A Critical Analysis of New Estimates by International Cont'd

III.B. Trends Regional Averages 1989-2005 If absolute measures of poverty differ between data sources, can we find common trends? Data from all three sources suggest that poverty has declined in the region as a whole since 1990. CEPAL, the World Bank and CEDLAS all provide estimates of extreme and moderate poverty in 1990, and according to all three sources, in no year since then has the regional average exceeded the 1990 estimate (see Figures 2 and 3). Figure 2 Extreme Poverty in Latin America 1990-2004 % of Population 60 50 40 30 CEPAL 20 10 World Bank 0 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 Figure 3 Moderate Poverty in Latin America 1990-2004 % of Population 60 CEPAL 50 40 30 20 World Bank 10 0 1988 1990 1992 1994 1996 1998 15 2000 2002 2004 2006 This improvement has been heralded by some as evidence that the region’s turn toward market liberalization has proven successful. But unless income distribution deteriorates rapidly, more income per capita should pull a larger share of the population over a fixed poverty line. Regionally, Latin American economic growth per capita has averaged about 1 percent per year since 1990. This is very slow compared to most Asian countries, but it is positive. That one must ask whether poverty has fallen over the past decade and a half, and not by how much, suggests concern about regressive trends in inequality. It is also hard to assess this data without asking, compared to when? Much debate about progress depends on the choice of a base year. 1990 arguably sets a low bar for the region, coming at the end of the debt crisis and widespread hyperinflation. According to CEPAL, the regional poverty rate only finally fell below the 1980 level in 2006.27 Choosing a more recent base year can be equally problematic. Since 1990, financial shocks have raised poverty rates in several countries, although at different times. Among the larger countries that dominate the regional average, shocks hit Mexico (1995), Brazil (1999), Argentina (2002) and Venezuela (1999, 2002-3). Progress on poverty in the intervening years is at least partly attributable to macroeconomic recovery and a return to trend. These crises make the choice of a base year difficult and complicate efforts to assess progress across countries. The Aggregation Problem: Mexico, Brazil, and Everyone Else Aggregation may be appealing as a tool to manage the stories of eighteen-plus countries, but it is almost meaningless from a policy perspective. Countries in the region do not move in sync economically, even if they adopt common policy stances like neoliberalism. Nor do the richer countries provide much aid to the poorest countries, notwithstanding Venezuela’s aid to a few allies. Major social policies are set at the national level, and thus poverty is best viewed from a national perspective. Together, Mexico and Brazil account for more than 40% of the region’s moderately poor population, and regional averages are dominated by trends in these two countries. In fact, because Brazil’s poverty rate has been stagnant over the past decade, the regional story is largely one about Mexico. When one disaggregates the data, as we do below, it seems less reasonable to generalize about progress in the region. According to both CEPAL and the World Bank, Mexico has made dramatic headway against poverty. This trend was interrupted by the Peso Crisis of 1995, but gains since then have brought both moderate and extreme poverty levels below those of the 1980s. In fact, World Bank data support the controversial claim by some Mexican 27 ECLAC (2006) Social Panorama of Latin America: 2006. Note that comparisons between 1980 and 1990 are unreliable because of incomplete surveys, the difficulty of adjusting prices, and inconsistent methodologies. 16 policymakers that extreme poverty has been almost eradicated.28 By contrast, CEPAL data indicates that some 12 million Mexicans still cannot afford the cost of a basic food basket. Table 4: Mexico - Poverty Headcount Ratios, 1981-2004 Percent of population 1981 Extreme Poverty Moderate Poverty CEPAL World Bank CEPAL World Bank 10.9 34.7 1989/1990 18.7 6.2 47.7 24.3 1996 22.0 7.8 52.9 27.8 2001/2002 12.6 4.3 39.4 21.2 Some readers will wonder about this evidence. How can such progress be consistent with recent political strife? Perhaps it is about the speed and not the direction of poverty reduction, or the fact that, as indicated by CEPAL (but not the World Bank), Mexico’s poverty rate is far above what one would expect given its income level. Discord about equity in Mexico, even when the data are consistent about the direction of trend, highlights the complexity of poverty discussed earlier.29 The hardships associated with neoliberalism have unevenly affected low-income Mexicans, depending on their geographic region, economic sector and occupation. The basic data provide no information about the distribution of households around the poverty line, nor about chronic poverty and ethnic segregation, and they rely on a poverty line that may be too low for an affluent country like Mexico.30 Brazil has also made considerable progress since 1990, but very little, if any, progress over the past decade. This is surprising, given liberal leaders’ implementation of highly visible anti-poverty programs, including Bolsa Familia. Yet national poverty rates, both extreme and moderate, have changed very little – and may even have increased – since 1996. (The World Bank data suggests an increase in extreme poverty; CEPAL data shows an increase in moderate poverty.) Progress toward achieving the Millennium 28 Damián, A. and J. Boltvinik point out that the absolute number of Mexicans living below the official food-based poverty line actually rose between 1992 and 2000. They also show that the city-size cutoff between rural and urban areas may skew the evidence of declining headcount ratios. Damián, A. and J. Boltvinik (2003). "Evolución y características de la pobreza en México." Comercio Exterior 53(6): 519531. 29 Peasant farmers in southern Mexico have been hurt, while many low income Mexicans in the north have gained access to industrial jobs and migrant work in the US. Work by Laos (2000) also suggests that the restructuring of the Mexican economy may have more negatively affected the working class (or near-poor) than the poor. For a concise overview of geographic inequality and the consequences of neoliberalism in Mexico, see “Mexico’s Mezzogiorno: What is Needed to Bridge the Gaping North-South Divide,” The Economist, November 18, 2006, Survey of Mexico, p. 5. 30 See Bolvinik (2001), “Opciones methodologicas para medir la pobreza en México,” Comercio Exterior, Octubre. 2001, pp. 869-878, for a discussion of appropriate poverty lines in Mexico. 17 2004 11.7 1.9 37.0 12.5 Development Goal of halving the 1990 extreme poverty rate by 2015 largely reflects recovery from the hyperinflation crisis of 1989-91.31 Table 5: Brazil - Poverty Headcount Ratios, 1981-2004 Percent of population 1981 1989/1990 CEPAL 23.4 Extreme Poverty 14.0 World Bank 11.8 48.0 Moderate CEPAL Poverty 32.3 World Bank 31.1 Sources: ECLAC 2007; Chen and Ravallion 2007. 1996 13.9 6.9 35.8 21.7 2001/2002 13.2 6.7 37.5 21.2 One explanation for Brazil’s stubborn poverty is its stagnant economy. GDP per capita has grown very slowly over the past decade and a half ─ 0.6 percent, compared to 1.6% in Mexico ─ and distributional shifts, while favorable, have been miniscule. Furthermore, its efforts to reduce poverty are set in a much poorer country: per capita GDP in Brazil is 42% less than in Mexico. Even if the depth of poverty (the poverty gap) were the same ─ it is worse in Brazil ─ it would take faster growth or proportionately greater redistribution in Brazil to achieve comparable progress. Based on the preponderance of evidence, the gap in progress between Mexico and Brazil is almost certainly real. But if Mexico’s wealth, more rapid growth and more equitable income distribution explains its faster progress against poverty, it does not answer why, according to CEPAL, the two countries have similar levels of poverty (about 37% in 2004). For that, one must return to table 1, which shows that CEPAL sets Mexico’s poverty line at $70.60 and Brazil’s at $26.10, as well as to the earlier discussion of how imputed rents, adult equivalence and under-reported income are treated. In other words, this may be a statistical artifact. Everyone Else If we take out Mexico and Brazil, or slightly more than half the regional population,32 is there a consistent story in the remaining countries? Among the remaining countries, the regional aggregate poverty rate has climbed. Aggregated World Bank data is presented in Table 6.33 31 There has been notable progress in reducing rural poverty in Brazil. 32 No other country comes close in size to Brazil (186 million) or Mexico (103 million). Colombia, with 46 million people, accounts for 8% of the region’s population of 551 million. Argentina (7%), Peru (5%) and Venezuela (5%) follow. Other countries each hold less than 3% of the region’s population (World Bank, 2007, WDI Online). 33 We have included Uruguay in regional averages, but because its poverty data is so sparse, we have not included it in the discussion of individual countries. 18 2004 12.1 7.6 37.7 19.8 Interestingly, this deterioration is not mainly about the poorest countries getting poorer. It is in the next four largest countries – Colombia, Argentina, Peru and Venezuela, none of which are poor by regional standards – that poverty has risen dramatically, at least according to the World Bank. Together, they account for 25% of the region’s population. Table 6: Aggregate Poverty Ratios, 1981-2004, World Bank PovcalNet Data Percent of population 1981 1990 1996 2002 2004 Extreme Poverty 10.2 9.6 8.1 8.4 8.0 Latin America Latin America without Brazil and 8.5 7.4 9.3 11.7 11.1 Mexico 4.0 1.7 6.9 10.9 10.1 Medium Four Countries 24.0 26.8 27.4 27.0 25.7 Poorest Six Countries Small Countries with Signs of 15.6 13.0 7.5 7.0 7.0 Progress Moderate Poverty 27.9 25.6 24.4 24.0 21.4 Latin America Latin America without Brazil and 21.8 20.5 25.0 27.5 26.7 Mexico 13.1 11.0 21.9 26.7 25.3 Medium Four Countries 46.3 49.2 49.0 49.4 48.0 Poorest Six Countries Small Countries with Signs of 34.5 29.9 21.5 18.7 19.1 Progress Notes: “Latin America” includes Argentina, Bolivia, Brazil, Chile, Colombia, Costa Rica, Dominican Republic, Ecuador, El Salvador, Guatemala, Haiti, Honduras, Mexico, Nicaragua, Panama, Paraguay, Peru, Uruguay, and Venezuela. “Small Countries with Signs of Progress” include Chile, Costa Rica, Dominican Republic, El Salvador, Guatemala, and Panama. “Medium Four Countries” include Argentina, Colombia, Peru, and Venezuela. “Poorest Six Countries” include Bolivia, Ecuador, Haiti, Honduras, Nicaragua, and Paraguay. Sources: Chen and Ravallion 2007; World Bank 2007. 19 If the lesson here is to disaggregate data by country, can we draw firm conclusions about any of the four medium size countries? Again, the data source matters. There is a clear deterioration in poverty levels in all four countries since 1989/90 according to World Bank data, but CEPAL data actually shows improvement in Colombia, and much less deterioration in Venezuela (see Table7).34 Table 7: Medium Four Countries, Moderate Poverty, 1980-2004 Percent of population 1980-81 1989-91 1995-97 2000-02 2003-04 ARGENTINA (urban only) PovcalNet (World Bank) BADEINSO (CEPAL) 1.4 8.5 8.3 .. 9.8 .. 21.6 45.4 17.4 29.4 COLOMBIA PovcalNet (World Bank) BADEINSO (CEPAL) 20.2 42.3 11.7 56.1 (MC) 18.9 50.9 19.4 51.1 (MC) 19.4 51.1 PERU PovcalNet (World Bank) 9.6 10.4 28.4 32.2 30.6 BADEINSO (CEPAL) .. .. 47.6 54.8 (MC) 54.7 VENEZUELA PovcalNet (World Bank) 22.6 14.5 36.6 40.2 40.2 BADEINSO (CEPAL) 25.0 39.8 48.0 48.6 45.4 Note: “MC” indicates significant change in survey methodology, such that the estimate is not considered comparable to previous years. Sources: Chen and Ravallion 2001 and 2007; ECLAC 2007; World Bank 2007. 34 Particularly in Venezuela, the absence of recent data makes it difficult to assess the impact of radical shifts in economic and social policy. 20 Among the poorest six countries, only in Honduras is there a significant improvement reported by both the World Bank and CEPAL data (see Table 8). However, the striking progress in Honduras reported by the World Bank is not confirmed by CEPAL, which shows much smaller gains, or by CEDLAS, which shows increasing poverty. (See the appendix for detail.) Data from the World Bank suggest that Nicaragua and Haiti have seen slight declines in poverty. For Nicaragua, this finding is supported by (admittedly sparse) estimates from CEPAL. In the case of Haiti, poverty estimates are not available from CEPAL or CEDLAS. Data from Bolivia, Ecuador and Paraguay show deterioration in poverty rates since 1990 across sources, but each shows markedly different rates of change within subperiods. Data sparseness is a particularly severe problem for Ecuador, while Paraguay’s data suffer most from inconsistency. Table 8: Poorest Six Countries, Moderate Poverty, 1980-2004 Percent of population BOLIVIA PovcalNet (World Bank) BADEINSO (CEPAL) ECUADOR PovcalNet (World Bank) BADEINSO (CEPAL) HAITI PovcalNet (World Bank) BADEINSO (CEPAL) HONDURAS PovcalNet (World Bank) BADEINSO (CEPAL) 1980-81 1989-91 1995-97 2000-02 2003-04 24.0 .. 28.7 .. 40.1 62.1 42.9 62.4 42.8 63.9 28.8 .. 31.2 .. 37.4 .. 36.9 49.1 35.2 51.2 77.3 .. 79.0 .. 81.7 .. 77.6 .. 75.8 .. 59.8 .. 64.2 80.8 31.6 79.1 36.0 77.3 36.0 74.8 60.7 .. 84.3 73.6 (1993) 77.2 .. 81.6 69.3 79.8 .. NICARAGUA PovcalNet (World Bank) BADEINSO (CEPAL) PARAGUAY PovcalNet (World Bank) 23.9 26.3 34.3 33.2 29.8 BADEINSO (CEPAL) .. .. .. 61.0 65.9 Note: “MC” indicates significant change in survey methodology, such that the estimate is not considered comparable to previous years. Sources: Chen and Ravallion 2001 and 2007; ECLAC 2007; World Bank 2007. 21 Among the small middle income countries, several have made gains against poverty (see Table 9). In the cases of Chile, Costa Rica, and Guatemala, the gains have been dramatic, although Guatemala’s data is incomplete. To a lesser degree, El Salvador and Panama have also reduced poverty. Progress in the Dominican Republic is contradicted by CEDLAS data, which shows a rise in moderate poverty from 8% in 1995 to 16.4% in 2003. World Bank data indicate that, as a group, these countries have reduced moderate poverty from 30% to 19%, and extreme poverty from 13% to 7%. Unfortunately, together they account for only 9.5% of the region’s population. Nonetheless, some of them may provide replicable policy models for other Latin American countries. Table 9: Small Countries with Signs of Progress, Moderate Poverty, 1980-2004 Percent of population 1980-81 1989-91 1995-97 2000-02 2003-04 CHILE PovcalNet (World Bank) BADEINSO (CEPAL) 17.9 .. 14.1 38.6 8.2 23.2 5.6 20.2 5.6 18.7 COSTA RICA PovcalNet (World Bank) BADEINSO (CEPAL) 32.0 23.6 16.1 26.3 13.3 22.5 8.9 20.3 9.6 20.5 29.5 .. 21.4 .. 11.7 8.0 12.1 .. 16.2 .. 39.7 .. 43.0 .. 51.9 55.5 40.5 48.9 40.4 47.5 PovcalNet (World Bank) BADEINSO (CEPAL) 68.1 71.1 58.8 69.4 34.4 .. 32.6 60.2 31.0 .. PANAMA PovcalNet (World Bank) BADEINSO (CEPAL) 12.7 .. DOMINICAN REPUBLIC PovcalNet (World Bank) BADEINSO (CEPAL) EL SALVADOR PovcalNet (World Bank) BADEINSO (CEPAL) GUATEMALA 24.1 18.5 17.5 16.8 41.0 .. 34.0 31.8 (1986) Note: “MC” indicates significant change in survey methodology, such that the estimate is not considered comparable to previous years. Sources: Chen and Ravallion 2001 and 2007; ECLAC 2007; World Bank 2007. At this point, it must be apparent that the authors have only drawn cautious conclusions about the direction of change. While policy debate ought to be about the pace of progress in the context of economic growth, we would not hazard guesses about the exact pace of progress anywhere. Inconsistencies between the data sources are simply too pervasive to support such an analysis, and in some cases, even the direction of progress is questionable. Poverty has probably declined in Mexico, Chile, Costa Rica, Panama, El Salvador, Nicaragua, Honduras, Guatemala, and Brazil; it has probably increased in Bolivia, Ecuador, Argentina, Peru, Colombia, Paraguay, Venezuela and Uruguay. 22 Skeptical readers might find even these tentative interpretations unfounded, given the erratic nature of the data. Equally difficult is an assessment of data on the ‘poverty gap,’ or the average shortfall of income from the poverty line. Where the poverty gap is small, it is often argued that households move back and forth across the poverty line, affecting poverty rates but having little impact on the quality of the life for the poor/nearly poor. Poverty gap data is published by the World Bank, CEPAL and CEDLAS, but interpreting it is treacherous. In 1998, for example, the World Bank reported that the poverty gap for Guatemala, one of the poorest countries in the region, was the same as that of Mexico, and lower than that of Costa Rica. Moreover, it is not uncommon for poverty gap estimates to be as little as zero or one percent, even where poverty is prevalent. It is possible to construct a few credible stories from this data (e.g., the poverty gap in Argentina rose in the early 2000s), but a systematic analysis would also yield as many implausible interpretations.35 III.C. Progress, Growth and Distribution Explanations for changes in poverty levels typically come down to two issues: growth and distribution. These in turn depend on productivity, education, social mobility, land use, employment, the distribution of assets, and the structure of taxes and subsidies, among other factors. Which of these policies is most important in overcoming poverty, and whether they are at odds with one another or can operate in concert, has been the subject of intense debate, particularly in the context of neoliberal reform. Table 10 provides data on growth and distribution by country for readers who hope to make sense of overall trends in the poverty data, however tentatively. Compared to other regions of the world, Latin America has been doubly cursed: growth has been slow, and nearly every country ranks among the most unequal in the world. 35 World Bank (2007) Millennium Development Goals Quick Query, http://ddp-ext.worldbank.org /ext/DDPQQ/member.do?method=getMembers&userid=1&queryId=27. 23 Table 10: Growth and Distribution in Latin America 1990-2004 Country (Gini dates) GDP/cap growth, Gini coefficient, Gini coefficient, 1990-2004 early 1990s early 2000s 1.9 50.1 53.1 Argentina (1990, 2004) 1.3 53.8 61.4 Bolivia (1989, 2002) 0.6 62.7 61.2 Brazil (1990, 2003) 4.0 55.4 55.2 Chile (1990, 2003) 1.0 60.1 57.7 Colombia (1994, 2004) 2.3 43.8 47.8 Costa Rica (1990, 2004) 2.7 51.7 58.6 Dom. Rep. (1997, 2004) 0.8 .. 51.3 Ecuador (2004) 1.9 50.7 49.3 El Salvador (1995, 2004) 1.2 58.2 54.3 Guatemala (1989, 2002) 0.4 61.5 58.7 Honduras (1990, 2003) 1.5 53.6 51.6 Mexico (1989, 2004) 0.8 58.2 57.9 Nicaragua (1993, 2001) 2.8 54.5 54.8 Panama (1991, 2004) 0.6 44.7 54.8 Paraguay (1990, 2004) 1.5 53.2 50.5 Peru (1997, 2004) 1.3 49.2 46.4 Uruguay (1990, 2004) 0.2 47.1 47.0 Venezuela (1990, 2004) 1.1 .. .. Latin America Sources: GDP: World Bank, WDI Online; Ginis: CEPAL, BADEINSO online (July 2007, Chile Gini data from March 2007) The relatively fastest growing countries, Chile, Costa Rica, Panama and El Salvador have all seen reductions in poverty since 1990. At the opposite extreme, Brazil, Ecuador, Honduras, Nicaragua, Paraguay, and Venezuela have had very slow growth, and none have had unambiguous evidence of declining poverty. Distribution has improved, at least a little, in several countries. But this is not the case in Argentina, Bolivia, Costa Rica, the Dominican Republic, El Salvador, Paraguay and Venezuela, at least prior to the early 2000s. The failure to improve distribution has been associated with higher poverty rates in all of these cases except Costa Rica and El Salvador. Even in Argentina and Bolivia, the salutary effect of decent growth has been entirely undermined by rising inequality.36 What kind of progress should we expect if an economy is growing? Better data would support informed debate on the pace of progress under various macroeconomic scenarios, and the effectiveness of strategies to overcome persistent poverty. Recent 36 Curiously, Peru has seen both decent growth and a modest improvement in equality, but poverty has increased. What might be going on? The Gini coefficient captures distribution across all levels of income, while poverty affects only the bottom quintiles (except in the poorest countries). Thus, Gini coefficients can decline in the absence of gains to the poor. Decile-by-decile distributional data provides a better sense of the role of distribution in fighting poverty. 24 efforts have been made to estimate the elasticity of poverty with respect to growth. 37 Crucially, the effect of growth on poverty depends on income distribution. If income is inequitably distributed, fewer of the benefits of growth reach the poor, and progress is slower than in more equal societies. The fact that poverty is as high and as stubborn as it is in Latin America reflects the fact that trickle down is extremely slow in the context of income inequality. In the absence of rapid growth and progress on overall inequality, social policy is essential to change poverty rates. Much has been made of the impact of Mexico’s Oportunidades program on extreme poverty rates, even as it costs less than half a percent of its GDP. Combined with an influx of worker remittances, cash payments from Oportunidades have helped push many Mexicans over the poverty line. In wealthier countries like Mexico, it may be possible to reduce poverty, as it has been narrowly defined here, with minimal redistribution. Creating jobs and incorporating the poor into a common ‘middle income’ economy will be more challenging. In countries where poverty is deeper, progress will require tackling inequality. The reduction of poverty through conditional cash transfer programs like Oportunidades returns us to questions about how we measure poverty. Small cash transfers to the poor obviously reduce official estimates of poverty. The associated incentives to increase school attendance and health checkups may also raise standards of living in the long run. Yet if policy makers are forced to choose between these programs and food subsidies, sewerage construction or road paving in poor neighborhoods, accurate accounting of such trade-offs is unlikely to occur in our poverty data.. In other words, even as we argue that a dramatic reduction in poverty lies well within the grasp of wealthier Latin American countries, the measurement of that success is sure to provoke controversy. 38 IV. Conclusions Poverty in Latin America is complex, and progress requires an ongoing effort to set priorities that genuinely improve the quality of life, whether through better labor laws, lower crime rates, or rural electrification. Measuring progress, a task essential to assessing policy, is no simple matter. Even if one focuses narrowly on the percent of households that fall below an absolute poverty line, this headcount approach provides no information about the depth of poverty below that line, nor much information about the quality of life in households 37 Using World Bank data, A. Kalwij and A. Verschoor estimate the elasticity of poverty reduction with respect to growth in "Not by growth alone: The role of the distribution of income in regional diversity in poverty reduction." European Economic Review, May 2007. 38 For a review of conditional cash transfer programs, see S. Handa and B. Davis (2006) “The Experience of Conditional Cash Transfers in Latin America and the Caribbean,” Development Policy Review 24(5): 513-536, September 2006. 25 with such meager economic resources. Yet even a straightforward estimate of absolute poverty is elusive. The three main international sources of data, CEPAL, the World Bank, and CEDLAS, turn out to be neither independent, in the sense that they draw on common household surveys by national governments, nor particularly consistent. Different poverty lines and methodological choices lead to markedly different estimates, despite common survey data. As Székely, et al, comment, “Given the possible range of results and the lack of a standard, widely accepted methodology, these statistics become practically meaningless if the user does not have at least some minimum guidance and explicit information on the underlying choices that are necessary to interpret them.”39 Readers will undoubtedly take away from this a strong dose of skepticism about the usefulness of poverty data. Unfortunately, skepticism is no substitute for understanding how macroeconomic and social policies affect poor families. Thus we find ourselves trying to parse reality from the patchwork of available data. It is tempting to dismiss differences between the estimates, and to focus exclusively on trends. Doing so means giving up the hope of understanding the scale of deprivation. In a relatively rich region, it should matter whether 10% or 20% of the population lacks the resources needed to buy a basic basket of food, even if we quibble about what should be included in the contents of that basket. This should matter not only to domestic policymakers, but also to international relief organizations that hope to direct their aid to countries where it is most needed. There are some consistent trends across sources that may satisfy readers seeking a balance between agnosticism and blind confidence in the data.40 Most hopefully, Mexico has reversed the sharp increase in poverty that followed the Peso crisis. Panama also made significant headway over the past decade and half. But poverty has probably risen, despite economic growth, in Argentina, Bolivia, Ecuador, Paraguay, Peru, Uruguay and Venezuela. Elsewhere, the gains are very modest, or direction of trend depends on the data source used and the sub-period under consideration That said, many countries have experienced unambiguous ‘progress’ that is little more than a recovery from macroeconomic shocks. Others have fallen into prolonged periods of stagnation and can claim progress only by reaching back to gains from the early 1990s. Where we are able to make such statements confidently, we at least have the option of informed policy discussion. In many instances, however, we simply cannot find consistent data. 39 Szekely, et al., 7. 40 John Sheahan has pointed out that there have been many individual country studies in which the authors have gone to great lengths to sort out what has happened to poverty in the particular country studied, with results that often seem valid. These studies, many by historians, political scientists, and sociologists, can offer a fuller understanding of the context in which the poor live. 26 The purpose of collecting poverty data is to identify policies that successfully ameliorate the hardships faced by the poor. International institutions can create a stronger basis for assessing anti-poverty policies by creating common, transparent approaches to data collection. Until then, conclusions from empirical research must be tempered with awareness that not only the pace of change, but the very direction of change may be contradicted elsewhere. This is not to suggest that we should reject the evidence entirely. Rather, we counsel caution in the use of this data. None of the three sources seems more reliable or sensible in its handling of the data. It is common to turn to ‘facts’ in the heat of an ideological debate, and poverty is one of the most contentious topics in development economics. The evidence presented here suggests that few arguments will be resolved with the international data on poverty. If one must resort to picking one of these data sources over another to demonstrate progress, much less adequate progress, the poor themselves are right to cry foul. 27 Appendix Table A-1: Extreme Poverty Percent of population 1980-81 1989-91 1995-97 2000-02 2003-04 .. .. .. 0.3 .. 1.0 1.1 .. 3.7 (MC) 7.5 20.9 6.9 (MC) 6.6 11.1 5.5 (MC) BOLIVIA PovcalNet (World Bank) BADEINSO (CEPAL) SEDLAC (CEDLAS) 3.3 .. .. 5.7 .. .. 21.1 37.2 18.3 24.0 37.1 28.3 23.9 34.7 .. BRAZIL PovcalNet (World Bank) BADEINSO (CEPAL) SEDLAC (CEDLAS) 11.8 .. .. 14.0 23.4 11.7 6.9 13.9 8.7 6.7 13.2 8.5 7.6 12.1 6.9 CHILE PovcalNet (World Bank) BADEINSO (CEPAL) SEDLAC (CEDLAS) 2.7 .. .. 2.0 13.0 3.3 0.0 5.7 1.5 0.5 5.6 1.6 0.5 4.7 1.4 COLOMBIA PovcalNet (World Bank) BADEINSO (CEPAL) SEDLAC (CEDLAS) 7.9 .. .. 2.6 26.1(MC) .. 5.6 23.5 8.8 7.6 24.6 (MC) 12.4 7.6 24.2 14.8 (MC) COSTA RICA PovcalNet (World Bank) BADEINSO (CEPAL) SEDLAC (CEDLAS) 14.8 .. .. 5.2 9.9 6.1 3.6 7.8 3.4 1.6 8.2 4.3 1.8 8.0 3.6 DOMINICAN REPUBLIC PovcalNet (World Bank) BADEINSO (CEPAL) SEDLAC (CEDLAS) 11.0 .. .. 3.9 .. .. 1.8 22.1 2.1 1.9 20.3 3.9 (MC) 2.8 29.0 3.4 12.4 .. .. 13.6 .. .. 16.3 .. .. 15.6 .. 23.0 14.7 22.3 14.7 (MC) 19.6 .. .. 21.4 .. 24.9 25.3 21.7 .. 20.4 22.1 19.8 20.4 19.0 16.4 41.4 .. .. 34.9 42.0 .. 13.9 .. .. 13.9 30.9 15.4 13.1 .. 12.8 ARGENTINA (urban only) PovcalNet (World Bank) BADEINSO (CEPAL) SEDLAC (CEDLAS) ECUADOR PovcalNet (World Bank) BADEINSO (CEPAL) SEDLAC (CEDLAS) EL SALVADOR PovcalNet (World Bank) BADEINSO (CEPAL) SEDLAC (CEDLAS) GUATEMALA PovcalNet (World Bank) BADEINSO (CEPAL) SEDLAC (CEDLAS) 28 HAITI PovcalNet (World Bank) BADEINSO (CEPAL) SEDLAC (CEDLAS) HONDURAS PovcalNet (World Bank) BADEINSO (CEPAL) SEDLAC (CEDLAS) MEXICO PovcalNet (World Bank) BADEINSO (CEPAL) SEDLAC (CEDLAS) 1980-81 1989-91 1995-97 2000-02 2003-04 52.6 .. .. 54.8 .. .. 58.8 .. .. 52.9 .. .. 50.7 .. .. 33.0 .. .. 37.8 60.9 .. 12.3 54.4 12.0 14.1 54.4 .. 14.1 53.9 16.8 10.9 .. .. 6.2 18.7 10.9 7.8 22.0 19.4 4.3 12.6 10.0 1.9 11.7 10.4 29.5 .. 57.7 48.4 (1993) 31.8 (1993) 44.5 .. 47.7 42.3 44.9 .. .. 18.0 .. 7.9 .. 6.1 17.4 6.0 .. .. 11.8 19.7 (1986) 17.0 11.4 11.5 6.1 4.0 .. .. 4.9 .. .. 17.5 33.2 11.0 16.4 .. 12.6 13.6 36.9 10.2 NICARAGUA PovcalNet (World Bank) BADEINSO (CEPAL) SEDLAC (CEDLAS) PANAMA PovcalNet (World Bank) BADEINSO (CEPAL) SEDLAC (CEDLAS) PARAGUAY PovcalNet (World Bank) BADEINSO (CEPAL) SEDLAC (CEDLAS) .. 2.1 .. PERU PovcalNet (World Bank) 1.1 1.4 8.9 12.9 10.5 BADEINSO (CEPAL) .. .. 25.1 24.4 (MC) 18.6 SEDLAC (CEDLAS) .. .. 14.3 14.9 (MC) 10.0 URUGUAY PovcalNet (World Bank) .. .. 0.8 0.0 0.0 BADEINSO (CEPAL) 3.3 3.4 1.7 2.5 4.7 SEDLAC (CEDLAS) .. 0.3 0.8 0.3 0.8 VENEZUELA PovcalNet (World Bank) 6.3 3.0 14.8 18.7 18.7 BADEINSO (CEPAL) 8.6 14.4 20.5 22.2 19.0 SEDLAC (CEDLAS) .. .. .. .. .. Note: “MC” indicates significant change in survey methodology, such that the estimate is not considered comparable to previous years. Sources: Chen and Ravallion 2007; ECLAC 2007; CEDLAS and LAC Poverty Group 2007. 29 Table A-2: Moderate Poverty Percent of population 1980-81 1989-91 1995-97 2000-02 2003-04 ARGENTINA (urban only) PovcalNet (World Bank) BADEINSO (CEPAL) SEDLAC (CEDLAS) 1.4 8.5 .. 8.3 .. 3.5 9.8 .. 8.9 (MC) 21.6 45.4 24.7 (MC) 17.4 29.4 14.9 (MC) BOLIVIA PovcalNet (World Bank) BADEINSO (CEPAL) SEDLAC (CEDLAS) 24.0 .. .. 28.7 .. .. 40.1 62.1 36.2 42.9 62.4 43.1 42.8 63.9 .. 31.1 32.3 48.0 21.7 35.8 21.2 37.5 19.8 37.7 CHILE PovcalNet (World Bank) BADEINSO (CEPAL) SEDLAC (CEDLAS) 17.9 .. .. 14.1 38.6 14.4 8.2 23.2 6.8 5.6 20.2 5.9 5.6 18.7 5.1 COLOMBIA PovcalNet (World Bank) BADEINSO (CEPAL) SEDLAC (CEDLAS) 20.2 42.3 .. 11.7 56.1 (MC) .. 18.9 50.9 16.7 19.4 51.1 (MC) 23.2 19.4 51.1 26.3 (MC) COSTA RICA PovcalNet (World Bank) BADEINSO (CEPAL) SEDLAC (CEDLAS) 32.0 23.6 .. 16.1 26.3 13.3 13.3 22.5 8.5 8.9 20.3 9.6 9.6 20.5 8.3 DOMINICAN REPUBLIC PovcalNet (World Bank) BADEINSO (CEPAL) SEDLAC (CEDLAS) 29.5 .. .. 21.4 .. .. 11.7 8.0 .. 12.1 .. 16.5 16.2 .. 16.4 28.8 .. 31.2 .. 37.4 .. 29.3 36.9 49.1 35.2 51.2 37.2 (MC) 39.7 .. .. 43.0 .. 51.1 51.9 55.5 .. 40.5 48.9 41.4 40.4 47.5 38.7 68.1 71.1 .. 58.8 69.4 .. 34.4 .. .. 32.6 60.2 34.3 31.0 .. 34.9 77.3 .. .. 79.0 .. .. 81.7 .. .. 77.6 .. .. 75.8 .. .. BRAZIL PovcalNet (World Bank) BADEINSO (CEPAL) SEDLAC (CEDLAS) ECUADOR PovcalNet (World Bank) BADEINSO (CEPAL) SEDLAC (CEDLAS) EL SALVADOR PovcalNet (World Bank) BADEINSO (CEPAL) SEDLAC (CEDLAS) GUATEMALA PovcalNet (World Bank) BADEINSO (CEPAL) SEDLAC (CEDLAS) HAITI PovcalNet (World Bank) BADEINSO (CEPAL) SEDLAC (CEDLAS) 30 HONDURAS PovcalNet (World Bank) BADEINSO (CEPAL) SEDLAC (CEDLAS) MEXICO PovcalNet (World Bank) BADEINSO (CEPAL) SEDLAC (CEDLAS) 1980-81 1989-91 1995-97 2000-02 2003-04 59.8 .. .. 64.2 80.8 .. 31.6 79.1 32.6 36.0 77.3 .. 36.0 74.8 37.4 34.7 24.3 47.7 27.8 52.9 21.2 39.4 12.5 37.0 60.7 .. 84.3 73.6 (1993) 58.6 (1993) 77.2 .. 81.6 69.3 79.8 .. .. 45.8 .. 18.5 .. 17.5 34.0 16.8 31.8 .. 24.1 41.0 (1986) 26.3 19.6 17.7 15.8 23.9 .. .. 26.3 .. .. 34.3 .. 24.6 33.2 61.0 25.7 29.8 65.9 26.0 NICARAGUA PovcalNet (World Bank) BADEINSO (CEPAL) SEDLAC (CEDLAS) PANAMA PovcalNet (World Bank) BADEINSO (CEPAL) SEDLAC (CEDLAS) PARAGUAY PovcalNet (World Bank) BADEINSO (CEPAL) SEDLAC (CEDLAS) .. 12.7 .. PERU PovcalNet (World Bank) 9.6 10.4 28.4 32.2 30.6 BADEINSO (CEPAL) .. .. 47.6 54.8 (MC) 54.7 SEDLAC (CEDLAS) .. .. 32.6 36.1 (MC) 30.2 URUGUAY PovcalNet (World Bank) BADEINSO (CEPAL) SEDLAC (CEDLAS) VENEZUELA PovcalNet (World Bank) 22.6 14.5 36.6 40.2 40.2 BADEINSO (CEPAL) 25.0 39.8 48.0 48.6 45.4 SEDLAC (CEDLAS) .. .. .. .. .. Note: “MC” indicates significant change in survey methodology, such that the estimate is not considered comparable to previous years. Sources: Chen and Ravallion 2007; ECLAC 2007; CEDLAS and LAC Poverty Group 2007. 31