53546 SOCIAL CONSEQUENCES OF THE GLOBAL FINANCIAL CRISIS IN LATIN AMERICA: SOME PRELIMINARY, AND SURPRISINGLY OPTIMISTIC, CONJECTURES Francisco H.G. Ferreira and Norbert Schady1 November 24, 2009 Abstract Surprisingly, the most severe economic crisis the world has seen since the Great Depression does not appear to have had as dramatic an impact on poverty in Latin America as might have been expected. The exceptions to this heartening assessment are the countries geographically and economically closest to the United States ­ chiefly Mexico. Elsewhere, although poverty statistics for 2008/09 are not yet available, the data on output, unemployment and real wages suggest relatively modest changes in poverty. There are two candidate explanations for the smaller-than-expected increases in poverty in Latin America: lower output declines, deriving from enhanced protection against external shocks; and a lower output elasticity of poverty. If the latter is indeed observed when the required data becomes available, we conjecture that it may reflect both the lower inflation rates now prevalent in the region, and recent reforms in the social protection system. For all their faults, the social protection systems in many LAC countries now reach the poor rather than only the middle-classes. The note concludes arguing against complacency, and pointing to areas where (i) further research; and (ii) greater policy reform and experimentation are needed. 1. Economic crises and poverty in LAC: the historical context Economic crises are not new to Latin America or the Caribbean. Since 1980, twenty-seven countries in the region have had no fewer than 43 (finance-related) recessions, defined as periods of at least one year with negative GDP growth.2 A number of these recessions were extremely severe: Uruguay's GDP 1 This crisis brief draws in part on a presentation we made at the Inter-American Development Bank in October 2009. We are grateful to participants at that event for comments, to Tatiana Didier for giving us access to the LCR financial crisis database and classification, and to Carlos Prada for excellent research assistance. The views expressed here are those of the authors, and do not necessarily represent the views of the World Bank or of the Inter American Development Bank. 2 This classification of crises departed from a database of all debt, banking and currency crises, compiled by the LCR Office of the Chief Economist. These episodes were then consolidated by generating a country/year matrix where an element would be designated a crisis year if it fell under any of the three original compilations. Successive crisis years in a single country were then consolidated as a single crisis episode. All episodes with zero or positive growth were discarded. Finally, if years contiguous to those in the crisis classification also presented negative GDP growth, they were included in the relevant episode. The resulting "crisis matrix" is presented as Annex 1. fell by 19% during a five-year recession in the early 1980s. Argentina's GDP declined 16% between 1999 and 2002. Peru's output fell 12% in 1982-83, and another 16% in 1988-90. Just as economic growth generally leads to poverty reduction, economic contractions are almost always associated with falling earnings and employment, and thus with temporary increases in income poverty.3 Although this is a general phenomenon, poverty has been particularly sensitive to output declines in Latin America. Figure 1 illustrates the sharp upswing in the incidence of poverty during two relatively recent Latin recessions: the Colombian crisis of the late 1990s, and the Argentine crisis of the early 2000s. It plots the per capita GDP and the poverty headcount series (for a poverty line of U$2.50 a day) alongside each other. Figure 1: Poverty is markedly counter-cyclical: Argentina and Colombia, 1991-2007 Source: Author calculations from CEDLAS and World Development Indicators data. The main transmission mechanism from collapses in aggregate demand and output to household incomes is, as in other countries, falling demand for labor which translates, in turn, into higher unemployment, lower real wages, fewer hours of work among those still employed, or any combination of the three. In Latin America (as in other developing countries) the informal sector typically acts as a buffer, expanding during recessions, as it absorbs some ­ but not all ­ of the labor that cannot enter the formal sector.4 Despite this role of the informal sector, however, most recessions in the region in the last two decades have seen both substantial rises in unemployment and declines in real wages. This is shown in Table 1, which presents changes in GDP, open unemployment and real manufacturing wages for all recessions since 1980 in Latin America's six largest economies: Argentina, Brazil, Chile, Colombia, Mexico and Peru. Data is also included for the 2008-2009 period, for comparison.5 3 Recessions also affect deprivation in other dimensions of well-being, but their effects on child health and education are more complex than has often been assumed. The existence of a dominant substitution effect, for instance, implies that school enrollment often rises during recessions. See, e.g. Ferreira and Schady (2009). 4 Falling formal sector employment is often the result of a substantial reduction in hiring in the sector, rather than of greater involuntary labor shedding. Flows into the formal sector are then diverted into the informal sector or unemployment. See Bosch and Maloney (2009). 5 The crisis episodes are taken from the Table in Annex 1, and a full description of the variables and sources is given in the notes to Table 1. 2 Of course, falling labor demand is the main, but not the only transmission channel from a recession to falling household incomes and rising poverty. Lower demand for the goods and services produced by poor self-employed workers is another, as are pro-cyclical social expenditures, from which Latin America has long suffered. See, e.g. Wodon et al. (2003) and De Ferranti et al. (2004). In light of this history, it is natural to ask what the likely effect of the 2008-2009 crisis on poverty in Latin America will turn out to be. The real answer to this question is as yet unknown, since poverty statistics for 2009 are not yet available for any country in the region. No single nationally representative household survey for 2009 is yet publicly available. One way in which available information has been used to estimate the likely poverty impact of the recession is to combine current growth forecasts for 2009 with past estimates of the growth elasticity of poverty reduction in the region. Using this approach, the Latin America Poverty and Gender Group (2009) estimated in March that the incidence of extreme poverty (using a poverty line of $2-a-day) in the region as a whole might increase by between 0.05 and 0.53 percentage points, i.e. by between 0.5% and 6%, in 2009. The incidence of moderate poverty ($4-a- day) was forecast to rise between 0.07 and 1.16 percentage points, or 1.5% and 5.5%.6 A recent update by Latin America Poverty and Gender Group using more recent (October) consensus growth forecasts suggests somewhat higher poverty increases, a result which appears to be driven to a large extent by worsening growth prospects for Mexico. All such prognoses rely on the assumption that average estimates of the growth elasticity of poverty, obtained both from past downturns and expansions, are a good guide for the future. As the authors are careful to acknowledge, this is by no means guaranteed.7 The incidence and pattern of growth and recession spells can be remarkably different, and lead to great instability in a highly aggregated summary index like the "growth elasticity". See, e.g. Bourguignon, Ferreira and Lustig (2005). In this note, we follow a different route, and consider some of the most recent evidence on both the labor market and social expenditure trends in Latin America's six largest economies. We do not attempt to predict poverty changes in 2009 numerically, but simply speculate more generally about the likely social effects of the 2008-2009 global financial crisis in Latin America. Our approach is descriptive, and should be seen as complementary to that of Latin America Poverty and Gender Group (2009), and others.8 2. The social impact of the 2008-2009 global financial crisis. In addition to confirming that recessions in LAC have typically been associated with increases in open unemployment and ­ often very large ­ declines in real wages, Table 1 permits an interesting comparison between the effects of the global financial crisis of 2008-09 and previous recessions in Latin America's major economies. Three features seem noteworthy. 6 If compared to the trend decline in poverty during 2002-2008, the "forgone poverty reduction" is larger. 7 Indeed, they note that "some of the factors that appear to have contributed to the recent significant gains in poverty between 2002 and 2008... were either not present or at much lower levels during ... previous downturns." (LCSPP, 2009, p.9) 8 For a recent discussion of the response of Latin American labor markets to the crisis, see Freije-Rodríguez and Murrugarra (2009). 3 I. Based on the October 2009 Consensus Forecasts for economic growth, it appears that the global financial crisis did not translate into a severe recession in most of Latin America. The clear exception to this statement among the LAC 6 economies in Table 1 is Mexico, where the expected 7% contraction in GDP would make this the severest downturn since 1980. In the region more broadly, Caribbean and Central American economies are also expected to suffer sizable GDP declines. This reflects these countries' greater integration with the US economy, both in terms of trade and remittances. In the other five countries in Table 1 (as in most of South America), the recession turns out to have been relatively mild, with growth still in positive territory in Brazil and Peru, and limited contractions in Colombia and Chile. II. The latest monthly or quarterly unemployment data shown in column 5 is consistent with that relatively benign picture. Outside Mexico (and Chile), unemployment has increased fewer than two percentage points in all LAC 6 economies. Chile's 3.2 pp. increase is a downside exception, particularly since it implies a much higher employment elasticity with respect to output. Whether that reflects differences in labor absorption by Chile's informal sector, or other institutional differences, remains a question for future research. III. Real wage declines in manufacturing during 2008-2009 were also subdued in the LAC 6 economies, including Mexico. The largest decline was 5.3% in Brazil. This compares with declines in excess of 20% in a number of past episodes, including a 45% decline in Brazil (in 1990) and an 80% decline in Peru (in 1988-1990). The magnitudes of the real wage declines from the late 1980s and early 1990s may be somewhat suspect, given the measurement challenges in hyperinflationary contexts. Nevertheless, it appears certain that manufacturing real wages were much less affected during 2008-09 than in previous crises. This clearly reflects the more moderate nature of this downturn, but it is possible to speculate that, even controlling for changes in GDP, real wages may have become stickier in a low- inflation environment. Turning from labor market to social expenditure trends, there is also reason to suspect that the policy response to this last crisis has been less pro-cyclical than in previous episodes. Five of the six countries in Table 1 implemented "fiscal stimulus packages", consisting of additional expenditures intended explicitly to sustain aggregate demand. All six countries also put in place a series of "emergency social measures", designed to protect the incomes of the poor, and of those most vulnerable to fall into poverty. Such measures have included direct support to labor demand, as in the case of Chile's "youth wage subsidy", or Mexico's "Employment Preservation Program". But in many other cases the measures consisted of expansions in existing anti-poverty programs that are less directly linked to formal employment, such as conditional cash transfers or workfare schemes. Examples include the extension of Brazil's Bolsa Família to another estimated 1.3 million families (by raising the income means-test) and the increase in the value of the benefit. Colombia also increased the coverage of its CCT program, Famílias en Acción, and Mexico has expanded its "Temporary Employment Program" and is currently discussing a reform of Oportunidades. Table 2 presents a selective summary of these policy responses for the LAC 6 countries. 4 3. Moving forward, in both research and policy. As noted above, it is clearly too early to claim that the poverty impact of the global crisis of 2008-2009 on Latin America was small. Nevertheless, both the labor market and social policy information available to date suggests that this is a plausible hypothesis, deserving of further investigation once the relevant household survey data become available. If, at that time, the conjecture is borne out by the data, this will have been a remarkable outcome. The 2008-2009 global financial crisis is now widely regarded as the most severe global economic downturn since the Great Depression. And Latin America is generally seen as a highly volatile region, where internal output fluctuations tend to amplify, rather than dampen, external stimuli, and where poverty responds acutely and pro-cyclically. What was different this time? Answers to this question will have to wait for the poverty data to become available. But there are two likely sets of candidate explanations. First, the "transmission at the border" from external trade, investment and remittance shocks to domestic activity has become less direct, at least in South America. This is likely to reflect (i) greater diversification in trading partners; and (ii) greater space for both fiscal and monetary policy responses. A recent Crisis Brief has argued that fiscal policies in the last two years may not yet be properly anti-cyclical in most countries in the region, but at least they are less pro- cyclical than they used to be.9 In addition, stabilization funds, such as Chile's Social and Economic Stabilization Fund, have also contributed to dampening external volatility "at the border", partly insulating the domestic economy from external shocks. Finally, the region has now largely stopped relying on one of the most regressive of all taxes, namely the inflation tax. Where its effect has been studied, it has generally been found that the elimination of high inflation has been substantially pro- poor.10 Second, social policies may have played a part in dampening the transmission of that part of the aggregate shocks which did reach the domestic economy to the livelihoods of the poor. While this too remains speculative until more data are available, it is not implausible. The expansion in the coverage of social assistance transfers across Latin America in the last ten to fifteen years has been substantial. While Latin America's social protection system in the 1990s was described as a "truncated welfare state" (De Ferranti et al. 2004), a battery of new and expanded social programs now have much greater penetration among the poor. These famously include a range of conditional cash transfers (Fiszbein and Schady, 2009), but also various other non-contributory pension and health insurance schemes (Walker, Robalino and Ribe, 2009). Not only did these programs provide a measure of automatic stabilization which LAC economies typically lacked, they were also much better targeted than in the crises of the 1980s and 1990s, protecting a larger number (although far from all) of those already in or near poverty. And, as Table 2 indicates, in many cases they provided a set of existing programs, with ready-to-use beneficiary rosters and payment systems, that could be expanded (in a discretionary way) as the crisis evolved. These conjectures suggest a number of questions for policy research in LAC, once more data for 2009 and beyond become available. But even if the rosiest possibilities raised here turn out to be true, they 9 Calderon and Fajnzylber (2009). 10 For example, on the case of Brazil, see Ferreira et al. (2008) and Ferreira et al. (in press). 5 should not be seen as providing any room for complacency. The performance of existing social protection instruments during the 2008 and 2009 crisis also revealed a number of serious shortcomings, including: The absence of programs that can successfully insure informal sector workers (who do not qualify for unemployment insurance, or severance payments) from loss of employment or sharp income reductions. This is a role often played by public employment or workfare schemes in other regions. In LAC "temporary employment schemes" have tended to be both small and regarded with skepticism. Why is that, and is it justified? The inefficiencies created by the dual nature of LAC's social protection system grow with its size. While traditional programs tied to formal sector jobs are generally financed by payroll taxes or other employment-related contributions, many of the new targeted, non-contributory programs which reach a larger fraction of the poor are financed out of general tax revenues. This introduces a de-facto subsidy to informality, financed by a tax on formal sector firms, and likely leads to a misallocation of both capital and labor towards the informal sector (Levy, 2008). Although the provision of basic safety nets has been an important achievement in some countries (while remaining woefully inadequate in others), much more can be done in terms of providing opportunities for and encouraging the poor to invest in long-term upward mobility. The "opportunity-ropes" agenda that began with microfinance and conditional cash transfers, and is expanding with the current enthusiasm for early childhood development, still has much room to develop. In short, the lessons from the region's apparent success in cushioning the poor and vulnerable from the impacts of a severe world recession are twofold: (i) Improved policymaking works ­ both at the macroeconomic and social policy levels; and (ii) there is still a long way to go, and no time to waste in improving the effectiveness and efficiency of the region's social protection systems. References Bosch, Mariano and William Maloney (2009): "Cyclical movements in unemployment and informality in developing countries". World Bank, LCR Office of the Chief Economist, unpublished. Bourguignon, François, Francisco Ferreira and Nora Lustig (2005): The Microeconomics of Income Distribution Dynamics in East Asia and Latin America. Washington, DC: Oxford University Press and the World Bank. Calderón, Cesar and Pablo Fajnzylber (2009): "How much room does Latin America and the Caribbean have for implementing counter-cyclical fiscal policies?" LCR Crisis Brief (April), The World Bank. De Ferranti, D., G. Perry, F. Ferreira and M. Walton (2004): Inequality in Latin America: Breaking with History?, Washington, DC: The World Bank. Ferreira, Francisco, Phillippe Leite and Julie Litchfield (2008): "The rise and fall of Brazilian inequality: 1981-2004", Macroeconomic Dynamics 12 (S2): 199-230. 6 Ferreira, Francisco, Phillippe Leite and Martin Ravallion (in press): "Poverty reduction without economic growth? Explaining Brazil's poverty dynamics, 1985-2004", Journal of Development Economics. Ferreira, Francisco and Norbert Schady (2009): "Aggregate Economic Shocks, Child Schooling and Child Health", The World Bank Research Observer 24 (2): 147-181. Fiszbein, Ariel and Norbert Schady (2009): Conditional Cash Transfers: Reducing Present and Future Poverty. Washington, DC: The World Bank. Freije-Rodríguez, Samuel and Edmundo Murrugarra (2009): "Labor markets and the crisis in Latin America and the Caribbean: A preliminary review for selected countries". LCR Crisis Brief (June), The World Bank. Latin America Poverty and Gender Group (2009): "How has poverty evolved in Latin America and how is it likely to be affected by the economic crisis?" World Bank, LCSPP, unpublished. Levy, Santiago (2008): Good intentions, bad outcomes: Social Policy, informality and economic growth in Mexico. Washington, DC: The Brookings Institution Press. Walker, Ian, David Robalino and Helena Ribe (forthcoming): From right to reality: Achieving social protection for all in Latin America. Washington, DC: The World Bank. Wodon, Q., N. Hicks, B. Ryan and G. Gonzalez (2003): "Are governments pro-poor but short-sighted?: Targeted and social spending for the poor during booms and busts" in Quentin Wodon (ed.) Public Spending, Poverty and Inequality in Latin America. Washington, DC: The World Bank. 7 Table 1: GDP, unemployment and real wage changes during recessions, LAC 6: 1980-2009 Change Country Event Crisis Duration GDP Unem (p.p) Real Wage (1) 1980 - 1982 -10.36% 3.40 N.A (2) 1985 -7.59% 1.90 N.A (3) 1988 - 90 -9.72% 2.00 N.A Argentina (4) 1995 -2.85% 6.20 N.A (5) 1999 - 2002 -15.50% 7.70 -26.79% (6) 2008 - 2009 -2.66% 1.00 -3.09% (1) 1981 - 83 -7.82% 1.56 N.A Brazil (2) 1990 -4.30% 1.95 -44.90% (3) 2008 - 2009 0.14% 1.30 -5.33% (1) 1980 - 1983 -9.63% 9.2* N.A Chile (2) 2008 - 2009 -1.50% 3.20 -3.04% (1) 1997 - 1999 -3.66% 7.80 -1.92% Colombia (2) 2008 - 2009 -0.10% 1.20 -2.83% (1) 1981 - 1983 -4.80% 2.6* N.A (2) 1985 - 1986 -3.75% 1.00 N.A Mexico (3) 1994 - 1995 -6.22% 3.80 -22.45% (4) 2008 - 2009 -7.10% 2.08 -2.65% (1) 1982 - 1983 -11.80% 2.4* N.A Peru (2) 1988 - 1990 -16.24% 3.5* -80.22% (3) 2008 - 2009 1.03% 1.60 -2.81% Notes: Gross Domestic Product (GDP) is the growth rate for the crisis episode. It is the change between years t and t+k of the cumulative value of GDP. For example, during the first crisis episode for Argentina, the growth rate is equivalent to the cumulative value between 1980 and 1982. Frequency: Annual Data. Source: World Development Indicators - World Bank. Consensus Forecast (October 2009). Unemployment Rate (pp) is the largest change, in percentage points, observed during the crisis episode, except for starred entries (*) which denote total changes over the entire interval. Frequency: Quarterly data for Argentina and Colombia; Monthly for Brazil, Chile, Mexico and Peru. Source: Argentina (INDEC-Encuesta Permanente de Hogares); Brazil (IBGE-Pesquisa Mensal de Emprego); Chile (Banco Central de Chile - CASEN); Colombia (Banco de la Republica - Encuestas de Hogares); Mexico (INEGI - Encuesta Nacional de Empleo Urbano and Encuesta Nacional de Ocupación y Empleo); Peru (INE - Encuesta Permanente de Empleo). The data for Peru includes only Metropolitan Lima. For 2009, the latest information refers to June in Argentina and September elsewhere. Real Wage is the real change of monthly wages in the manufacturing sector during the crisis episode (all wages were deflated with domestic CPI). Only hourly wages were available for Chile. The data for Peru includes only Metropolitan Lima. Frequency: Quarterly data for Argentina; Monthly for Brazil, Chile, Colombia, Mexico and Peru. Source: Argentina (INDEC-Encuesta Industrial Mensual); Brazil (IBGE-Pesquisa Industrial Mensual); Chile (Banco Central de Chile-Encuesta Mensual Industrial); Colombia (Banco de la Republica-Muestra Mensual Manufacturera); Mexico (INEGI-Encuesta Industrial Mensual); Peru (Ministerio de Trabajo y Promoción del Empleo). 8 Table 2: Selected fiscal and social policy responses to the 2008-2009 crisis, LAC 6 Total fiscal Country stimulus Some Social Measures Source package Argentina US$4.4 billion, 1) Subsidy of 10% of labour cost for 12 months extendable by a UNDP, 1.27% of GDP further 12 months (at 5%). ECLAC 2) Promotion of worker formalization (through incentives). 3) Plan to create 100.000 job places. Brazil US$ 8.67 1) Increase "Bolsa Familia" transfer amount. UNDP, billion, 0.5% 2) Expansion of Social Program "Bolsa Familia" to an additional 1.3 ECLAC of GDP million families. 3) Extension of unemployment insurance for fired workers from December 2008. Chile US$4 billion, 1) Employment subsidy for low-wage young workers, as well as UNDP, 2.2% of GDP additional cash transfers to low income households. Payment of ECLAC US$ 70 per family dependent made available for most vulnerable households in March 2009. 2) Extension of Unemployment Solidarity Fund to provide access to all unemployed workers. Colombia NA 1) Total Government investment in public works in 2009 will reach ECLAC over US$ 2.4 billion. 2) Increase number of families covered by Famílias en Acción by 1.5 million. Social programs estimated to grow by 42% in 2008. 3) Growth of social programs of 42% compared to 2008, which spent $ 910 million Mexico US$13.3 1) The temporary employment program at the federal level was UNDP, billion, 1.49% expanded by 40% over what had been planned, bringing it up to US$ ECLAC of GDP 160 million in 2009. 2) US$ 140 million earmarked under the Employment Preservation Program for protecting employment in vulnerable businesses. 3) Support to unemployed urban workers (US$110 per month) for a period of four to six months through the Urban Temporary Employment Program. Peru US$4 billion, 1) Special retraining program to support the reintegration of workers UNDP, 3.2% of GDP who lose their jobs. ECLAC 2) Additional resources invested in maintaining and equipment for education and health institutions, and social program budgets. 9 Annex 1: Economic Crisis Episodes in Latin America, 1980-2008 Country 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2008 Antigua and Barbuda Argentina -10.4% -7.59% -9.72% -2.85% -15.5% Belize Bolivia -1.85% Brazil -7.82% -4.30% Chile -9.63% Colombia -3.66% Costa Rica -2.26% Dominica Dominican Republic -2.12% -5.45% -0.25% Ecuador -0.57% -6.30% Grenada -6.53% Guatemala Guyana -13.19% Haiti -3.43% -13.19% -3.95% -1.04% Honduras -1.89% Jamaica Mexico -4.80% -3.75% -6.22% Nicaragua -1.72% -0.09% -0.39% Panama -4.49% -13.38% Paraguay -3.72% Peru -11.80% -16.24% St. Vincent and the Grenadines Suriname -0.49% -0.87% Trinidad and Tobago -3.28% -0.83% -1.45% Uruguay -18.71% -9.10% Venezuela, R.B. -2.38% -8.86% Source: This classification of crises departed from a database of all debt, banking and currency crises, compiled by the LCR Office of the Chief Economist. These episodes were then consolidated by generating a country/year matrix where an element would be designated a crisis year if it fell under any of the three original compilations. Successive crisis years in a single country were then consolidated as a single crisis episode. All episodes with zero or positive growth were discarded. Finally, if years contiguous to those in the crisis classification also presented negative GDP growth, they were included in the relevant episode. 10