative program design built on conditional transfers and cross-sectoral synergies with rigorous monitoring and evaluation. A new wave of social policy initiatives coincidedwith the first structural reforms of the late 1980s and early 1990s, initiating a move away from generalized subsidies towards targeted poverty alleviation incorporating decentralized,local participation. Acknowledging that the poor were the most disadvantaged both in terms o f coping under economic stress and in terms o f adapting to structural adjustments, the Salinas administration developed Programa Nacional de Solidaridad(PRONASOL). This program was one of the first to specifically target the poor, representing an important break with earlier policy. A second innovative feature o f PRONASOL was its focus on community participation, coupled with the decentralization o f both resources and decision-making. Despite these advances, the majority of poorly-targeted generalized subsidies remained in place.36 Furthermore, programs within PRONASOL lacked well-specified objectives as well as corresponding monitoring and evaluation, contributing to the perception o f PRONASOL as a vehicle for ~ l i e n t e l i s m . ~ ~ By the 1990s the policies of PRONASOL had been discredited, paving the way for innovative social policies that turned the focus of social assistance towards building the human capital and income generating capacities of the poor. The move away from general subsidies toward targeted social assistance initiated by PRONASOL continued throughout the 1990s. Changes were made to some but not all subsidies. Generalized food subsidies such as the universal tortilla subsidy were phased out giving way to food subsidies targeted to the poor. In addition, important advances were made interms o f increased transparency and accountability, by ways o fmore systematic monitoring and evaluation. The launch in 1997 of the Programa de Educacidn, Salud y Alimentacidn (Progresa, now called Oportunidades),a social assistance program targeting rural infants and youth, is the clearest manifestation of this new wave of policy initiatives. Progresai s a conditional cash transfer (CCT) program that provides money to extremely poor families conditional upon investments in human capital such as sending children to school or bringing them to health centers on a regular basis. This conditionality makes CCT programs an instrument not just for short-term social assistance, but for longer term human capital investments. CCT programs are 36Lopez-Calva refers to an evaluation of generalized subsidies realized by the SECOFI in1992,which shows that of every peso disbursed in the Programa de Subsidios Generalizados, only 50 centavos reached the beneficiaries due to highadministrative costs. Inaddition, by favoring the urban middle class, the subsidies left behind large portions of the population. Conference proceedings from "Innovations in Social Policy Conference", SEDESOL, Mexico City, April 2003. 37Perceptions of clientelism were also generated by the practice of having earmarked transfers sent directly to PRONASOL committees, bypassing mayors and municipalities, following `a distinctive political logic' (Haggard and Webb, 2004). Diaz-Cayeros and Magaloni (2003) found that the timing of PRONASOL expenditure responded to the federal electoral cycle, and allocation was in part determined by the competitiveness of electoral contests. 45 part o f a new generation o f social programs that use demand-side financing to target the poor that includes school voucher programs and subsidized health insurance schemes. These programs' reliance on market principals, usingdemand-side interventions to directly support beneficiaries, i s a marked departure from traditional supply-side mechanisms such as general subsidies or investmentsinschools, health centers and other providers o f social services. Oportunidades pioneered a new generation of social assistance programs (conditional cash transfers programs) that have been emulated internationally. It also introduced consistent monitoring along with rigorous, independent evaluation. In addition to enforcing the trend towards targeted poverty alleviation, the program reoriented the focus o f social assistance to building the productive capacity and human capital o f the poor. Finally, by linking nutrition, health and education, the program marked a transition towards a more integrated approach to poverty alleviation. A cross-sectoral approach was seen as more effective than isolated programs in responding to the multidimensional needs o f the poor. While Oportunidades targets infants and youth, such an integrated approach has been used in recent regional development programs developed under the current administration, including Habitat (urban areas) and Microregiones (remote and marginalized rural areas). One o f the main current challenges in reforming the social protection system i s to consolidate these new trends in social assistance and to disseminate the principles and lessons from Oportunidadesto all parts o f the social protection system. So far, take-up has been low. Figure3.10. DistributionofHouseholdsbenefitingfrom public ScholarshipPrograms, 2002 45% ~ 40% 35% 30% 25% I 20% 15% 10% 5% 0% 3 4 5 6 7 8 9 10 OOportunidades Mother Source: Mexico Public Expenditure Review,World Bank 2004. Increased spending on education has been highly progressive as a result of Oportunidades and the increase in means-tested scholarships. Figure 3.11 shows how Oportunidades has grown to become the largest public scholarship program in Mexico. More importantly, the graph also illustrates how progressive the program is. O f all households 46 receiving Oportunidades transfers, a total o f 93.4 percent were in the bottom half o f the income distribution. 63.6 percent were among the poorest 20 percent o f the population and 39.5 percent among the poorest 10 percent. More than 80 percent o f Oportunidades households were in rural areas in2002, however this has changed inthe past two years due to the roll-out o f the program to urban areas (see Figure 3.11). Households benefiting from other scholarships are concentrated in urban areas (76.4 percent). This positive trend holds mainly for primary and lower secondary, while the distribution o f scholarships for upper-secondary and tertiary education remains regressive. Figure3.11. Extensionof Oportunidadescoverage, 1997-2004 (thousands of families) 6000 , 5000 ~ 10004 0h 1997 1998 1999 2000 2001 2002 2003 2004 W Rural .Semi-urban -Urban Source: Anexo del Cuarto Informe de Gobierno 2004. Coverage of Oportunidadeshas increased considerably in the past couple of years as the program moved to include upper-secondary students in 2001, followed by an expansion into urban areas in 2002. In2002, the program provided more than 2.5 million scholarships to primary school children and 1.3 million scholarships to secondary school adolescents n a t i ~ n w i d eCoverage was about 60 percent more than in2000. The original program design has . ~ ~ not changed as a result o f urban expansion, but recent experience in adapting the program to urban areas without considerations o f how the model would function in a different context suggests that some program modifications may be needed. Figure 3.11 shows the number o f beneficiaries over the time period 1997-2004, broken down by rural, semi-urban andurban. In 2002, the Mexican government introduced the Contigo strategy, an `umbrella' framework designed to achieve greater coherence and impact in social protection policy. Programs are designed around age and gender groups and action areas so as to ensure that individuals' needs o f government services and assistance at different times in their life and in differentwelfare dimensions are covered. Its stated goal i s to provide basic social benefits to all Mexicans, including quality healthcare and education, adequate nutrition, housing, employment security and pensions (Secretaria Te`cnica, 2003) (see Figure 3.12 below). In addition, the conceptual framework includes four transversal principles, applicable to all areas o f public action: (i)equity to allocate more resources to disadvantaged groups; (ii)transparency to avoid discretionary or political use o f government and society; (iii) responsibility between the joint 38Rubio and Soloaga (2002) 47 three levels o f government (federal, state and municipal); (iv) integration to encourage efficient social spending by exploiting synergies, eliminating duplication and ensuring coherence in differentprogram objectives. Figure 3.12. Contigo:Interactions-BetweenAreas of Public Action and the Life Cycle .. ~~ ~ /Cumulative Cvcle) (Cumulative - 4 cycle) r PrenataI Every action takes into accountgender and ethnic differences across people 1 ChiIdhood The individualand the family are the unitfor action 1 Teenagers and Youth Adults Elderly r Recent social protection initiatives illustrate the new social protection framework, in particular in terms of delivering integrated interventions drawing on synergies between sectors and programs: The inclusion of upper-secondary students in Oportunidadeswas accompanied by a savings plan for participating high school students called Jdvenes con Oportunidades. The program offers a bonus which grows each year from ninth grade through graduation and turns into a savings fund if the student completes high school before turning 22. The program has important links with other services. Students can use the bonus to buy health insurance such as the Seguro Popular, for collateral either in micro-credit or to enter into the government's housing credit program, Tu Casa. Alternatively, the savings can help fund higher education. By 2004, J6venes con Oportunidades operates close to 40,000 savings accounts on behalf o f Oportunidades beneficiaries. Microregiones i s a new, largely rural crosscutting program focusing on local development in remote and marginalized areas often outside of the reach of Oportunidades.The main objective o f the program is to buildbasic social infrastructure complemented by strategies to integrate the marginalized area into a broader economic network. The program i s participatory inthat local communities, via local councils, take part in the decision-making process by proposing and prioritizing investments. The program also seeks to identify and support strategic centers within each region. 48 As a complement to Microrregiones, the Estrategia de Abasto, Alimentacidn y Nutricidn offers food subsidies to families living below the food poverty line in remote and marginalized areas, specifically outside the reach of Oportunidades. While there i s evidence that the percent o f underweight children was practically halved over the course o f the decade, estimates for the share o f children under 5 years o f age that still suffer from stunted growth in its moderate form range as high as 18 percent.39 Moreover, 18.6 percent o f total households had a per capita income below the minimum requirement to fulfill their daily energy food intake included in the INEGI-ECLAC's basic food package.40About 35 percent o f these households are located in remote areas with less than 500 inhabitants, and the program focuses on reaching these marginalized communities which often do not systematically receive any other support. The food subsidy is distributed using the network of Diconsa stores and with the support of the Microrregiones program. A third cross-sectoral social assistance program, Habitat, was launched in 2003 to combat urban poverty. The program combines and coordinates poverty reduction programs and urban development projects and targets both poor neighborhoods and poor households. Hhbitat has been instrumental in improving the quality and access to basic services such as water, electricity and sewage in targeted urban areas, and has helped create community development centers (Centros de Desarrollo Comunitario) and child care centers. Special attention i s given to female heads o f households, the elderly and the disabled. The program covers about 180 cities and the latest extensions o f the program have focused on border cities in the North and South and 10 historic city centers. The budget for 2004 was MX$5 billion, up from MX$2billion in2003. Even though Mexico's various social assistance programs may be justified by the different demands of a heterogeneous population, their sizeable number reflects potential problems of overlap and coordination. Many programs are aimed at the same target population but vary in terms of the geographical region attended, the economic sector in which the population i s located, the type o f instruments employed, or merely the entity responsible for its operation. The number o f programs, by itself, generates a serious problem o f inefficiency. Coordination problems grow with the number o f programs, efforts are duplicated, and as a result some segments o f the target population may be left unattended or covered simultaneously by more than one program. Finally, programs formally belonging to other sectors perform central social protection functions, as seen in the income security and redistributive functions o f Procampo (agricultural subsidy). Aside from Oportunidades, Procampo i s the government's main poverty- targeted transfer program. In general, the Mexican state has been heavily involved in rural development. Expenditure in agricultural and rural development represents 34 percent o f all line- ministry expenditure and almost 2 percent of GDP (Rural Poverty in Mexico, 2004). Naturally, many Agricultural and Regional Development (ARD) policies and programs are not directly targeted to poverty reduction, however while pursuing other goals (e.g., farm modernization or competitiveness), they directly or indirectly affect ruralpoverty and inequality. A subsidy such as Procampo also provides income security for the poor. Government transfers, both in terms o f social assistance and ARD, constitute a considerable share o f the incomes o f the rural poor. There i s an increasing trend inARD expenditure and some evidence o f increased targeting o f programs. The dependency of ruralpoor on transfers is contrasted by the relative absence of transfers inthe 39 National NutritionSurvey 1999.2001 edition andWorld Bank. Mexico Poverty Assessment. July2004. 40 Ministry for Social Development. Technical Committee for Poverty Measurement, 2002. 49 --- incomes o f urban poor, new, smaller initiatives such as Hhbitat notwithstanding. Finally, Urban PoverQ in Mexico (World Bank, 2004) underscores the importance o f the labor market as a risk coping mechanism for the urban poor. Box 3.3: Procampo Aside from Oportunidades, Procampo is the government's main poverty-targeted transfer program. PROCAMPO, introduced in 1993, is a program o f direct payments scheduled to end in 2008, which compensates producers for the loss o f input subsidies, price supports, and import protection incurred by Mexico's entry into NAFTA. The program covers 2.8 million producers out o f which 2.3 million are farmers on ejidos. Planned expenditure for the program was 12.4 billion pesos in2002. Main objectives: 0 Support domestic producers o fbasic staples to face competition from the UnitedStates HelpMexican farmers switch to more competitive crops under a liberalizedcontext Procampo has a progressive overall distribution as poor subsistence farmers represent a large proportion o f producers o f basic crops, the poor (as defined by the food poverty line) benefit from 33 percent of transfers. Estimates referred to in the PER show that for the poorer ejidatarios, the Procampo transfer represents up to 40 percent o f income. But over 80 percent o f households in the poorest households do not receive program benefits, and incidence stretches significantly across the income distribution. DistributionofProcampo Transfers acrossincome deciles, 2002 18% Despite a progressive distribution of 16% transfers, Procampo reaches few poor 14% households not covered by either 12% Oportunidades or public social security IO% schemes. Less than 2 percent of extreme 8% poor households not covered by 6% Oportunidades or IMSS receive Procampo 4% transfers (ENIGH2002). 2% 0% 1 2 3 4 5 6 7 8 9 10 Source: Mexico Public Expenditure Review (World Bank 2004) and Poverty in Mexico (World Bank, 2004) In sum, although the Contigo strategy works well as a conceptual framework, earning the government deserved acclaim, the strategy has yet to be fully operationalized (World Bank, 2004 - Poverty Assessment Report). With the exception o f a few cross-sectoral programs including Oportunidades and Hhbitat, it i s unclear how the wide variety o f social assistance programs fit in the Contigo strategy. Most social and economic policies affecting the poor seem to continue along sectoral and ministerial lines; as seen in Table 2.2, social assistance policy today i s delivered via almost 90 different programs, from a multitude o f responsible entities, incorporating little o f the Contigo framework into decision making. 50 WHERE DOES THE SOCIAL PROTECTIONSYSTEMSTANDTODAY? To conclude, Mexico's social protection system i s at a critical juncture. The 1997 IMSS reform o f the pension system for private workers moved in the right direction by linking contributions and benefits thereby improving the financial underpinnings o f the system. The ISSSTE system covering public sector employees, however, i s still unreformed. As a whole, the social protection system in Mexico, as in many Latin American countries, i s still fragmented on the basis o f labor market status and a large fraction o f the population still has no or inadequate coverage. Social insurance programs which constitute the bulk o f social protection spending continue to suffer from financial instability and insufficient coverage among the poor, and despite progress in Oportunidades,social assistanceremains fragmented. Mexico's social protection system has seen major progress in the past two decades, as illustrated by major innovations such as the Oportunidadesprogram, and the conceptually strong framework now provided by Contigo. Innovative approaches have emerged dealing with the multidimensional characteristics o f poverty, notably through cross-sectoral programs such as Oportunidades, Microregiones and Habitat. Monitoring and evaluation practices are becoming increasingly rigorous. However, the majority o f social assistance programs have yet to incorporate these concepts as rigorously as the government's flagship social assistance program Oportunidades, nor i s it clear how some o f the smaller social assistance programs fit into an integrated Contigo framework. Box 3.4: Argentina's experience with workfare programs in times of crisis - Jefas y Jefes deHogar Inresponse to the 2001 crisis, the Argentine government launchedthe workfare program, Jefas y Jefes de Hogar Desocupados (Unemployed Heads o f Household). The program provides 150 pesos per month to unemployed household heads or their spouses in exchange for 20 hours per week of community service work, job training or work as a temporary employee of a private company. Eligible households are those with at least one o f a child under the age o f 18, a pregnant woman andor a handicapped member. Either the husband or the wife can participate in the program, provided that their spouse i s not working. Program eligibility criteria were not tightly enforced- about one third o f those receiving the program did not satisfy eligibility criteria. Yet, given the evidence that a fall in real wages, rather than unemployment, was the significant factor behind the decline in living standards participationby people who were not formally eligible may not have been a badthing. In addition, the fact that beneficiary unemployment status is hard to verify in economies with high rates of informality makes this eligibility requirementunenforceable. Indicators o f structural poverty such as having dependants andor living in households with no members in the formal labor markets may have been more effective eligibility criteria in ensuring pro-poor targeting. Overall, targeting was good as about over half o f the participants came from the poorest fifth o f Argentine families, and all but 10 percent fell below the official poverty line. The program is estimated to have reduced Argentina's unemployment rate by only about 2.5 percentage points. However, the effect on poverty, and more particularly on extreme poverty, was significant: close to 10 percent o f the participant would have fallen below the food poverty line without the program. The program helped participants who would have suffered an appreciably sharper drop intheir income without the program. Among the lowest quintile, income to males from work programs increased from 2 to 16 percent o f total household income; for females, the equivalent increase was from 3.4 to 21.8 percent. Source: Adapted from Poverty in Urban Mexico, M c Kenzie (2003); Galasso and Ravallion (2004). 51 With regard to social security, workers in the formal sector and their families are provided social protection by social security institutes, the most prominent o f which are the I M S S and ISSSTE.41The private pension system has the challenge o f increasing coverage, while the pension system for public sector workers provides generous benefits remains unreformed and maintain severe financial imbalances. Informal workers and their families, in contrast, have relatively limited access to social protection. For health care, these citizens rely mostly on the Secretaria de Salud (SSA) or out-of-pocket fee-for-service inthe private sector. Inold age, these citizens do not have a pension and hence rely on own savings and intra-family support. A large group o f people, however, neither have access to these programs nor to the formal social security institutes; these citizens represent the "missing middle" o f Mexico's social protection system. Chapter 2 raised the issue o f increasing risk and vulnerability to job loss and unemployment: Mexican labor markets seem to be increasingly absorbing output shocks through higher unemployment rates, yet current unemployment protection system only protects formal employees.42 The PoverQ in Urban Mexico report takes note o f this gap in social protection, highlightingthe need to develop safety nets that protect the urban poor and help them mitigate risks, particularly workfare and other social insurance programs that can be quickly deployed when a crisis hits. Lacking unemployment insurance and the limited access and uncertain future o f workfare programs such as PET, may become an obstacle to the provision o f comprehensive social protection o f Mexican workers. In short, despite importantadvances,Mexico faces substantial problemsinits social protectionsystemthat must be acknowledgedand addressed. Reformingthe social insurance system remains a priority both in terms o f addressing the coverage gaps particularly among the poor in the informal sector, as well as in response to growing financial obligations under the present system. Cast in terms o f the framework established in Chapter 1, the bulk o f Mexico's social protection interventions (and resources) flow towards providing risk-pooling mechanisms where markets fail to provide. Financingfor these programs, particularlyin the provision of retirementinsurance,involvesbothcontributionsfrom affiliatesand a subsidy from general tax revenues, mainly for public sector workers pensions while a substantial proportionof - the population is excludedfrom coverage - thus, the persistent dualistic nature ofthe Mexican social protection system. There have been important reforms in social protection initiatives targeting the chronic poor, but resource flows are still insufficient and the major assistance program-Oportunidades-does not in and o f itself constitute a shock insurance mechanism. Two other major programs, Procampo and the new SeguroPopular program, do provide some form o f riskmanagement for the chronic poor, yet implementation andfinancing questions remain, and an important proportion of the extreme poor population appear to be excluded from even these programs. The system as a whole i s less a `system' than a set o f new social assistance initiatives. Whereas Contigo presents a compelling and elegant framework, it has yet to be fully operationalized, andprograms under its umbrella remain largely fragmented and uncoordinated. Three key issues can be identified: First, the social protection system remains dualistic andfragmented, with large gaps in coverage for key vulnerable groups. Potentially promising new programs remain uncoordinated institutionally and present challenges o f financial viability. The Contigo strategy remains largely conceptual and the individual reforms remain ad hoc and cannot 41The Instituto Mexican0 de Seguro Social and the Instituto de Seguridad y Servicios Sociales de 10s Trabajadores del *Estado,current respectively. The unemployment system consists primarily of a defunct severance pay system. 52 be anchored within a coherent social protection systemwithout fundamental reformto the structure and composition o f the overall system. And without fundamental reform, these individual reforms will not be sufficient to address the core, structural issues outlined above. Second, Mexico faces the challenge of stepping-up and mainstreaming the social protection reform agenda. Many o f the laudable qualities Mexico brought to the reform o f its social assistance system through the introduction o f the flagship Oportunidades program need to be introduced throughout the social protection system more broadly. Indeed, without paying attention to strengthening the accountability relationships between clients, service providers and government throughout the system, the overall functioning o f the system will almost certainly suffer from quality and performance problems. From a public sector management perspective, Mexico needs to mainstream the practice o f introducing new programs while phasing out less viable and costly alternatives, allowing evidence-based policy to guide those decisions, and thus de- politicizing the process. Third, the reform must be grounded within a strongfiscal strategy with clear decisions made as to how reforms will be financed. Choices include reallocating expenditures within the current social protection expenditure envelope (in particular, reallocation of existing subsidies embeddedinthe system), or, addressing existing tax structures in order to generate greater public resources to finance the social protection system. If these questions are not addressed, new add-on policy initiatives risk further weakening the solvency and stability o f the social protection system as a whole. 53 4. TOWARDS AN INTEGRATEDSOCIAL PROTECTION STRATEGYINMEXICO:A DISCUSSIONOFISSUESAND OPTIONS The objective of this chapter is to lay out the most important issues that need to be addressed in the current social protection debate. The princ@al social protection challenge facing Mexico today is how to place risk management mechanisms within the reach of the population not currently covered by the existing social security system (thepoor and the informal sector), while securing the financial underpinnings of the system. Choices used by different countries range from higher efJiciency of theformal sector system in order to improve the structure of incentives faced by workers, to supplemental programs and benefit packages for informal sector and non- working groups, to de-linking the system entirely from the labor market, moving towards a unijied system of universal coverage. Different policy avenues pose huge institutional andfiscal challenges that should bepart of thepolicy debate. This chapter identifiesfour priority issuesfor the Mexican social protection reform agenda which emergefrom our analysis, including: (i) the fiscal and institutional challenges of the current social security systems; (ii) inadequate health care and health insurance coverage for the poor; (iii) the lack of a strategy for addressing Mexico's high old-agepoverty rates; and (iv) the needfor an integrated approach to balance risk managementand chronicpoverty reduction. Thechapter closes with a discussion of key questions for thefuture. Most social protection systems in the region, including Mexico's, are dualistic, leaving a significant share of the poor population with insufficient access to health insurance, old-age pensions, and social assistance. At the center o f the system are social insurance programs comprised o f retirement and health plans for formal sector employees. Mexico i s facing a classic problem o f low-coverage and high-cost formal social insurance systems that are due to both the transitional costs of the IMSS' reforms and, more worryingly, the rising subsidies neededto pay benefits inthe as-yet unreformed pension systems for public sector workers, mainly ISSSTE. Choices on how to address this core issue mustbemade. The remainder o f the system i s comprised of social assistance programs, with insufficient resources, are fragmented, and characterized by gaps in coverage, missing, incomplete or overlapping interventions for different vulnerable groups outside o f the formal sector. The prevalence o f and the number o f extreme poor outside the reach o f Mexico's flagship Oportunidades anti-poverty program, twinned with the limited capacity o f that program to act as a risk management tool, all point to the prevalence o f important at-risk groups that might be better served by public policy. Oportunidades clearly follows best practice program design principles, and successklly targets and reaches many o f the poorest households. However, it falls short o f offering an integrated strategy for risk reduction among the chronic poor, and some key groups, notably the elderly, are left uncovered by this initiative. The central government has also taken the lead in beginning to address the high exposure o f the poor to catastrophic health shocks, as witnessed by the new Seguro Popular health insurance initiative, yet the question remains of whether the dualistic system should in fact be maintained. 54 PENSIONREFORM: ISSUES FOR THE FISCAL AND INSTITUTIONAL POLICY DEBATE The current pension systems require urgent reforms in order to assure financial stability. The particularly advantageous retirement benefits enjoyed by IMSS' own employees constitute a significant and rapidly increasing financial burden.43In the case of ISSSTE, the increase in federal transfers was driven by a growing gap between current contributions and pensions payments. In2003, ISSSTEhad a deficit o f 21.2 billion MX pesos, set to increase to 35 billion by 2006. For every peso o f income generated by the system, 2.5 pesos i s needed in a subsidy from the federal budget, a precarious and unsustainable financial stance.44The liabilities emanating from IMSS added up to 42 percent o f GDP in 2002 and will diminish in time. The liabilities arise from the transition cost o f the reforms and the annual cost will beginto decline in the next five years. The unreformed pension system in ISSSTE represents a contingent liability that now stand at around 45 percent o f GDP and continue to increase at a rate o f close to 2% of GDP a year.45 A country like Mexico may consider several broad avenues of policy reform, not necessarily exclusive from each other, and each one with different institutional and fiscal challenges that have to be part of the country's policy debate.A first avenue i s to increase the coverage o f the current system by improving it. A second avenue i s to shift toward financing from general revenues and de-linkingsocial protection from labor-market status. A third avenue i s to create new institutions to cover the currently excluded population. Increasing coverage through the first avenue requires a set o f reforms aimed at lifting the current obstacles to expanding coverage: improving value for the money in social protection services, reducing costs o f formality imposed by labor legislation rigidities, more effectively sanctioning tax evasion, and, at least for a segment of the population, increasing subsidies form general revenues. When workers and firms value the benefits attached to a formal contract less than the costs, the job will be informal. So improving service quality could increase the benefits o f formality and more workers will be willing to pay for formal social protection. Likewise, relaxing rigid labor legislations could reduce the costs o f accessing social protection, while unbundling health and pensions - which allows for a better alignment o f the system with workers' preferences - may also promote formalization. Inaddition, more effectively sanctioning non-compliance could encourage formality. These measures by themselves would lead to a gradual increase in coverage o f the formal social security institutes, financed by payroll taxes. This implies an enlargement o f the relative size o f the formal sector which will gradually, cover larger groups of the population. Still, despite the reform efforts that this option entails, additional efforts to improve tax collection would be needed in order to provide adequate social protection to the poor, at least during a transitional period. Another avenue, implemented in many OECD countries, implies de-linking social protection from labor-market status and financing the system from general revenues. In this system, coverage would be based on citizenship, not on the labor-market status o f individuals. Costs o f formality unrelated to social protection would no longer impact social protection 43 Insection3.3, we show that the pension of anex-IMSS employee ison average 8.3 times the size of that of any other affiliated worker -inspite of havingworked on average 12years less. 44 World Bank staff estimates presented to government on October 1,2004. 45 The numbers for IMSS come from Sphere Institute and Universidad Iberoamericana, 2004, referred to in "Social Protection reform in Mexico: Key Issues and Policy Options", World Bank Policy Memorandum , November 2004. The estimations for ISSSTE were made by the Ministry of Finance inMexico. 55 coverage. In fact, the distinction between a formal and informal worker would be irrelevant in terms o f social protection. The challenge in this case i s to improve revenue collection capacity and expand the tax base dramatically in order to be able to finance the system from general taxes only. By reducing payrolltaxes, this option would increase demand for labor and thus potentially create jobs and increase efficiency. A transition would have to be carefully planned, so as to provide incentives for workers to join contributory risk-pooling schemes and to incrementally foster the use o f general taxation to replace these contributions over time. The institutional, financial, and political challenges in terms o f reforming the tax system are large and would need to be debated nationally. The final avenue entails creating new programs to cover population groups currently excluded from the formal social security system, even if that implies maintaining a fragmented system. This option would avoid de-linking and may require finding the fiscal space to finance new expenditures. Any new program has to be carefully designed so as to be consistent with the current systems and would need to focus on two sometimes contradictory goals: expanding coverage and maintaining incentives for formality. Such a design, which would provide benefits to older people who did not have a chance to accumulate savings for their own pensions, will have non-contributory benefits, and should accomplish the anti-poverty goals without beingmore generous than the formal-sector benefits. BETTERCOVERAGE OF HEALTHRISKSFOR THE POOR The issue of how to offer better health access and financial protection for the poor i s at the forefront of the social insurance policy debate in Mexico. The introduction o f the Seguro Popular program raises two issues central to the policy debate. The first i s whether to operate a dualistic or `tiered' system which offers different categories or quality o f benefits to different populations based on their income level or employment status, or to provide a single, universal health care system. The second issue is, in the case o f a dualistic system, whether it i s optimal to have different components o f the system operated by differentpublic institutions (as i s currently the case in Mexico), or whether there are significant benefits to vertical integration. Again, countries have chosen very different approaches to the challenge o f extending health care to the poor. This section offers some examples from international experience that may be useful to consider inthe Mexican context: the case o f Thailand, which has opted for a transitional phase where the dual system is maintained while moving towards a universal system, and the case of Costa Rica, where vertical integration of the health insurance system has accompanied expanded coverage. "Financial Protection" as a health system goal. Financial protection can be defined as the instance in which all households are protected from falling into poverty as a result o f health costs -and where, likewise, no household i s prevented from moving out o f poverty due to spending on health care.46Achieving a higher degree of financial fairness -e.g., fairness of contributions to health costs- i s key to making health systems beneficial for the poor.47Mexico's new and promising initiatives in the health sector were inpart inspiredby findings inthe World Health Report (2000) in which Mexico had an overall ranking of 51 out o f 191 countries, yet ranked 144 on financial fairness, i.e. the fairness o f contributions to Since the ranking was 46"Beyond Survival: Protecting Households from the Impoverishing Cost of Health Shocks". World Bank Regional Study, forthcoming. 47Fairness is measured in terms of out-of-pocket expenditure as a percentage of disposable income. In a perfectly fair system the share of health expenses is equal across the income distribution. 48World Health Organization, 2000. "World Health Report 2000. 56 made, the Fox administration has launched a new health insurance scheme inwhich contribution varies with income and tends to zero for the poorest two deciles. Coverage has expanded rapidly, inparticular among the target population -the poor. What is fair financing? According to WHO an easy way to gauge this i s according to the share o f disposable income that households have to pay to improve or protect their health, and that high shares (e.g., >50 percent) are unfair. According to this logic, a system i s fair if all households pay the same proportion o f disposable income on health. The most important determinant o fhow fairly a health system i s financed i s the share o frepayment intotal spending. As noted by WHO (2000), out-of-pocket payment i s usually the most regressive way to pay for health care, and the way that most exposes people to catastrophic financial risk.Therefore, how revenues are collected has a great impact on the equity o f the system. Figure4 Poolingto redistribute risk, and cross subsidy for gre ter equity Contribuiim Hettranrfer Uiilbttion Podirq l n r m I A A Source: World Health Organization, 2000 How revenues are combined so as to share risks also matters: the argument in favor o f a single pool or a small number o f pools o f adequate size, and against fragmentation, concerns the financial viability o f pools, the administrative costs o f insurance, the balance between the economies o f scale and (when there i s little or no competition) the risks o f capture and unresponsiveness, and the limitation o f risk selection (which i s a matter o f efficiency as well as equity). Inefficiencies in collecting and pooling revenues reduce both the funds available for investmentand for providing services, andpeople's access to those services that can be financed. Purchasing, finally, also affects both equity and efficiency, by determining which investments are made and which interventions are bought, and for whom. Revenues may be collected fairly and with minimal waste, and be pooled so as to assure that the healthy help support the sick and the rich help support the poor. The performance o f the system will still fall short o f its potential if the pooled resources are not used intelligently to purchase the best attainable mixture o f actions to improve health and satisfy people's expectations. A second set of challenges involves the institutional designof health systems. A high degree o f vertical segmentation characterizes health systems in most Latin American countries, including Mexico. Different organizations such as the Ministry of Health, social security funds, the armed forces, charitable organizations or the private sector may pay their own providers, raise and allocate funds and provide services for differentpopulations, leading to higher administrative 57 costs and inefficiencies than may be optimal. At the other end of the spectrum are health systems inwhich one single organization raises, pools and allocates funds to a fairly monolithic group of independent service providers. Norway operates a health system similar to this. Some countries with multiple health insurance (social security) organizations have introduced central collecting agencies in charge o f risk equalization between population pools (Colombia, Germany and the Netherlands). International experiences show how the optimal institutional design differs with a country's socio-economic characteristics and its existing institutional framework. In addition, each system designcarries its advantages and disadvantages. While central control o f funds inthe case o f Norway makes sense, there i s still scope for improved efficiency through better regional cooperation and planning. Key challenges inMexico also include the incentives (or disincentives) to formality created by the existing dual health system (as well as changes to the incentive structure that would be brought about by further reform), federal-state relations, single vs. multiple insurers and providers, the question of portability of benefits across systems and publidprivate providers, and lastly, the incidence o f subsidies. Many developing countries are facing challenges similar to Mexico, with some countries lendingparticularly useful lessons from their efforts to address these challenges via reform. Box 4.1 highlights international experience with reforming health and social security systems. As seen in the four cases discussed below, countries have approached the issue o f expanding health insurance coverage differently. In addition, the four cases all show how vertical integration o f health systems has accompanied or followed reformto expand coverage. Box 4.1: International experience with extending coverage of health services and insurance Costa Rica: One-step universalization of coverage, followed by the creation of a single risk pool. However, Costa Rica's universalization may have come at the expense o f fiscal sustainability. Health services and social security coverage was universalized in Costa Rica in 1971. Affiliation to the Caja Costarricense del Seguro Social (CCSS) i s mandatory for formal sector employees while self-employed and informal workers are encouraged to join voluntary plans with different contribution structures depending on income. Voluntary affiliation i s high and CCSS today covers 89 percent o f the total population. The one-step universalization was complemented by an incremental vertical integration o f functions to address inefficiencies and overlaps in a fragmented health system. The mandate to provide health services, together with MinistryofHealth facilities and employees, was transferred to the CCSS, which today operates as a national health insurance and pension fund, and administrator o f the public health infiastructure. CCSS now has a legal mandate to extend its coverage to 100 percent o f population, regardless o f nationality or employment status. Thailand: Building on a dualistic system in the transition to universal health care provision. A heavily subsidized insurance scheme for the poor, children and elderlywas launched in 1975, andhas since beengradually expanded. After adding two compulsory insurance schemes for public and private employees, and voluntary insurance schemes including private health insurance, over 80 percent o f the population was covered in 1998. ARer a series o f studies, the Thai government decided that the best way to move towards universal health insurance coverage was to start with a dual system for formal and informal sector before moving to a single-payer, national health insurance in the future. In 2001, the government launched a new health insurance policy, "30 bhat treat all", to target the 20 percent o f the population not yet covered by health insurance. For incidents other than accidents and emergencies, the insuredmustpay a contribution o f 30 bhat per period. With this contribution, accessible health services are practically the same as for the other insurance schemes. After one year, 96.6 percent o f the target population was registered under the new scheme. The Republic of Korea: Gradual extension of compulsory pension and health care coverage in the 1980s, followed by vertical integration in the late 1990s. In the early 1980s, 58 pilot programs to cover the rural and urban poor were introduced and these were generalized by the end of the decade to cover all rural and urban areas. Likewise, pension coverage has been extended incrementally. First, self-employed workers in rural areas were covered, with the government assuming two thirds o f the 9 percent contribution, a subsidy that was phased out over 10 years. Later, self-employed workers in urban areas were covered on the same basis. After expanding coverage via this segmented approach, Korea pursued an integrated system. In 1998, the health insurance schemes for the self-employed (92 in rural and 135 in urban areas) merged with the public insurance program for government and school employees. In 2000, the 142 schemes covering industrial workers merged into the National Health Insurance Corporation, which became the single insurer. This resulted in all affiliates receiving the same statutory benefit package and all health care providers beingreimbursedaccording to a standard schedule. In2001, the schemes of government and school employees merged with that of industrial workers; however, the incorporation o f the fund o f the self-employed was delayed to 2004. Taiwan: Fromfragmented to universal coverage with a mandatory health insurance scheme. In 1995, when Taiwan introduced legislation to create a mandatory national health insurance scheme, only half o f the population was covered by a social security scheme. At first, the process seemed extremely rapidgiventhe fact that one year later 92 percent ofthe population was covered. However, the process inreality started more than a decade before in the Council for Economic Planning and Development (CEPD). The first planning stage took two years o f studies and the original proposal included a project to phase in the nationwide insurance program progressively until reaching universal coverage by the year 2000. The first pilot project for the expansion started with well-organized farmers groups in 1987. Political events in the first half o f the 1990s created a strong political incentive to give priority to the fast expansion of social security to the whole population. A careful analysis o f the pilot projects and the lessons leamed from the farmers' experiences and studies on trends o fhealth expenditures allowed introducing the legislation in 1995. Sources: Angela Lisulo (2003), Nutta Shreshthaputra and Kaemthong Indaratna (200l ) , "Social Protection inMexico: Key Issues and Policy Options", World Bank Policy Memorandum (2004). PROVIDING OLD-AGEINCOME SECURITY FOR THE POOR The future of old age security inMexico will dependon two key sets of decisions and actions: (i) focused reform and rationalization o f the existing social security system (as discussed in section 4.1), and (ii) tackling Mexico's old-age poverty problem. Mexico's high rates o f poverty among the elderly given its level o f development, combined with the progressive aging o f the Mexican population under current demographic trends, highlight the importance o f developing a pro-active approach to reducing the risk o f poverty in old age as an integral part o f Mexico's social protection system. Options used by different countries are related to the implementation o f targetednon-contributory pension. Inwhat follows there i s a discussion o f the pros and cons o f these systems andthe experience o f several countries. Pension systems, both public and private, are the primary means of providing financial security to the elderly and to those unable to work due to disability. Because pensions are designedto replace income previously earned, they usually apply only to those inthe labor force. Because they are often funded by payroll taxes, they also have implications for labor costs and employment. The three main goals of pension systems are: (i) to provide a savings vehicle (to allow individuals to redistribute across their own lifetime to avoid poverty in old age); (ii)providing a social safety net for those who become unable to work (insurance); and (iii) to redistributefrom the rich to the poor to preventpoverty inold age. When the challenge is to expand pension coverage, if more than one program or mechanismis devised, they have to be designed as part of one integral system. For example, 59 Non-contributory systems may reduce incentives to formalization for both businesses and workers. Consequently, when using these systems it i s essential to structure benefits for subsidized pensions so that only the poor want them, i.e., a minimum pension or a very basic package o f health services. Modest benefits thus prompt a form o f self-targeting mechanism. If a focus i s kept on contributions, the informal workforce that generally includes the poorer segments o f the population remains uncovered. Most systems operate with a combination o f contributory social insurance and non-contributory social assistance structured around minimum benefits. Giving more weight to the latter can increase coverage but requires a country to generate the fiscal space needed to increase current expenditures permanently. Box 4.2 provides a summary discussion of the pros and cons o f implementingtargeted, non-contributory benefit plans for the elderly. Box 4.2: Pros and Cons of introducinga targeted, non-contributorybenefit program intothe social protectionsystemincludethe following: The major strengthso fintroducing a targeted non-contributory benefitprogram include: (i)Covering the risk o f poverty among the elderlywith a system that is "blind" to labor market history; and (ii) "Erasing" the distinction between "formal" and "informal" at least with respect to poverty in old age. The major weaknesses o f introducing a targeted non-contributory benefit program are: (i) system,dependingonthebenefitlevelsitprovides,canintroducedisincentivesforpeople The to participate in and contribute to the contributory system, which they need to do to help them smooth consumption, and which i s desirable because the more participants, the better the system works (this i s true whether public pooling or private individual savings); and (ii) costsFiscal could quickly get out of hand since at least one of the eligibility criteria (old age) will be met by most people; this is no small "con" and there are plenty of examples worldwide where, because of poor targeting, the non-contributory component ends up creating enormous fiscal outlays. Inwhat ways canthese weaknessesbe addressed? The incentive issue canbe addressedbymaking the non-contributorybenefitmodest. (theChilean benefitis 30 percent oftheminimumwage) Fiscal and incentive issues can be addressed by adopting an "insurance" concept o f coverage, i.e., only those that suffer the bad state (poverty in old age) actually get the benefit, but all are "covered" (just as in any insurance program, participants are covered though the risk o f the eventuality occurring i s small) Fiscal problems can also be minimized by making the non-contributory benefit taxable, along with all other sources o f income (this also encourages participation in thetax system -to receive the benefit, one mustfile atax claim). Costs can also be minimized -particularly targeting costs - by piggybacking on an existing, well-targeted poverty program, like Oportunidades.Chile has done this by designing a special version o fthePuente program (see below) for elderlyhouseholds Large savings can arise from creating time-flexible parameters around age criteria for eligibility, as life-expectancy changes, both for the contributory and the non- contributory systems. Many fiscal problems arise simply because old-age benefit programs are not designed at their inception to take into account the reality o f changes inlife expectancy over time. 60 Non-contributory,universal benefits are simple to administer and may be effective in combating old-age poverty but impose high fiscal burdens. As pointed out by Packard (2004), a universal benefit system does away with the entire covered/uncovered debate, as it by definition covers the entire population. Universal benefits have many advantages over both orthodox, contributory systems and non-contributory systems targeted to the poor - they are the simplest public poverty-prevention mechanism to administer, with the lowest transaction costs.49 At the same time, universal benefits can be extremely costly and are viewedby many as a luxury only available to high-income nations such as New Zealand (see section 4.5, Box 4.5). While many countries, includingDenmark and the Netherlands, have universal elements to their pension systems, New Zealand appears to be the only one where all state provision i s made through a universal flat pension. With its current low capacity to raise general revenue, it i s not clear how Mexico could sustain the financial burden o f universal pension benefits. Non-contributoryand/or minimumpension guarantees targeted to the poor through means-testing can provide a more efficient use of funds by funneling scarce resourcesto the elderly poor, but are more challengingin terms of efficient administration.Non-contributory and/or minimum pension guarantees are often more akin to social assistance than insurance, and have been implemented in many Latin American countries (see Table 4.1). Minimum pension guarantees can be nested within existing contributory social security systems as seen in the Chilean example below or in the Mexican experience with IMSS, and typically target poor workers whose low contributions have resulted in an income after retirement that falls short of some pre-definedminimumlevel. The gap betweenthe actual pensions received based on earlier contributions and the minimum level o f income i s filled by a social assistance cash transfer - financed by general tax revenue. Non-contributory pensions often function outside the formal social security system; others are nested in existing formal pensions systems, such as Brazil's ruralpensionprogram (Box 4.3). Table 4.1. Expenditureand Coverage ofNon-ContributoryPensionPrograms Country Expendit Expenditure as Expenditure as Percentage Coverage: ure as percentageof percentageof financed Old Age percenta total expenditure total Social from Beneficiariesas ge of on Social Sector general percentageof GDP Security Spending revenues Elderly Poor Argentina 0.2% 4% 1% 100% 47" Brazil 0.3% 5% 2% 100% 86 Chile 0.4% 6% 2% 92% 37" and 79" CostaRica 0.3% 7% 2% 48% 45d Uruguay 0.6% 6% 3% 100% 17 and 12" Notes: a. Share o f indigent aged 65 and over, EncuestaNacional de Desarrollo Social 1998 b. Share o f recipients 67 and over with per capita income less than % minimumwage receiving a pension (assistance or rural), 1999 c. Share o f recipients 65 and over in deciles 1 and 2, urban and m a l areas, 2000 d. Share o fpoor aged 65 and over, 2000 Source: ILO (2002), from Gill et al. (2004) 49Willmore (2000). 61 Box 4.3: Latin-American Experience with Non-Contributory Pensions and MinimumPensionGuarantees Extending pension coverage to the poor when informality is high: Brazil's experiencewith non-contributory pensionprograms. Brazil has achieved remarkable declines in old-age poverty, largely due to two non- contributory pension programs with uniform benefits. A rural scheme covers about 7 millionwith benefits, including pensions for old age, survivors and invalidity, as well as benefits for maternity and occupational accidents. Benefits are calculated based on the minimumwage, while eligibility i s based on length o f service in agriculture, fisheries etc. The BeneJicio de PrestaqBo Continuada provides a pension o f one minimum salary paid to the elderly and disabled in households with per capita family income less than one-fourth the minimum salary. The BeneJicio de PrestaqBo Continuada is the largest spending category within social assistance inBrazil and accounted for 45 percent o f social assistance spending in2002. Estimates o f effective poverty reduction do not differentiate between the different types o f pensions; however, when looking at the impact o f all pensions on beneficiaries, destitution i s reduced by about 96 percent and poverty by 29 percent. While the programs strongly reduce poverty, they also contribute to Brazil's fragile fiscal situation Chile: Reducing poverty with a combination of targeted non-contributory pensions and minimumpensionguarantees for affiliates of public social security The Chilean pension system combines a minimum pension guarantee (MPG) for registered workers with a social assistance pension (PASIS) targeted to poor, informal workers. The MPG is available for all insuredworkers that have reached a contribution threshold o f 20 years, as long as the sum o f pensions, income and remuneration is equal to less than the minimum pension. The workers have earned the rightto an annuity initially financed out o fthe funds intheir individual accounts. When these funds are exhausted, the shortfall is financed with a transfer from the government. Both the MPG and the PASIS are financed fiom general revenue, primarily VAT." Coverage o f PASIS increased significantly in the period 1990-2000, and statistics show that the social assistance pension also increasingly reduced poverty over the decade.51The effect i s strongest amongst the extreme poor, where by 2000 poverty had fallen by 69 percent in households benefitingfrom the PASIS. Finally, the program also became more progressive, with beneficiaries from the first quintile growing from 35 percent in 1990 to 53 percent in2000. Sources: Bertranou2002, Truman Packard et a1(20041, Gill et al. (20041. What are the key linkages and complementarities between contributory and non- contributorypensions?The growing importance of non-contributory pensions will in the long run lead to an unsustainable financial situation for the overall pension system. As such, tax- financed benefits should not replace contributory pension schemes. Along the same lines, in a financially sustainable pensions system, contributory schemes needto cover more than a minority o f the population. Improving the efficiency o f collection o f contributions remains crucial. Along the same lines, non-contributory pension benefits must be structured in a way so as not to provide disincentives to pay contributions. This has shown to be a particular concern when non- 50 Arenas de Mesa, Albert0 and Fabio Bertranou, 1997. "Learning from Social Security Reforms: Two different cases, Chile and Argentina", Washington DCWorld Bank. Bertranou, Fabio, 2002. "Filling the Protection Gap: The Role of MinimumPensions and Welfare Benefits", Paper presented for ISSA Seminar on financial and actuarial bases of pension schemes. 62 contributory pensions are combined with a minimum benefit component for contributory programs, as workers can continue to evade contribution, and still be guaranteed o f receiving the same minimumamount. The ability to operate non-contributory schemes hinges on the ability to finance them via general tax resources, which explains why they are most widely used in developed countries. Most developing and middle-income countries struggle with thin tax bases and problematic tax collection in general. In addition, scarce revenues lead to a range o f competing priorities, and the sustainability o f a universal scheme i s easily questioned from one year to another inresponse to changing economic and political conditions. For instance, the gross cost o f the New Zealand universal pension scheme represented 5 percent of GDP in 2001, and i s estimated to rise to about 10 percent o f GDP by 2051. Reforming social insurance systems with a view to expand coverage inmost cases entails givingmore weight to non-contributory pensions. STRIKING THE RIGHT BALANCEBETWEENADDRESSING VULNERABILITY AND COMBATINGCHRONICPOVERTY The priority reforms discussed above - reforming the social security systems for public sector workers to restore fiscal soundness, extending health insurance coverage to the poor, and addressing old-age poverty - are all crucial steps toward a more effective and equitable social protection system. Yet, these changes alone will likely be insufficient in creating a cohesive, comprehensive SP strategy that can successklly address the needs o f the most vulnerable in society the chronic poor. - From international evidence we know that the chronic poor are equally, if not more, prone to experiencing income shocks. At the same time they are the least endowed with risk management tools to respond effectively to those shocks in the absence o f effective public policies. As a result, transitory movements in income are more frequent among the chronic poor, recovery periods tend to be longer, while movements out o f poverty are often short-range, with a highprobability of recidivism. A comprehensive social protection strategy will include both the traditional risk management mechanisms and targetedpolicies designedto address the heightened vulnerability and thus, persistent poverty o f the chronic poor. First, coverage o f traditional risk management instruments need to be extended. Second, targeted redistributive transfers need to address the income vulnerability and lower physical and human endowments among the chronic poor such policies would involve. Ideally, all components o f the social protection system- whether aimedprimarily at riskmanagement or primarily at redistribution andpoverty reduction- need to be linked and form an integrated approach consistent with the nation's broader social protection strategy. Chile's experience in the 1990s offers an example o f how traditional social policies, regardless o f how aggressive, can fail in reaching a hard core o f chronic poor. To meet the challenge posed by this particularly resilient concentration of poverty, Chile switched toward a new and innovative poverty eradication strategy focused on integrated, family and community- centered interventions (Box 4.4). How could Mexico proceed to strike the right balance between interventions addressing short-term vulnerability to risks and programs aimed at alleviating long-term structural, poverty? To meet this challenge, traditional social insurance mechanisms will need to be linked to other components o f the social protection system, particularly assistance components. The entire system will also need horizontal coordination and design features that promote dynamism, flexibility and positive incentives to move up and out o f the assistance 63 system into the traditional social insurance network and labor market. Finally, designfeatures that promote coordination across programs in response to the multi-dimensional needs o f the poorest families are fundamental. Chile Puente i s designed for exactly this purpose. As shown above, it promotes tailored bundles o f programs from the Chile Solidario strategy to answer the specific needs o f particularly hard to reach households, providing a `bridge' for these neediest families into the social insurance network. The past decade has seen impressive efforts in designing asset generating and multidimensionalsocial assistanceinterventionsto combat extreme poverty in Mexico (Le., Oportunidades), yet these programs do not constitute a perfect substitute for formal risk pooling mechanisms, nor do they represent a concerted move toward a more cohesive overall social protectionsystem. Conditional cash transfer programs facilitate risk management insome areas, e.g., by loweringthe risk of school non-attendance and protecting consumption in times o f crises, however they often do not correlate with other risks frequently faced by households, particularly adult household members, such as job loss, natural disasters or health problems (Rawlings 2004).52Recent evidence suggests that while Oportunidadesenables income smoothing over time, the program does not appear to provide any additional insurance or protection benefit over and above that o f existing programs and informal mechanisms (Skoufias, 2004). In addition, Oportunidadeshas worked largely in isolation from other social assistance or insurance initiatives. A key challenge here i s to build links between Oportunidades and other programs and to structure the set of programs in such a way that incentives for moving up and out o f social assistance are built into the programs. Box 4.4: Chile's Puente Program Despite steady declines in poverty over the last two decades in Chile, the share o f the population in extreme poverty has remained virtually constant at just above 4.2 percent. The existence o f this hard core o f chronic poverty seemingly unresponsive to existing policies and programs, spurred the formulation o f a new poverty reduction strategy, Chile Solidario. The proactive approach o f the strategy and its most prominent component-Chile Puente-differ substantially from the norm for social programs inthe region. Puente acts as an entry point to the social protection system for the 225,000 households targeted by the Chile Solidario initiative. It combines a temporary cash transfer with intensive family-based psychosocial support and is based on proactivity, a strong family focus, and a coordinated response. Apoyos Familiares (AF) follow assigned households for 24 months, providing psychosocial support as the family seeks to improve its living conditions. The characteristics o f Puente beneficiaries largely coincide with the characteristics o f structural or chronic poor in Chile. The four principal components o fthe program are: Psychosocial support to the household via the relation that the AF establishes with each family assigned to them inthe homes o f these families; Training and education of the AF, based on developing knowledge of the benefits and characteristics o f the Program and to transfer a methodology for working with the families; A regionalfund for initiatives made up o f private and public sources to finance services or benefits currently not in place, that are required by the families to reach one or more o f the defined minimumconditions ofwell-being; and s2 Rawlings, Laura B., 2004. "A New Approach to Social Assistance: Latin America's Experience with Conditional Cash Transfer Programs", The World Bank. 64 Monitoring and evaluation: An online system used by the AF gathers all family information that emerges from the work sessions. Based on this system, the advancement o f the program can then be viewedin each municipality, province, and region, or at the country level at whole. The system also provides statistics on main characteristics o fparticipating families, achievements o f minimum conditions after intervention, and the mobilization o fresources in favor o fthe families. Sources: MIDEPLAN (2003), Secretaria Ejecutiva Sistema Chile Solidario (2003); Packard (2004). One way to improve coordination i s to tackle existing overlap among agencies providing benefits and services, particularly in social assistance. Social assistance in Mexico i s dominated by the flagship program Oportunidades. A few o f other programs have taken the same integrative approach, including Hhbitat and Microrregiones. However, the efficiency o f operating a multitude o f other programs i s questionable especially with regard to potential overlap and duplication o f efforts. The effectiveness i s also uncertain due to the fact that many of these smaller programs lack rigorous monitoring and evaluation mechanisms, and are so small that in fact extensive impact evaluation would not be cost effective. These concerns have spurred debate on possible rationalization and consolidation, especially when put in the context o f the more integrated and dynamic approach to assistance envisaged by the Contigo strategy. Indeed, one o f the transversal issues emphasized in the Contigo strategy i s the need for integration, i.e., the need to encourage efficient social spending by exploiting synergies, eliminating duplication and ensuring coherence in different program objectives. In general, poor implementation o f the principles applied in Oportunidades and conceptualized in the Contigo strategy i s a problem across the social assistance spectrum. How can Mexico build upon its established success in Oportunidadesto bringreform to other areas o f the system? The strengtheningand expansion o f good programs such as Oportunidadesi s a positive trend and in many aspects a direct result of rigorous monitoring and evaluation. The continued existence and resilience to change o f the not- so-good programs i s a problem, and often linked to the lack o f standardized monitoring and evaluation. Finally, the Contigo strategy itself i s being challenged on different fronts. By focusing on responding to changing needs and risks across the lifecycle, important dimensions o f risk and vulnerability may be lost, such as spatial differences, ethnicity and the distinction between chronic and transient poor. Addressing the needs o f the chronic poor likely requires a greater emphasis on redistribution via social protection policies, more than what the current social assistance framework and expenditure envelope allows for. Finally, beyond SEDESOL, from which the Contigo framework emerged, there i s less ownership and therefore acceptance o f the basic principles, which could in part explain why the framework has not been operationalized more extensively. A second challenge is posed by the insufficient linkages and coordination both between programs within the social protection systems (internal coherence) and between social protection and policies in other sectors of the economy (external coherence). Coherence and coordination also affect how dynamically the system i s able to perform. This debate can be illustrated by taking the example o f Oportunidades and its linkages to other programs and sectors. The program has been very successkl in its impact on the health and education outcomes for extreme poor infants and youths in rural areas. The poverty alleviation impact i s however determined by whether increased human capacity translates into a higher earning capacity, which again depends on interactions between Oportunidadesgraduates and the labor market and/or the agricultural sector. While the return to human capital i s determined by other sectors o f the economy, graduates are still in need of insurance and assistance. Ifgraduates enter into the informal sector, having no health insurance makes them as vulnerable to catastrophic health shocks as the next person. At the same time, there i s no reason why poor 65 mothers should be in less need o f health benefits after their children graduate from the program. Jbvenes con Oportunidades i s an attempt to build dynamic linkages that addresses some o f these issues. First, it creates bridges to other programs inthe social protection system. For instance, the savings earned and received upon graduation can be used to buy health insurance (Seguro Popular) or to enter into the government's housing credit program (Tu Casa).A dynamic link to other economic sectors i s created by supplementingthe human capital o f graduates with access to credit markets, as the savings can also be used as collateral for micro-credit or as a down payment on a house. Jbvenes con Oportunidades i s a nascent program and its impact has not yet been formally evaluated. However the approach taken illustrates how linkages between sectors and programs can be usedto create incentives and assistance for beneficiaries to move `up and out' o f the system. Clarity of objectives and a transparent operation of programs are keys to a more coherent and balanced social protection strategy. In order to build a more cohesive social protection strategy, the objectives o f individual programs need to be clearly stated to facilitate evaluation, avoid possible duplication o f efforts, and assist overall coordination. Finally, to reach special groups and to increase the responsiveness of the social protection system to particularly excluded groups, countries have tailored their social protection delivery systems to meet `customer' needs rather than mimicking government structures.Australia and Canada are among the leading countries in terms o f customer-centered delivery o f social services (see Box 4.7). The principle for both countries' delivery systems is to provide services along the lifecycle, rather than on the basis of governmental structures. Their experience also shows that Information Technology (IT) offers great opportunities to support the social protection system, in particular in terms o f increasing responsiveness through a restructured delivery system. Again, IT systems used in social protection have been focused on the requirements and expectations o f citizens rather than on the organizational structures o f government institutions. For customer-centered solutions to be successful, stove-piped operations are often turned into a single-point-of-contact system (one-stop shop) that allows integrated delivery o f social services via physical or electronic channels. Chile Puente also takes a one-stop shop approach by providing the extreme poor with a single entry point to the programs and benefits o f the country's social protection system. The Chilean program pushes the customer- centric approach one step further by actively seeking out the hard-to-reach, extreme poor, in a sense `recruiting' them for the program. Structural determinantsof chronic poverty, such as discrimination and high levels of inequality, call for a broader set of policy interventions. While a mix o f risk pooling mechanisms and redistributivetransfers i s necessary to address the problem o f chronic poverty, it might not be sufficient in dealing with social and economic structures that obstruct escape from poverty for certain groups or individuals. Tackling chronic poverty requires a multidimensional, integrated response, beyondthe narrow scope o f social protection. Box 4.5: Australia's Centrelink: A governmentagency set up to provide a "One-stop shop" for a range of governmentservices Centrelink i s a public service provider, established to produce more efficient and streamlined delivery of government services. The agency offers a one-stop shop for a range of services on behalf of 25 government agencies including ten ministries (departments). M a i n features of this one-stop shop are: 66 A customer-centered organizational structure: The organization o f Centrelink services is based on the requirements and expectations o f citizens rather than on the organizational structures o f government institutions. Customer service networks include customer service centers, call centers, and other specialist customer service outlets and community agents. Using information technology: Centrelink's comprehensive website operates as an information hub for all the different governmental services and programs. Customers can access and update personal information, including income statements. They can also estimate eligibility for different programs as well as file requests for certain services. In addition to the website, Centrelink Call comprises 27 networked call centers located around Australia, and automated systems are widely used. 0 Social justice and equity in access and service delivery: Centrelink uses a range o f measures to ensure that customers have equitable access to products and services, and that people are not disadvantaged by service delivery arrangements. Department o f Immigration, Multicultural and Indigenous Affairs customers review the center each year to ensure that customers receive fair and equitable treatment. Source: www.centrelink.gov.au FINAL REMARKS To summarize, this chapter has focused on four key emerging issues within social protection in Mexico: (i) social security reform, (ii) risks among the poor, (iii) health old-age poverty, and (iv) the need to balance public risk management mechanisms with a more integrated strategy to combat chronic poverty. Inorder to tackle these challenges, Mexico faces arange ofpolitical choices. Firstly,how much redistribution? As discussed below, the role of social protection policy in promoting redistribution i s not clear-cut. While largely a question o f national preference, the answer also hinges on the ability to raise tax revenues and the political will to redirect subsidies currently benefiting less needy segments. A second set o f questions emerge in relation to the choice o f institutional design. The current system based on a payroll-financed social security system for formal sector employees may be reformed in order to increase its efficiency. Expanding coverage requires, among other, improving the value for money, improving the quality o f services. Raising the quality in service provision i s critical to achieving better outcomes, yet requires a political commitment to altering accountability relationships between different levels o f government, service providers and the ultimate beneficiaries. It also requires complementary reforms to labor market legislation as to reduce the costs o f formality. Also, even if it maintains a dualistic structure, it may be complemented by targeted social assistance to the very poor, strengthening the latter. Alternatively, some countries move to a single benefit system with universal -at least minimum- coverage, de-linkedsocial protection benefits from the labor market. Implicitly, this alternative also questions Mexico's ability to undertakethe tax reforms neededto support a model funded bygeneral revenues. How much redistribution? How much will Mexico seek to use social protection policy as a vehicle for redistribution? As discussed in Chapter 1, the state has a clear role in providing risk-pooling mechanisms where private insurance markets fail (e.g., old age and health insurance), but the role o f social protection policy in promoting redistribution i s more an issue o f national choice. That choice, however, has clear consequences for equity and poverty reduction outcomes. The 67 economic trade-off betweengrowth and equity, and thus between growth and redistribution, has been refuted by both theoretical advances and international experience in recent years. At the same time, redistribution i s boundto result in some economic efficiency trade-off, as exemplified in the design of social security systems where redistribution typically creates labor market distortions (Orszag and Stiglitz, 2000). For instance, the change from a defined benefit system where everybody gets the same benefits regardless o f contribution, to a defined contribution system where benefit levels are linkedto contributions, enhances incentives to participate. At the same time it reduces the progressiveness o f the system. Indeed, international experience points to examples where redistribution has been achieved in a wide variety o f ways. It was done via land reform and education in Korea, via affirmative action in Malaysia or via aggressive social policies in Chile, to name but a few examples. The choice i s therefore two-fold: how much distribution; and through which means? The relative importance given to redistribution and the role assigned to the social protection system in achieving equity objectives i s a question o f national objectives, but i s crucial to guidingthe direction o f social protection policy reform. Should risks be bundled? While the Seguro Popular and Seguro de Saludpara la Familia are designed to extend health insurance to the uncovered, this population i s also inneed o f protection against other risks, in particular related to old age. Despite national recognition of the need for extended pension coverage, the country faces several challenges as to how to do this, i.e., whether increase the efficiency o f the current social security system, use targeted social assistance pensions or move towards universal pensions. However, Seguro Popular sets a precedent and critical issues o f institutional and operational design will have to be addressed before moving towards more coherent and integrated social insurance. For instance, the institutional design o f Seguro Popular suggests a continued dualistic system; however, how would a pension's scheme for the uncovered fit in this picture? Should old-age insurance be bundledtogether with the Seguro Popular, thus reinforcing the dualistic character o f the system? Ifbundledwith health insurance, would it make sense for the ministry o f health to maintain stewardship? If not, i s there a need for a new, fully- fledged social security institute operating alongside IMSS? And finally, should operation and provision o f the pension benefits be decentralized to states as has been the case for the Seguro Popular? Answers to these questions depend on political factors and an institutional context particular to Mexico. However, it i s clear that the current fiscal situation limits the range o f feasible options and that fiscal reform will be a prerequisite for the establishment o f a truly integrated social insurance system. Should socialprotection be de-linkedfrom the labor market? A review of international experience shows that countries choose to do this very differently, but few (infact, only N e w Zealand) actually go all the way infully de-linking social insurance schemes from the labor market. Far more commonly, countries vary in how far they choose to go in de-linking the system, that is, how they mix and match contributory and non-contributory schemes, and whether the system design is dualistic (with separate providers for formal and informal sector) or universal (national insurer that covers both sectors). Evidence indicates that if the aim is to improve low coverage of social insurance among the poor, relatively more emphasis needs to be placed on developing non-contributory schemes. However, many countries have contributory programs for informal workers and extending coverage to the informal sector - poor and non-poor - often implies tailored schemes that accommodate the different needs and capacity to contribute of distinct population groups. 68 s s 2 i od vi s s 2 c: i W s s 9 '" i i W s s i i i 0 W s s 09 c? i r- m s s c: i 2e s s "! z i e s s ? '9 i i vi 2 3 0 0 i i FL I s c? i s '" i s c? i s '9 i s 2 s '" i s 2 s 2 W m m U 0 0 0 0 0 0 U I s rd r- s 2 i F hl s x s 9 m i s 9 s vi c? m i s 2m s c? i s 9 s vi c? i i s ? s m i hli s 2 s 9 hli s 2 s 9 hli ne s : s s s "! '9 9 hl m vi s s s s 09 c: '9 c: i i hl m s :: 2 s s s '4 i + s s s s F 0, 0, 0, i i i 2 n n i sI 9 9 i v1 v1 WI 2 W W W W i i i i s3 s3 s3 s3 a a a E E E aE +- +- +- +- +- 0 +- 0 +- 0 +- 0 #a #a #a #a 2 2 2 2 0 i i i i 3 F 2 F2 F 2 F2 v1 v1 ew 3 20 20 20 20 o 0 0 0 0 ij+ 4 s "! vi i s 2 hl s '" m i s i d: hl s 9 vi s i w s ? i i s : i W9v1 is3 aE +- +- 0 #a 2 i F2 z2 z2 0 0 0 0 m d N m 0 0 d 0 1 5 0 0 0 0 N N N N h I Bibliography Adato, Michelle (2000). The Impact of Progresa on Community Social Relationships. - Washington, D.C. : International Food PolicyResearchInstitute. Addison, Tony and JorgenLevin (2004). Tax PolicyReform inDeveloping Countries.- Sweden : UNU/WIDER :Helsinki, Finland and Orebro University, (mimeo). Albarran, Pedro and Orazio Attanasio (2002). Do Public Transfers Crowd out Private Investment? : Evidence from Randomized Experiment inMexico. - World Institute for Development Economic Research :UnitedNations University. Arenas de Mesa, Albert0 and Fabio Bertranou (1997). Learning from Social Security Reforms : Two Different Cases, Chile andArgentina. -Washington, D.C : The World Bank. Ariza, Marina and Rolandia de Oliveira (2004). Familias, Pobreza y Necesidades de Politicas Publicas en MCxico y en CentroamCrica. - Santiago de Chile : CEPAL. -Conferencia Reunion de Expertos (October 28-29 ). Arriagada, Ana-Maria and Gillette Hall (2001). "Social Protection : Managing Social Risk in Mexico". -- In Mexico : A Comprehensive Development Agenda for the New Era. -- Washington D.C. : The World Bank. pp. 479-568. - Barrientos, Armando, David Hulme and Andrew Sheperd (2005). "Can Social Protection Tackle Chronic Poverty". -In:Background Paper for the Social Protection for Chronic Povertv Conference.-UnitedKingdom :University of Manchester,(Forthcoming). Baulch, Bob and John Hoddinott (2000). "Economic Mobility and Poverty Dynamics in Developing Countries". - In:Journal ofDevelopment Studies 36(6): 1-24. Behrman, Jere, Alejandro Gaviria andMiguel SzCkely (2001). Intergenerational Mobility inLatin America. - Washington D.C : Inter-American Development Bank. Bertranou, Fabio (2002). "Filling the Protection Gap : Non-Contributory and Social Assistance Pensions inArgentina, Brazil, Chile, Costa Rica and Uruguay". - In: Background Paper for International Social Securitv Association Seminar (Santiago de Chile November 21- 22) Binder,Melissa and Christopher Woodruff (2002). "Inequality and Intergenerational Mobility in Schooling : The Case of Mexico". - In: Economic Development and Cultural Change (50): 249-267. Brewer, Mike (2000). Comparing In-work Benefits and Financial Work Incentives for Low- income Families in the U S and the UK. London : Institute for Financial Studies. - - (Working PaperNo. 00/16). Tom Clark and Michael Myck (2001). Credit Where It's Due?: An Assessment of the New Tax Credits. - London : The Institute of Fiscal Studies London. - (IFS Commentaries No. (286). Castaiieda, Tarsicio and Enrique Aldaz-Carroll (1999). The Intergenerational Transmission of Poverty : Some Causes and Policy Implications. - Washington, D.C. : Inter-American Development Bank. (Discussion Paper). - Centrelink (2003). Annual Report :2002-03 www.centrelink.gov.au 76 Cercone, James (2001). "Reform of the Mexican Health System" In: Mexico : A Comprehensive Development Agenda for the New Era. Washington, D.C : The World Bank. pp. 409- - - 446. Chaabane, Mohamed (2002). Towards the Universalization of Social Security : The Experience of Tunisia. - Geneva : International Labor Organization. (Working PaperNo. 4). - Chronic Poverty Research Centre (2004). The Global Chronic Poverty Report 2004-05. -- United Kingdon : The Chronic Poverty ResearchCentre. Coady, David, Margaret Grosh and John Hoddinott (2003). Targeting Outcomes Redux. - Washington, D.C : The World Bank :IFPRI. Cox, Donald and Emmanuel Jimenez (1990). "Achieveing Social Objectives Through Private Transfers :A Review". - In:The World Bank ResearchObserver (5)2 Julv: 205-218. Courchene, Thomas, Alberto Diaz-Cayeros, and Steven Webb (2000). "Historical Forces : Gerographical and Political". - In: Achievements and Challenges of Fiscal Decentralization : Lessons from Mexico. - Washington, D.C. : The World Bank. - pp. 123-138. Cox, Donald and Emmanuel Jimenez (1990). "Achieving Social Objectives Through Private Transfers : A Review". - In:The WorldBank ResearchObserver(5)2 Julv: 205-218. D e Ferranti, David [et al.] (2000). Securing Our Futureina Global Economy. -Washington, D.C. :The World Bank. (Latin America andCaribbeanViewpoints Series). - [et al.] (2003). Inequality in Latin America and the Caribbean : Breaking with History?. - Washington, D.C. ; The World Bank. - (World Bank Latin American and Caribbean Studies). D e Janvry, Alain and Elisabeth Sadoulet. (2000). "Growth, Poverty and Inequality in Latin America : A Causal Analysis, 1970-1994". - In: Review of Income and Wealth (46)3 September: 267-287. Diaz-Cayeros, Alberto and Beatriz Magaloni (2003), "The Politics of Public Spending - Part 11. The Programa Nacional de Solidaridad (PRONASOL) in MCxico". - In: Background Paper for the World Development Report 2004. -Washington, D.C : The World Bank. Duran Fernandez, Roberto, (2003). "La Movilidad a1 Interior de la Distribucion del Ingreso en MCxico Durante e l Period0 1984-2000". -In:Gacetade Economia(Aiio 9)17. Erlich, I.and G.S. Becker (1972). "Market Insurance, Self-Insurance, and Self-Protection". - In:JournalofPoliticalEconomv 80: 623-648. Espinosa-Vega, Marco A. and Tapen Sinha (2000). "A Primer and Assessment of Social Securitv Reform in Mexico". - In: Federal Reserve Bank of Atlanta Economic Review, First Quarter. Fiscal Relations : Argentina, Brazil and Argentina compared (2000?). - Washington, D.C : The World Bank. Fiszbein, Ariel, ed. (2005). Citimes, Politicians and Providers : the Latin American Experience with Service Delivery Reform. -- Washington, D.C. : The World Bank. Fuentes, Ricardo and AndrCs Montes (2003). "Mexico Country Case Study Towards the MillenniumDevelopment Goals at a Sub-national level". - In: Background Paper for Human Development Report 2003. New York :UNPD. - 77 Galasso, Emanuela and Martin Ravallion (2004). Social Protection in a Crisis : Argentina's Plan Jefes y Jefas. - Washington, D.C. : The World Bank.- (Policy Research Working Paper No. 3165). 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. Giugale, Marcel0 M. and Steven Webb ed. (2000). Achievement and Challenges of Fiscal Decentralization :Lessons from Mexico. -- Washington, D.C : The World Bank. Glewwe, Paul and Gillette Hall (1998). "Are Some Groups More Vulnerable to Macroeconomic Shocks? : Hypothesis Tests Based on Panel Data from Peru". - In: Journal of Development Economics (56)1998 : 181-206. Govinda Rao, M.(2000?). Poverty Alleviation under Fiscal Decentralization. - Washington, D.C. The World Bank. http://siteresources.worldbank.org/EXTSAREGTOPDECENTRALIZATION/Resources/ 496899-1095189811806/521763-1095191184981/PovertvAllev.pdf Grandolini, Gloria and Luis Cerda (1998). The 1997 Mexican Pension Reform : Genesis and Design Features.-Washington, D.C. : The World Bank. Haggard, Stephan and Steven Webb (2004). Political Incentives and Intergovernemental. - Washington, D.C. : The World Bank. Hall, Gillette and Christopher Humphrey (2003a). Southern States Development Strategy : Synthesis Report. -Washington, D.C. : The World Bank. Heathr Layton and GuadalupeToscano (2003). "Social Protection Note". -In:Southern States Development Strategv. -Washington, D.C. :The World Bank. and Harry Patrinos (2005). Indigenous Peoples, Poverty, Human Development in Latin America. -London :PalgraveMcMillan, (Forthcoming November 2005). Hernandez Licona, Gonzalo (2000). Dinamica del Mercado Laboral en el MCxico Urbano. - MCxico : Secretariade Desarrollo Social. Hotz, V. Joseph and John Karl Scholz (2001). "The Earned Income Tax Credit". - In: NBER Working Paper No. 8078. Hulme, David and Andrew Shepherd (2003). "Conceptualizing Chronic Poverty". - In: World Development l(3): 403-425. Karen Moore and Andrew Shepherd (2001). Chronic Poverty : Meanings and Analytical Framework.. -- United Kingdom : International Development Department : University ofManchester. IMSS (2004a). Financiamiento Social en MCxico. MCxico : IMSS, (Processed) - (2004b). Informe a1 Ejecutivo Federal y a1 Congreso de la Union sobre la Situacion Financiera y 10s Riesgos del Instituto Mexican0 del Seguro Social. http://www.imss. e;ob.mx/IMSS/IMSS-INF/2004/fina04.htm Krebs, Tom, Pravin Krishna and William Maloney (2005). "Trade Policv. Income Risk, and Welfare". - In:NBERWorking PapersNo. 11255. Krishna, Anirudh (2004). "Escaping Poverty and Becoming Poor : Who Gains, Who Looses and Why?". -In: World Development (32)l: 121-136. 78 Laurell, Asa Christina (2003). "What Does Latin American Social Medicine do when it Governs? :The Case ofMexico City Government". - In:American JournalofPublic Health(93)12. Lederman, Daniel, William F. Maloney and Luis Serven (2004). Lessons from NAFTA for Latin America and the Caribbean : A Summary of Research Findings. - Washington, D.C. : The World Bank. (Latin America andthe Caribbean - Regional Studies). Lisulo, Angela (2003). "Costa Rica : Health Policies". - In: Background Paper to the World Development Report :2004. Washington, D.C. :The World Bank. - Lopez-Calva, Luis Felipe (2003). "El Caso de MCxico". -- Presentation from Social Policy Best Practices Conference.-MCxico : SEDESOL. Maloney, William F., Wendy Cunningham and Mariano Bosch (2003). "Who Suffers Income Falls During Crises? : An Application of Quantile Analysis to Mexico, 1992-95". - In: Background Paper to Securing our Futureina Globalized Economv .-Washington, D.C. :The World Bank. McKay, A. and D. Lawson (2002). "Chronic Poverty : A Review of Current Quantitiative Evidence". -In:CPRCWorking PaperNo. 15. McKenzie, David (2003). Aggregate Shocks and Labor Market Responses : Evidence from Argentina's Financial Crisis. - Stanford : Stanford University Center for Research on Economic Development andPolicy Reform. (Working PaperNo. 176). - MIDEPLAN (2003). Chile Solidario : Cuenta Publica 2002. - Santiago de Chile : Secretaria Ejecutiva. Morduch, Jonathan (1999). Issues on risk and poverty / presented at the Joseph Stiglitz. - In: Summer ResearchWorkshop onPoverty (Washington, D.C. July 6-8, 1999). Narayan, Deepa [et al.] (2000). Voices of the Poor : Crying out for Change. - Oxford : Oxford University Press : The World Bank. Nitsch, Manfred (1997). Recent Developments in Financing Social Security in Latin America. - Geneva : International Labor Organization, (Issues in Social Protection. Discussion Paper No. 1) O'Connell, Allison (2004). Citizen's Pension: Learning from New Zealand. Washington, D.C. : - Public Policy Institute. Oldman, Oliver (1965). "Controlling Income Tax Evasion". -In:ProblemsofTax Administration inLatinAmerica. -- OAS/IDB/ECLAC, 1965. Orszag, Peter and Joseph Stiglitz (2000). "Rethinking PensionReform : Ten Myths about Social Security Systems". - In:New IdeasAbout OldAge Security. - Washington, D.C. : World Bank. pp. 17-56. - Oxley, Howard, Thai Thanh Dang and Pablo Antolin (2000). Poverty Dynamics in Six OECD Countries. -Paris : OECD. - (Economic Studies No. 30,2000/1). Packard, Truman [et al.] (2004). Chile : Household Risk Management and Social Protection. -- Washington, D.C. :The World Bank. Paul, Samuel (2004). "Citizen Report Cards : An Accountability Tool". - In: Developmet Outreach(March). -Washington, D.C : World Bank. Pew Hispanic Center & MIF (2003). Receptores de Remesas en MCxico. - Washington, D.C. : Inter-American Development Bank. 79 Presidencia de la Republica (2004). Anexo Estadistico del Cuarto Informe de Gobierno 2004. http://cuarto.informe.presidencia. gob.mx/ Psacharopoulos, George and Harry Patrinos (1994). Indigenous People and Poverty in Latin America :An EmpiricalAnalysis. - Washington, D.C. : The World Bank.