Stagnant Poverty Reduction in Latin America

Public Disclosure Authorized Poverty and Inequality Monitoring: Latin America and the Caribbean Stagnant Poverty Reduction in Latin America Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized Poverty reduction in the Latin America (LA) region remained stagnant in 216 as the economic contraction deepened, while the middle class has slightly declined. Poverty rates were once again effectively unchanged for half the countries. Income growth of the bottom 4 percent of the Latin American income distribution was negative in both 21 and 216, driven by Brazil, but overall inequality decreased due to larger income declines among wealthier households. Preliminary data are ambiguous over the likely poverty outcomes in 217. Non-labor income will need to play an important role given the low labor income increases seen in preliminary labor market data. Poverty reduction in the Latin America (LA) region remained stagnant in 216 as the economic contraction deepened. The contraction of 1.4 percent in regional GDP per capita in 21 deepened into a contraction of 1.8 percent in 216. 1 Poverty rates remained flat relative to 21, increasing only.2 percentage points, resulting in one fourth of the population continuing to have incomes below the global upper middle-income poverty line of USD. per day (211 PPP) 2. Likewise, around percent of the population remained below the global lower middle-income poverty line (Figure 1). Since 214, the middle class has not only stopped growing, but has slightly declined. The middle class, defined as those living with between US$13 and $7 per day (211 PPP), decreased slightly from 36 percent of the population in Latin America in 214 to around 3 percent in 216. 3 The vulnerable 1 GDP per capita data from World Development Indicators, last updated March 1, 218 2 In October 217, the World Bank began reporting poverty rates for all countries using two new international poverty lines: $3.2/day (in 211 PPP) and $./day. These new lines have been derived by averaging the official poverty lines in lower middle-income countries and upper-middle income countries across the world, respectively. The new $3.2 and $./day lines are designed to complement, not replace, the International Poverty line of $1.9/day, and will facilitate comparisons between countries across the world. For more information see: Jolliffe and Prydz (216). population, those households who are not in poverty nor in the middle class, comprise almost 38 percent of the population, remaining the largest share in LA. Figure 1: Continued stagnation in poverty reduction and middle class growth in 216 4 3 2 2 22 24 26 28 2 212 214 216 Middle Class US$ 13-7 PPP Poverty US$. PPP Poverty US$ 3.2 PPP Vulnerable US$.-13 PPP Source: LAC Equity Lab tabulations using SEDLAC data (CEDLAS and the World Bank) 3 Previously, the thresholds to identify the vulnerable and the middle classes were set at $ and $ a day in 2 PPP. For the case of LAC, these thresholds were updated from 2 PPP to 211 PPP in the following way. Using the 2 PPP conversion factor of each country, the vulnerable and middle-class lines were converted to local currency units at 2 prices. These values were deflated to 211 prices using the CPI of each country and converted back to US dollars using their corresponding 211 PPP conversion factor. Finally, a simple average of the resulting lines was done to obtain a regional value. By rounding to the closest unit, the vulnerable and middle-class lines in 211 PPP for LAC would be US $13 and $7 a day, respectively. 1 JUNE 14, 218

Poverty rates in 216 were once again effectively unchanged for half the countries. Almost two-thirds of Latin American countries had worse GDP per capita growth rates in 216 relative to the previous year (Figure 2). At the household level, poverty decreased in of the 12 countries for which there is comparable data in 216, as well as in 4 of the projected countries. However, most of the changes were less than 1 percentage point and many were near zero (Figure 2). The most positive results can be seen in Honduras, where poverty decreased almost 3. percentage points, and in the Dominican Republic and Nicaragua. Mexico and Brazil undertook important changes in their 216 household surveys, leading to noncomparable data and a break in their respective poverty series (see Box 1). Both, therefore, had to be projected for inclusion in the regional aggregate. Figure 2: (a) GDP per capita growth in 216 worsens for most countries in the region. (b) Most countries show minor changes in poverty between 21-216 ($./day poverty line) 6% 4% 2% % -2% -4% -6% Dom Rep Nicaragua Panama Costa Rica Paraguay Bolivia Peru Honduras El Salvador Mexico Colombia Uruguay Guatemala Chile Ecuador Argentina Brazil Source: GDP data from WDI; poverty data from SEDLAC (CEDLAS and the World Bank) Note (*) Due to either methodological changes (Brazil and Mexico) or lack of an available survey (Chile, Guatemala, Nicaragua), the 216 regional estimate incorporates projections for these countries. The 21 data is also projected in the case of Argentina, Guatemala, Mexico, and Nicaragua. See Annex 3 for more details. Similarly, the income growth of the bottom 4 percent of the Latin American income distribution was negative in both 21 and 216. The economic slowdown and recent contraction in the LAC region is also evident in the slowing and negative growth of incomes of the bottom 4 percent of the population (Figure 3). In both 21 and 216, household per capita income of those in the bottom 4 percent of the distribution declined by around -.6 percent. Figure 3: Annualized income growth of the bottom 4 percent of Latin America remained negative in 216. 8 6 4 2 21 216 percentage points 2 1-1 -2-3 -4 Honduras Dom Rep (*)Nicaragua (*)Guatemala El Salvador Costa Rica Panama Ecuador (*)Mexico Bolivia (*)Chile Uruguay (*)Argentina Colombia Peru Paraguay (*)Brazil -2-4 26 27 28 29 2 211 212 213 214 21 216 Bottom 4 income growth (annual %) GDP p/cap growth (annual %) Source: World Development Indicators and SEDLAC data (CEDLAS and the World Bank) 2 JUNE 14, 218

BOX 1. New Poverty Series for 216 in Brazil and Mexico In 216, both the National Statistics Office in Brazil and in Mexico introduced improvements to the latest edition of their household surveys. These updated surveys aim to better capture income, among several other changes. Values of income produced in 216 from the new version of the household surveys are not directly comparable with the historical series. Therefore, World Bank monetary poverty rates using international lines derived from these data should not be compared with pre-216 numbers. Brazil In Brazil, national poverty rates have traditionally been estimated using the household survey PNAD (Pesquisa Nacional por Amostra de Domicilios), collected by IBGE (Instituto Brasileiro de Geografia e Estatística). The PNAD used to be collected every year during the month of October, and included 147 thousand households and 1 municipalities. In 216 the survey was completely replaced by a new survey, the PNAD-Continua. The PNAD-C is collected throughout the year, it is representative at every quarter, and it collects information from 211 thousand households, in 3 municipalities. The PNAD-C has been collected since 212. However, the nonlabor income has so-far been made public only for 216. Thus, it is impossible to assess how the measure of household welfare (total per-capita household income) compare across the two instruments. IBGE is planning to release the full series of non-labor income since 212 in mid-218. While the team waits for these, any meaningful analysis of the dynamics of poverty and inequality between 21 and 216 cannot be undertaken. Mexico In Mexico, official poverty estimates are produced and released to the public every two years by CONEVAL (Consejo Nacional de Evaluación de la Política de Desarrollo Social) based on data from the household survey MCS-ENIGH (Módulo de Condiciones Socioeconómicas Encuesta Nacional de Ingreso y Gasto de los Hogares), which is generated by INEGI (Instituto Nacional de Estadística y Geografía). Since 28, Mexico has been using an official multidimensional measure of poverty as the official poverty measurement, which combines monetary poverty (based on current income per adult equivalent) with non-monetary dimensions of well-being, called social deprivations. For comparisons across countries, however, the SEDLAC regional data harmonization effort uses only an income-based welfare measure. The main differences of the 216 survey compared to previous years are: (1) a larger sample size to allow urban/rural disaggregation at the state level; (2) both income and expenditures were collected for all households; and (3) field operation was improved, particularly better training for interviewers and an automated system to monitor data collection. Including the current surveys as they are without adjustments would be misleading, as there exist large differences with the surveys from previous years. While the team waits for the 217 and 218 household survey data, which would allow us to create a new series for each country, we have chosen to project poverty for Brazil and Mexico using changes in per capita consumption from National Accounts and changes in population from WDI. This results in a smaller increase in poverty in Brazil and a much smaller decrease in poverty in Mexico in 216 as compared to poverty measures produced by the country. 3 JUNE 14, 218

At the country level, the downturn in incomes of the bottom 4 percent was strongest in Brazil. Annualized income growth of the bottom 4 percent between 214 and 216 was generally positive across the Latin America region, with the exception of Brazil (214-21 data), Paraguay, Ecuador, and Bolivia (Figure 4). Countries with positive overall household income growth between 214-216 tended to also experience positive income growth for the bottom 4 percent of their population, and usually by higher rates. On the other hand, the negative impact on the growth of income of the bottom 4 percent tended to be less adverse than for the rest of the population in those countries facing negative overall growth rates. Panama experienced one of the highest income growth rates of both the bottom 4 percent and the overall population (around 7 percent). In the other extreme, Brazil experienced the lowest income growth rate of both populations (around. percent). Figure 4: Household income growth between 214-216 varied across the region 8 6 4 2-2 -4-6 -8 Panama Dom Rep Argentina El Salvador Colombia Costa Rica Peru Honduras Uruguay Bolivia Ecuador Income growth Bottom 4% Income growth overall population Paraguay Brazil (*) Source: SEDLAC and own calculations Note: Almost all estimates are based on data between 214 and 216. Brazil s data is for 214-21. However, given the continued economic contraction in the country in 216, it is likely that the growth rate looks similar for the 214-216 period. Nevertheless, inequality in Latin America continued to decline in 216. The Gini coefficient decreased from.8 in 21 to. in 216. This was due to a continued and larger income drop among the richest households in the income distribution. Although per capita income for the bottom 4 percent of the population in Latin America declined by.6 percent, per capita household income for the richest 2 percent decreased by 2.4 and 2.2 percent, respectively, in 21 and 216. Figure : Inequality in Latin America continued declining in 216 (Gini coefficient) 4 4 3 3 2 2 1 1 49 26 27 28 29 2 211 212 213 214 21 216 Source: SEDLAC and own calculations The income decline in the region would have increased poverty further, but it was partially offset by changes in the distribution of income. Per capita household income declined between 21 and 216 by -2.4 percent in the region as a whole. A growthdistribution decomposition of the small (.2 percentage point) increase in poverty between 21-216 shows that the contraction in mean income was the more important factor, pushing poverty up, but it was partly offset by changes in its distribution (Figure 6). The region in 216 would have faced a higher poverty increase of.6 percentage points if only the contraction in mean income had occurred. Taking a long-run view, the annualized decomposition results for the 26-216 decade suggest both income growth and its distribution helped decrease poverty. Income growth, however, continued to have the more important impact on the observed, annualized, 1.1 percentage point decrease in poverty over the decade. Lastly, whether in the current economic contraction or in the more positive past decade, the distributional impact is about the same in these cases. 4 JUNE 14, 218

Figure 6: The distribution of income is less important than income growth both now and over the past decade, but it partially offset the negative impacts of recent income declines. percentage points 1.2.8.4 -.4 -.8-1.2.6 Growth effect Current economic -.4 -.4 Poverty -.7 increase 21-216 Distribution effect.2 Past decade Source: SEDLAC and own calculations. Footnote: Shapely decomposition of changes in poverty into a growth and a distribution effect, building on Datt-Ravallion (1992). See Azevedo and Sanfelice (212) and Ferreira et al. (212). -1.1 Poverty decrease 26-216 Drilling down further, the negative income growth for the Latin America region is driven primarily by the economic contraction experienced in Brazil. In the 214-216 period, annualized household per capita income growth in Latin America averaged -1.7 percent (Figure 7). However, excluding Brazil, the Latin America region s household per capita income grew at an annualized positive rate of 1.3 percent. In addition, after also excluding Mexico, the region experienced pro-poor income growth of 1.4 percent, with growth as high as 2.1 pecent for the lowest quintile (Figure 7). Figure 7: Income growth in Latin America is positive and pro-poor if Brazil is excluded Change 214-216 (%) 3 2 1-1 -2-3 -4-6 Quintiles of Per Capita Household Income LAC Brazil Note: Brazil and Mexico 216 are projections. 1.4 1.2-1.7.2 LAC ex-bra&mex Mexico Mean growth Although half the region experienced on average low or negative income growth between 214-216, countries such as Uruguay and Colombia nevertheless experienced positive income growth of their poorest 4 percent of the population, improving the results for the region (Annex 4). Colombia, in particular, makes up about percent of the poor of Latin America, the third highest after Brazil and Mexico. Although Colombia had an overall negative income per capita growth rate (-1.9 percent), the lowest deciles experienced positive income growth of 2.8 and 1.4 percent between 214-216 (Annex 4), impacting in a beneficial way the poverty and inequality results for Latin America. The overall economic slowdown impacted Colombia s urban labor market, increasing unemployment rates during this period, while rural areas, where most of the poor live, saw an increase in both labor and nonlabor incomes in the same period. These differentiated urban-rural impacts need further investigation. The other half of the Latin American countries experienced income growth higher than 2 percent, however this growth was not pro-poor. Argentina, the Dominican Republic, Honduras, Peru, and Panamá experienced overall positive income growth (as did the projected estimates for Chile, Guatemala, and Nicaragua) (Annex 4). The Dominican Republic and Panama had average annualized growth rates of 8. and 7 percent, respectively. However, among all JUNE 14, 218

these countries, the first deciles generally experienced lower growth than the middle or higher deciles. In fact, in Honduras, the first two deciles had significant decreases in mean income while the rest of the population experienced a relatively high increase in income. The fact that poverty rates remained unchanged in Honduras is probably due to the increase in income received by people in the 2 nd and 3 rd quintile of the income distribution, a population that still forms part of the, relatively high, poverty rates in the country. Preliminary data are ambiguous over the likely poverty outcomes in 217. The Labor Income Poverty Index (LIPI), which is based on the most recently available quarterly data on labor incomes, provides preliminary information regarding possible poverty trends in LAC. 4 Figure 8 shows the LIPI indexed at 1 in 216. The upward sloping lines from 21 to 216 reflect increases in poverty based solely on labor incomes. In fact, as seen in 21 (see World Bank, 217), poverty did not increase the last two years thanks to growth in non-labor income that compensated for the declines in labor income. The LIPI between 216-217 shows generally flatter slopes than between 21-216 for most countries. This is particularly the case in Argentina and in Brazil. Nevertheless, the preliminary 217 LIPI results (most of which are currently Q2 and Q3) do suggest a possible small increase in labor income-based poverty in 217. Growth remains a critical component for poverty reduction, however, and the region seems to be pulling out of its economic contraction in 217. If the better outlook for economic growth holds, and if non-labor income continues to play a poverty-reducing role as in the past, then 217 may turn out to be the beginning of a positive turn-around in the region. Figure 8: The Labor Income Poverty Index (LIPI) suggests few poverty changes in 217 ($. per day, 211 PPP) Source: Own calculations using LABLAC (CEDLAS and the World Bank). 4 This index uses the fact that, in LAC, changes in labor income and poverty are highly correlated to estimate poverty movements based on the more frequently and recently available labor force survey data. In the case of Brazil, the LIPI series is based on a comparable labor income module in the new PNAD- Continua survey that extends back to 212, providing a longer set of comparable data. The other data currently available in the LIP includes: Argentina, Costa Rica, Ecuador, Mexico, Paraguay and Peru. 7 JUNE 14, 218

Annex 1. About this Brief This brief was produced by the Latin America and Caribbean Team for Statistical Development (LAC TSD) in the Poverty and Equity Global Practice of the World Bank (March 218). The core team was led by Carolina Díaz- Bonilla and Andrés Castañeda and included Laura Moreno, Jorge Soler, Natalia Garcia-Peña, and Pamela Gunio. The team worked under the guidance of Oscar Calvo-Gonzalez. The numbers presented in this brief are based on a regional data harmonization effort known as SEDLAC, a joint effort of the World Bank and CEDLAS at the National University of La Plata in Argentina (see Annex 4 for the list of surveys used in this brief). They increase cross-country comparability of selected findings from official household surveys. For that reason, the numbers discussed here may be different from official statistics reported by governments and national offices of statistics. Such differences should not be interpreted in any way as a claim of methodological superiority, as both sets of numbers serve the same important objectives: regional comparability and the best possible representation of the facts of individual countries. Indicators for LA are calculated using data from Argentina, Bolivia, Brazil, Chile, Colombia, Costa Rica, Dominican Republic, Ecuador, El Salvador, Guatemala, Honduras, Mexico, Nicaragua, Panama, Paraguay, Peru, and Uruguay (LA-17). JUNE 14, 218

Annex 2. Household surveys used from SEDLAC harmonization Country Name of survey Circa 216 Coverage Argentina Encuesta Permanente de Hogares- Continua 216 Urban-31 Cities Bolivia Encuesta Continua de Hogares- MECOVI 216 National Brazil Pesquisa Nacional por Amostra de Domicilios 21 National Chile Encuesta de Caracterización Socioeconómica Nacional 21 National Colombia Gran Encuesta Integrada de Hogares 216 National Costa Rica Encuesta Nacional de Hogares 216 National Dominican R. Encuesta Nacional de Fuerza de Trabajo 216 National Ecuador Encuesta de Empleo, Desempleo, y Subempleo 216 National El Salvador Encuesta de Hogares de Propósitos Múltiples 216 National Guatemala Encuesta Nacional de Condiciones de Vida 214 National Honduras Encuesta Permanente de Hogares de Propósitos Múltiples 216 National Mexico Encuesta Nacional de Ingresos y Gastos de los Hogares 214 National Nicaragua Encuesta Nacional de Hogares Sobre Medición de Niveles de Vida 214 National Panama Encuesta de Hogares 216 National Paraguay Encuesta Permanente de Hogares 216 National Peru Encuesta Nacional de Hogares 216 National Uruguay Encuesta Continua de Hogares 216 Urban- Montevideo and Interior >, inhabitants JUNE 14, 218

Annex 3. International Poverty Rates by Country Poverty Rates at USD. a day (%) - 211 PPP Poverty Rates at USD 3.2 a day (%) - 21 PPP 212 213 214 21 216 212 213 214 21 216 Argentina 8. 8.3 9.1 7.8 2.7 2.4 2.8 2.4 Bolivia 28.3 26. 24.9 2.6 2. 14.7 12.2 12.3 11.7 12.6 Brazil 2.8 19. 17.9 19.4 8.7 8.2 6.9 8. Chile.1.1 2.6 3.1 Colombia 34.2 32.1 29.8 28.7 28. 1.7 14.3 13.2 11.9 11.8 Costa Rica 12.3 12.3 11.9 11..7 4.2 4. 3.8 4. 3.8 Dominican Republic 31.6 3.8 26. 22.7 21. 11.. 9.9 7. 8.2 Ecuador 27. 26.8 23.8 24.7 23.9 11.4.1 8.6 9.4 9.4 El Salvador 38.9 3. 34.9 31.8 3.7 14.4 12.4 12.2 9.9.3 Guatemala 48.8 24.2 Honduras 7.1.3 3.9 3.8.4 36.1 32.9 31.8 32.2 3. Mexico 31.8 33.6 12.2 12. Nicaragua 34.8 12.8 Panama 18.9 18.3 17. 1.4 14.6. 8.1 8.4 6.7 7. Paraguay 23. 2.4 19.3 2. 2.1 9.9 6.7 7. 6.9 7. Peru 26.8 26.1 2.2 24.3 24.3 12.3 11.4.7.3. Uruguay.. 4.3 4. 3.7 1.1.8.7.6. LAC 26.2 2.4 24.4 24. 24.6 11.1.4 9.7 9.8 9.9 Note: The LAC regional estimate is calculated using pooled country-specific data. When household surveys are unavailable for a specific year and country, estimates (inter- or extrapolations) for that country-year are done using data from the World Bank s World Development Indicators (WDI). In particular, consumption per capita growth rates (National Accounts household final consumption expenditure) are used to adjust survey incomes and population growth rates are used to adjust survey weights. JUNE 14, 218

Annex 4. Growth Incidence Curves by Country (214-216, %) Countries with positive average growth in per capita household income Dominican Republic Panama 1 2 3 4 1 2 3 4 Argentina Honduras 1 2 3 4 1 2 3 4 Peru 1 2 3 4 Countries with positive but low average growth El Salvador Costa Rica 1 2 3 4 1 2 3 4 JUNE 14, 218

Countries with overall negative income growth (214-216) Uruguay Ecuador 1 2 3 4 1 2 3 4 Colombia 1 2 3 4 Bolivia Paraguay 1 2 3 4 1 2 3 4 JUNE 14, 218

Countries whose Growth Incidence Curves are based on projected data (214-216) Brazil Mexico 9 4-1 1 2 3 4 1 2 3 4-6 Nicaragua Guatemala 1 2 3 4 1 2 3 4 Chile 1 2 3 4 JUNE 14, 218

References Azevedo, J.P., V. Sanfelice, and M. C. Nguyen. 212. Shapley Decomposition by Components of a Welfare Measure. World Bank (mimeo). Datt, G. and M. Ravallion. 1992. Growth and Redistribution Components of Changes in Poverty Measures: A Decomposition with Applications to Brazil and India in the 198s. Journal of Development Economics, 38, pp 27 9. Ferreira, Francisco H.G.; Messina, Julian; Rigolini, Jamele; López-Calva, Luis-Felipe; Lugo, Maria Ana; Vakis, Renos. (213) Economic Mobility and the Rise of the Latin American Middle Class. Washington, DC: World Bank. Foster, James; Joel Greer; Erik Thorbecke (1984). "A class of decomposable poverty measures". Econometrica. 3. 2: 761 766. Gini, Corrado (1921). "Measurement of Inequality of Incomes". The Economic Journal. Blackwell Publishing. 31 (121): 124 126. Jolliffe, D.; Prydz, E. (216). Estimating International Poverty Lines from Comparable National Thresholds. Policy Research Working Paper 766. World Bank. Ravallion, M., and S. Chen. 23. Measuring Pro-poor Growth. Economics Letters 78 (1): 93 99. Shorrocks, A. F. (212) Decomposition procedures for distributional analysis: a unified framework based on the Shapley value. Journal of Economic Inequality. Shorrocks, A.; Kolenikov, S. (23) A Decomposition Analysis of Regional Poverty in Russia, Discussion Paper No. 23/74 United Nations University. JUNE 14, 218