2 Lessons from Past Crises—and How the Current Crisis Differs

International Monetary Fund
Published Date:
April 2010
  • ShareShare
Show Summary Details

Historically, periods of sharp contraction have been extremely harmful for human development. Social indicators tend to deteriorate rapidly during economic downturns and improve slowly during economic booms. Moreover, vulnerable groups, such as children and women, are more exposed to the effects of growth volatility.

The asymmetric response of social indicators to the growth cycle likely results from contractions associated with conflict or weak institutions that impair government services. With global crises, donor spending may also come under pressure.

There are several reasons, however, why this crisis may be different from previous crises for low-income countries—social spending has been largely protected so far; precrisis policies and institutions were better; and external shocks, not domestic policy failures, were the main cause of the current crisis for developing countries.

Nonetheless, the impact on the Millennium Development Goals (MDGs) is worrisome. In particular, several rapid and qualitative assessments find that households are already making painful adjustments, particularly in middle-income countries.

How growth volatility affects human development and gender indicators

It is commonly observed that human development indicators deteriorate during growth downturns. Also true, but more difficult to calculate, is that deteriorations in human development indicators during downturns tend to exceed improvements during economic booms (Box 2.1 explains the definition of growth cycles used here). For example, life expectancy is 2 years longer during growth accelerations than the overall average, but 6.5 years shorter during decelerations (figure 2.1). Infant mortality is 8 per 1,000 lower during accelerations, and 24 per 1,000 higher during decelerations. The primary school completion rate is 4 percent higher during accelerations but 25 percent lower during decelerations. Further evidence for asymmetry is the size of correlation coefficients relating social indicators with upturns and downturns (table 2.1). In general, the correlation between social indicators and periods of deceleration is stronger than the correlation between social indicators and periods of acceleration (for details, see annex 2A.1).

FIGURE 2.1Key human development and gender indicators plummet from their overall mean during growth decelerations, all countries

Source: World Bank staff calculations based on data from World Development Indicators. See annex Table A2.1 for levels and Arbache, Go, and Korman (2010) for more details.

Note: Differences of sample averages during growth accelerations and decelerations from overall sample means.

Table 2.1Correlation coefficients between growth acceleration and deceleration and human development indicators
Growth accelerationGrowth deceleration
IndicatorCoefficientSignificance levelCoefficientSignificance level
Life expectancy at birth, women (years)0.13**-0.22**
Life expectancy at birth, men (years)0.12**-0.25**
Life expectancy at birth, total (years)0.13**-0.23**
Infant mortality rate (per 1,000 live births)-0.17**0.21**
Child mortality under five rate (per 1,000)-0.15**0.22**
Primary completion rate, girls (% of relevant age group)0.16**-0.26**
Primary completion rate, boys (% of relevant age group)0.13**-0.23**
Primary completion rate, total (% of relevant age group)0.16**-0.26**
Ratio of girls to boys, primary enrollment0.17**-0.22**
Ratio of girls to boys, secondary enrollment0.1**-0.19**
Ratio of women to men, tertiary enrollment0.06-0.18**
Source: World Bank staff calculations based on data from World Development Indicators.Note: Tests for differences in the means of these variables among growth accelerations, decelerations, and all country-year observations show that they are almost all statistically significant at the 1 percent level (**).

BOX 2.1Defining growth cycles in developing countries

The historical growth patterns considered in this study are derived from a dataset for 163 countries covering 1980–2008. A growth acceleration episode meets three conditions for at least three consecutive years:

  • The four-year forward-moving average growth rate minus the four-year backward-moving average growth rate exceeds zero for each year.

  • The four-year forward-moving average growth rate exceeds the country’s average growth rate, meaning that the pace of growth during accelerations is faster than the country’s trend. Thus the definition of episodes of growth acceleration (or deceleration) is endogenous to each country’s long-run rate of growth.

  • Average GDP per capita during the four-year forward-moving period exceeds the average during the four-year backward-moving period, ensuring that the growth acceleration episode is not a recovery from a recession.

A growth deceleration episode meets these three conditions in reverse. The framework is from Arbache and Page (2007), which extends the methodology in Hausmann, Pritchett, and Rodrik (2005) by examining both accelerations and decelerations and by making each country’s long-run growth rate endogenous. Testing the sample means of development indicators for significant differences during periods of growth acceleration and deceleration can show whether countries that experience more growth fluctuations face slower progress on the MDGs and identify how growth cycles affect changes in development indicators.

Source: Arbache and Page 2007; Arbache, Go, and Korman 2010.

Economic downturns also have a disproportionate impact on girls relative to boys. Life expectancy of girls and boys increases by two years during good times but decreases by about seven years for girls and six years for boys during bad times. The primary education completion rate rises 5 percent for girls and 3 percent for boys during good times but decreases 29 percent for girls and 22 percent for boys during bad times. The female-to-male enrollment ratios for primary, secondary, and tertiary education rise about 2 percent during growth accelerations but fall 7 percent (primary), 15 percent (secondary), and 40 percent (tertiary) during decelerations. These differences may result from household time and resource allocations that favor boys over girls when household budgets shrink.1 The differential impact on child schooling and child survival is greatest in low-income countries, while gender differences are smaller in middle-income countries. Economic downturns also have different effects on the labor force participation of women and men, with important implications for how families adjust to economic crises (Box 2.2).

Despite some commonalities, the relationship between growth volatility and development outcomes varies across countries and regions. Initial conditions, regional spillovers, trade arrangements, economic geography, and other factors are associated with how countries and regions respond to economic downturns. For example, human development indicators in Sub-Saharan Africa are among the lowest in the world: infant and under-five mortality rates are almost three times higher than the global average, life expectancy is 29 percent lower, primary school completion is 66 percent lower, and the ratio of female to male tertiary enrollment is about half the global mean. But the difference between the average level of social indicators in good and bad times is smaller for Sub-Saharan Africa than it is for developing countries as a whole (compare figure 2.1 and 2.2). This finding may imply that at low levels of income, the ability to improve social indicators is particularly limited—and therefore so is the likely deterioration.

FIGURE 2.2.Key human development and gender indicators also fall below their overall means during growth decelerations in Sub-Saharan countries, if less so

Source: World Bank staff calculations based on data from World Development Indicators. See annex table A2.1.2 for levels and Arbache, Go, and Korman (2010) for more details.

Note: Differences of sample averages during growth accelerations and decelerations from the overall sample means.

BOX 2.2Aggregate economic shocks and gender differences: A review of the evidence

The labor market for nonagricultural wage work by women (often used as a proxy for women’s access to decent work) tends to behave very differently from the market for nonagricultural wage work by men. While the unemployment rate rarely differs between men and women, a much smaller proportion of working-age women are in the labor force (whereas most men of working age are either working or unemployed, women may be working at home). Thus in analyzing the impact of crises (see the figure below), it is important to take into account the response of women who are not in the labor force. In some crises, for example, in Indonesia in 1997, women entered the labor force to maintain household consumption (called the added-worker effect). In the Republic of Korea during the 1997 crisis, some women left the labor force (the discouraged-worker effect).

Possible transmission channels of economic crisis

The table below summarizes country studies of the impact of crises on women’s labor market participation and health and education outcomes.

Previous crises: Available evidence by country
Income level at time of crisisCountryLabor market effects for womenSchoolingHealth
Low-income countryCôte d’lvoireDecline in student enrollment for both boys and girls.Deteriorating child health for both boys and girls.
Pooled survey data from several low-income countriesGirls’ infant mortality more sensitive than boys’ infant mortality to fluctuations in GDP.
Lower-middle-income countryIndonesiaAdded-worker effect in 1997–98.Decline in student enrollment for young boys and girls and older girls.Higher neonatal mortality, but overall not much effect on child health.

Lower body mass index for adults.
PeruAdded-worker effect in Lima.Increase in student enrollment for both boys and girls.Higher infant mortality rate.
PhilippinesAdded-worker effect.Drop in high school enrollment for both boys and girls.

Drop in elementary school enrollment, more for girls than boys.

Increase in child labor, more for boys than girls.
Upper-middle-income countryArgentinaAdded-worker effect in urban areas during 1990s.
BrazilBoth added- and discouraged-worker effect in Sao Paulo during 1980s.
Costa RicaDecline in student enrollment for both boys and girls in rural areas; higher for girls than boys in urban areas.
MexicoAdded-worker effect during 1980s.Increase in student enrollment in 1995-96, stronger for boys than girls.Higher child mortality during crises.
Russian FederationDeterioration in weight for height for both boys and girls.
High-income countryKorea, Rep.Discouraged-worker effect in Seoul in 1997-98.
United StatesNo effect.Increase in student enrollment during Great Depression.Improved child health outcomes.

Explaining the pattern of past crises

Several factors contributed to negative human development outcomes during past economic downturns, including the high frequency of downturns in low-income countries; the poor policy environment in many countries during past crises, particularly in low-income countries; shrinking social spending during contractions and the lack of social safety nets; and declines in aid during crises that also affect high-income countries.

Poor countries suffer from frequent economic contractions and high growth volatility

One reason for the low levels of human development in low-income countries is that they experience numerous crises. Of all country-year observations for low-income, International Development Association (IDA)-eligible, and Sub-Saharan African countries, nearly a quarter are decelerations; for the middle-income countries in East Asia, Europe and Central Asia, and Latin America, less than 10 percent are decelerations (table 2.2). Similarly, the share of accelerations is 37-39 percent for poorer countries but 43-53 percent for middle-income countries.2 In addition, overall growth volatility is greater in low-income countries and in Sub-Saharan Africa than in middle-income countries (see table 2.2). The regional pattern suggests growth spillovers at the geographic level, which may be associated with economic geography, regional trade arrangements, natural disasters, regional migration, and regional conflicts.

TABLE 2.2Frequency of growth acceleration and deceleration, growth rates, and GDP per capita, 1980–2008
GDPGrowth accelerationGrowth deceleration
Region, incomeGDP per capita growth rate %Standard deviation of growthFrequency (country years)GDP per capita growth rate %Frequency (country years)GDP per capita growth rate %
East Asia and Pacific3.094.450.465.010.09-2.75
Europe and Central Asia2.206.650.534.790.08-7.19
Latin America and the Caribbean1.634.650.533.720.07-2.78
Middle East and North Africa1.415.510.432.890.06-3.44
South Asia3.722.870.364.69
Sub-Saharan Africa1.027.280.394.190.22-3.17
Country Income category
Developing countries1.676.370.464.330.14-3.87
IDA countries0.996.280.393.820.21-3.47
Low-income countries0.636.740.373.750.23-3.50
Lower-middle-income countries1.985.890.474.520.13-4.99
Upper-middle-income countries2.346.430.554.540.08-2.76
High-income, non-OECD countries3.027.410.425.900.02-4.62
High-income OECD countries2.192.590.543.310.03-2.32
Source: World Bank staff calculations based on data from World Development Indicators.Note: — = not available. OECD = Organisation for Economic Co-operation and Development.

Differences in the frequency of economic contraction explain a significant share of the differences in the average growth rate of different groups of countries. Growth in GDP per capita during 1980-2008 was 0.6 percent a year in low-income countries and more than 2 percent a year in middle-income countries. The slower growth in low-income countries stems from the greater frequency of decelerations,3 not from a marked difference in growth rates during booms and busts. For example, during periods of acceleration, low-income countries’ per capita GDP rose 3.75 percent, slightly less than the 4.5 percent growth rate for middle-income countries. During decelerations, low-income countries’ per capita GDP fell 3.5 percent, somewhat less than in middle-income countries. Thus “defensive” policies that prevent collapses should have substantial impacts on average growth by avoiding multiple collapses and their negative outcomes.4 The finding that the elasticity of poverty to growth is lower in high-poverty countries (see chapter 1) suggests that low-income countries, where poverty rates are high, need a long period of sustained growth to reduce poverty and improve other human development indicators.

Contractions tend to occur in severely unfavorable economic and policy environments

Contractions have a grave impact on human development because they are marked by an overall deterioration in government effectiveness. Similar to human development indicators, indicators of institutional quality (political stability, voice and accountability, regulatory framework, rule of law, and government effectiveness) in developing countries perform asymmetrically over the growth cycle. In other words, the deterioration during bad times is much greater than the improvement during good times relative to the sample averages for all times (figure 2.3). Causality likely moves in both directions: a deterioration in institutions impairs growth, which leads to further institutional weaknesses, and so on. In many cases both institutions and growth are affected by domestic violence or foreign wars. In Sub-Saharan Africa, for example, the frequency of major and minor conflicts is about 23 percent during growth deceleration and 13 percent during growth acceleration. In oil- and mineral-dependent economies, defects in institutional quality may be masked by the revenues generated by favorable commodity prices. These defects become clear when prices turn down and revenues dry up.5

FIGURE 2.3During growth decelerations, economic and institutional indicators diverge far from the overall means

Source: World Bank staff calculations based on data from World Development Indicators. See annex table A2.3 for levels and Arbache, Go, and Korman (2010) for more details.

Note: Differences of sample averages during growth accelerations and decelerations from the overall sample means.

Macroeconomic variables such as investment, savings, exports, imports, external finance, and inflation deteriorate more during downturns than they improve during upturns (see figure 2.3). During decelerations, both savings and investment, particularly fixed private investment, decline relative to average levels (as a share of GDP) by much more than they rise during accelerations. The increase in foreign direct investment as a share of GDP during accelerations is twice as large as the drop during decelerations (relative to the sample mean). The very high values for inflation during growth decelerations reflect the incidence of hyperinflation during several growth collapses in Africa before the 1990s and incidents of high inflation in the early 1990s (such as in Angola, Armenia, Azerbaijan, Belarus, Brazil, Democratic Republic of Congo, Peru, and Ukraine).6

Aid to poor countries is procyclical, suggesting that donors respond to an emerging policy failure by giving less aid. Macroeconomic and institutional variables are more closely correlated with the incidence of deceleration than of acceleration (table 2.3)—further evidence of the asymmetric behavior of these indicators over the economic cycle and of the extremely poor economic environment characterizing downturns in developing countries.

TABLE 2.3Correlation coefficients between economic cycles and economic and institutional indicators
Growth accelerationGrowth deceleration
IndicatorCoefficientSignificance levelCoefficientSignificance level
Final consumption (% GDP)-0.1**0.13**
Government consumption (% GDP)-0.07**0.04**
Gross capital formation (% GDP)0.09**-0.17**
Gross domestic savings (% GDP)0.1**-0.13**
Gross private fixed capital formation (% GDP)0.19**-0.19**
Imports (% GDP)0.06**-0.09**
Exports (% GDP)0.09**-0.11**
Trade (% GDP)0.08**-0.11**
Net foreign direct investment (% GDP)0.03-0.04*
Private capital flows, total (% GDP)0.04*-0.04*
Inflation (%)-0.06**0.13**
Political stability-0.07**-0.12**
Voice and accountability-0.07**-0.1**
Regulatory framework-0.06*-0.19**
Rule of law-0.1**-0.18**
Government effectiveness-0.05*-0.21**
Source: World Bank staff calculations. Indicators on institutions are from the World Bank Institute’s Worldwide Governance Indicators database, which relies on 33 sources, including surveys of enterprises and citizens, and expert polls, gathered from 30 organizations around the world. * Significant at the 5 percent level; ** significant at the 1 percent level.

Social spending under pressure

Drops in spending on social services like education and health care are an important reason for the sharp deterioration in human development indicators during crises. Cutbacks in social spending are more worrisome during crises because that is when people need these services most. Cutbacks during crises are also harmful because such disruptions have long-lasting effects. This section draws on evidence of the impacts of GDP downturns on public and private domestic spending on health and public education for over 108 developing countries for 1995–2007.7

The analysis points to three key results. First, social spending growth rates tended to be volatile. Second, per capita social spending levels nonetheless showed a steady upward trend. Third, social spending in poor countries was subject to more pressures than in richer countries during contractions in GDP. The last result confirms that it is the low-income countries that are more likely to need help in protecting social expenditures during crises.

The volatility of public and private health spending is evident from the unweighted average of growth rates of social spending by country over time, a calculation that gives equal weight to changes for each country and does not allow the larger countries to dominate the pattern. In particular, changes in public health spending are more volatile than GDP growth trends over time (figure 2.4). Historically, a drop in GDP growth of 2 percent or more has had a greater than proportional effect on the growth of public health spending. Growth of private health spending (insurance and out-of-pocket payments) responds in a similar way, although the pattern is more volatile than for government spending, especially before 2005.

FIGURE 2.4Health spending growth rate is more volatile than its per capita level or GDP growth

Source: Lewis and Verhoeven 2010.

Despite the fluctuations in growth rates of GDP and health spending, the trends in absolute per capita health spending continue to rise over time. In general, private health spending rises more slowly than public health spending. The steady rise of the mean per capita spending level (which is also an unweighted average across countries) suggests that the volatility of spending growth rates was affecting countries with lower levels of per capita spending more than countries with higher spending levels. Indeed, the data confirm that the negative impacts of crises on health spending are much stronger in the lowest-income countries, where growth in health spending is more likely to fall in response to a decline in GDP.

Education spending in developing countries appears to be less closely tied to GDP growth than is health spending (figure 2.5). In absolute terms, the public sector spends more, on a per capita basis, on education than on health. There is also a modest rising trend. The data on education spending are weak, however, so any conclusions concerning cross-country trends are subject to considerable uncertainty.8

FIGURE 2.5Public education spending is less closely tied to GDP growth than is health spending

Source: Lewis and Verhoeven 2010.

Donor funding under pressure

Does aid to developing countries rise during crises? The sharp deterioration in human development indicators and the decline in social spending during growth decelerations highlight a potentially important role for donors. Aid’s contribution to welfare in developing countries could be bolstered by increasing aid in bad times to compensate for shortfalls in government resources. The evidence on whether aid plays such a countercyclical role is mixed. Overall, aid to individual countries appears to be procyclical, increasing when growth rises and falling when growth slows. And growth in aid to health and education sectors does not appear to be closely related to GDP growth in developing countries (figure 2.6). However, after 2003, aid to education shows a small response to growth, with donor financing rising and falling as national education resources fall and rise. This countercyclical financing suggests that donor spending is modestly compensating for GDP growth shifts.

FIGURE 2.6Aid to education and health does not appear to be closely related to GDP growth, 1998-2007

Source: Lewis and Verhoeven 2010.

Evidence on aid funding during financial crises in donor countries is limited. A recent study that tracked donor allocations during and after banking crises (1998–2007) in developed countries suggests that donor funding is tied to economic prosperity in those countries.9 Aid flows decline 20–25 percent during banking crises in member countries of the Organisation for Economic Co-operation and Development (OECD) and take significant time to recover. Not all donor programs are equally affected. But some combination of the fiscal costs of crises, debt overhang after the crisis ends, and perhaps erosion in public support reduces aid flows from affected donor countries. To the extent that aid recipients and donors are simultaneously affected by crises (as in the recent crisis), cutbacks in donor funding could deepen the economic deterioration in developing countries.

Two deviations from these aggregate trends are instructive—and encouraging. During the Southeast Asian crisis of 1997, donors (most notably the U.K. Department for International Development) supported core social programs in Indonesia, slowing the declines in education and health spending and permitting social services to continue.10 During the current crisis, Mexico sought loans from the World Bank to compensate for budget reductions and to expand temporary safety nets. Latvia, Lithuania, Poland, and Romania all received policy loans or technical assistance from the World Bank to support reforms and continued financing of safety nets and education and health programs. These aid responses provided necessary finance for income support and social service programs—both critical for bridging financial gaps during a downturn.

Safety nets were uncommon in developing countries before previous crises

Few countries facing previous macroeconomic and financial crises in Asia (1997–99), Europe (Russian Federation 1998, Turkey 2001), and Latin America (1980s, 1994–95, 1999, 2001–02) had strong safety nets in place before the crisis. Countries had to scale up programs, regardless of the fit between the original target population and the population affected by the crisis, or quickly start new ones. Mexico scaled up retraining and employment programs and targeted food distribution in response to the “tequila” crisis, even though these programs probably did not target the most affected populations. Safety net programs set up after a crisis starts often suffer from poor initial implementation, as with Indonesia’s Labor-Intensive Public Works (JPS Padat Karya) program, or take too long to scale up, as with Colombia’s Families in Action (Familias en Acción) program.11

MAP 2.1Around 9 million young children die before their fifth birthday

Source: World Development Indicators.

Despite the difficulties, however, countries have managed to start effective safety net programs in response to crises. Argentina established a new workfare (Jefes de Hogar) program in response to the 2002 crisis and extended it to 2 million participants within a year. Incidence and coverage were good, with about 80 percent of the benefits concentrated among the poorest 40 percent of the population. Argentina benefited from extensive experience with an earlier, smaller workfare program. The Republic of Korea was able to quickly introduce a public works program in response to the Asian financial crisis, reaching more than 400,000 people within six months.12

Beyond emergency responses to natural disasters and humanitarian crises, safety nets have been uncommon in low-income countries, partly because they were viewed as taking away from more productive expenditures. But support for social safety nets in the poorest countries has risen as their importance in protecting the poor and vulnerable during crises has become evident. Low-income and fragile countries are devoting a larger share of lending to public works programs and increasing cash transfers and in-kind safety nets, with a renewed focus on school feeding programs.

A key lesson in previous crises is the importance of well-functioning safety nets in responding to a crisis and promoting growth and development afterward. When already in place, safety nets can be scaled up to meet increased needs and then scaled back as the crisis subsides. They can provide temporary protection for households by cushioning unemployment, contractions in public services, and falling demand for formal and informal work. But if safety nets are not in place when shocks strike, governments might respond with price subsidies or other suboptimal policies, which can leave an unwanted legacy of fiscal burden, economic distortions, lower growth, and greater poverty.

What is happening in the current crisis—and what is different?

There is some hope that human development indicators have not deteriorated as much during the current crisis as in previous crises. Because the current economic crisis did not reach most developing countries until 2009, it is too early to arrive at a definitive conclusion on its impact. However, rapid surveys and discussions with governments have yielded preliminary evidence showing that social spending may have held up better during this crisis than in previous ones and that there has been more reliance on social safety nets. Moreover, policy regimes in developing countries had improved considerably before the crisis, so governments might have had greater success in protecting their populations from the worst effects of the growth downturn.

Social spending held up in some regions

Impacts on social sector budgets for 2008–10 varied by country circumstances, specifically according to how the global downturn affected the economy and public revenues and whether countries prepared for a possible contraction.

Latin America and the Caribbean. In Latin America social spending has remained strong, partly because of the relatively modest size and scope of the downturn in much of the region and partly because of efforts to protect social spending. Some of the larger economies (such as Brazil and Chile) instituted social measures aimed at financing temporary employment and transfers to vulnerable populations. Chile’s Social and Economic Stabilization Fund provided a countercyclical boost in spending, blunting the effect of the external shock.13 In Mexico, the severe contraction imposed intense fiscal pressure, but education and health funding are nevertheless set to rise 10 percent in 2009-10 (although spending on education is expected to fall sharply as a share of total government expenditures; figure 2.7). El Salvador is not cutting education funding despite the severe recession, but health spending is expected to fall from 3.4 percent to 3.0 percent of GDP, largely because of reductions in the Social Security Institute’s health expenditures.14

FIGURE 2.7Despite intense fiscal pressures, Mexico’s federal funding for health and education is set to rise in 2009–10

Source: Mexican government statistics.

Note: Data are estimated for 2005-07, approved for 2008 and 2009, and projected for 2010.

Eastern Europe and Central Asia. Because Eastern Europe was the hardest hit of emerging market regions, countries there were the first to cut all areas of public spending. Neither education nor health has necessarily been spared. However, aggregate trends in social spending alone are not an accurate indication of the impact of the crisis on outcomes. Some countries have directed spending reductions to sectors with overcapacity, thus improving long-term efficiency and limiting the welfare impact of expenditure cuts (Box 2.3).

BOX 2.3Crises as opportunities for reform

Crises can present opportunities to achieve reforms in social sector spending that will improve efficiency and welfare over the long term. The current crisis sharply reduced GDP in many countries in Eastern Europe—by more than 15 percent in Latvia—making it impossible to sustain social spending at precrisis rates. Across-the-board cuts in education spending would have greatly impaired access to education, with dismal implications for the quality of the workforce and long-term productivity. Instead, Latvia and Romania directed spending cuts at areas of overcapacity, through reforms that previously had been blocked by political opposition.

Latvia is using the stringencies imposed by the crisis to right-size its teaching force. By shifting teacher financing and management to local governments and providing them with per capita student transfers, the central government is tackling overcapacity. This reform translates into an average 34 percent reduction in the number of teachers and a 45 percent reduction in teacher salaries. In the health sector, the government has embraced sources of efficiency gain through restructuring. Drawing on diagnostic work with the World Bank, the government has eliminated excess hospital beds, invigorated outpatient care, and prioritized the financing of effective health care procedures by adjusting the list of ineligible health services. The crisis made all these needed reforms possible, and policy research informed strategic investment decisions, largely avoiding across-the-board reductions or random cuts in social programs.

Romania responded to declining school enrollments and tighter budgets by substantially reducing education personnel (teaching and nonteaching positions) in 2009 and by curtailing supplements to base salaries. Some 18,000 teachers (6 percent) were laid off following adjustments in teaching norms and substantial cuts in the funds allocated to each county. The Ministry of Education, Research, and Innovation cut 15,000 additional public positions, consolidated schools, and reduced the number of scholarships for higher education. The staffing reductions will allow much needed adjustments to class size and better alignment of teachers, students, and budgets. The ministry also has reduced the number of fee-paying students, which shrinks the overall resource envelope for higher education at the same time that budgets are being cut.

Source: Lewis and Verhoeven 2010.

By contrast, some countries had increased planned spending heading into the crisis, necessitating painful reductions as government resources dwindled. In Moldova, education sector employees make up about 60 percent of public employees. During two election campaigns in 2009, the outgoing government raised teachers’ salaries 25-30 percent to align entry-level salaries with average national earnings but made no commensurate increases in class size or shifts in teaching loads. Other measures were contemplated that would further raise real wages if implemented. The new government’s challenge will be to implement corresponding increases in class size or shifts in teaching loads. Ukraine adopted a social standards law in November 2009 that calls for large increases in minimum wages and social standards throughout 2009-10. Public wages were adjusted accordingly at the end of 2009, but the subsequent wage hikes have not taken place because of challenges to the law in the Constitutional Court. More recently, the Cabinet of Ministers approved salary top-ups for secondary, vocational, and university teachers (equivalent to 20 percent of base salary). With no budget yet in place for 2010, budget operations are being executed on the basis of an operational budget that limits current monthly spending to one-twelfth of the 2009 appropriations.

One indicator of the impact of the crisis on health expenditures is that pharmaceutical spending (a good proxy for health sector spending) has declined sharply in Eastern Europe. World demand continued to rise from the first quarter of 2007 through the last quarter of 2009 (with the first quarter of 2008 considered the last quarter before the worldwide financial crisis), but expenditures in Eastern Europe declined in the first quarter of 2009 before beginning to rise again. The decline was most dramatic in the Baltics, with Latvia cutting back pharmaceutical expenditures by more than 25 percent between the fourth quarter of 2008 and the end of 2009 (figure 2.8).

FIGURE 2.8Average pharmaceutical expenditures fall in Eastern Europe, especially in the Baltics, before beginning to rise again

Information on social sector spending in other regions is extremely limited, but scattered information provides some examples. For example, 16 of 19 country programs initiated and monitored by the International Monetary Fund and implemented with the World Bank in 2008-09 budgeted higher social spending for 2009; 9 of those countries were in Sub-Saharan Africa (Burundi, Republic of Congo, Côte d’Ivoire, Liberia, Malawi, Mali, Niger, Togo, and Zambia.15 Several African countries with poverty reduction strategies have protected funding for social sectors. Some countries with adequate fiscal space (Kenya and Nigeria) have protected capital expenditure, mainly for infrastructure. But there are also examples of forced contractions in social spending. Countries with precrisis fiscal and debt problems (such as Ethiopia and Ghana) had to undertake fiscal tightening.16 The effects of the crisis have been relatively modest in East Asia, although qualitative evidence in six countries suggests that informal work has surged and that migrants have returned home temporarily, lowering overall income and reducing households’ ability to pay for social services. Households have responded by transferring children from private to public schools and reducing food consumption, although parents contend that they have tried to shield children’s nutrition.17

Spending to combat HIV/AIDS (human immunodeficiency virus/acquired immune deficiency syndrome) is a special case. Big increases in funding have made HIV/AIDS one of the most important items on the development agenda. Funding for HIV/AIDs programs during the current crisis has been largely sustained. The uptick in donor spending in 2008-09, when the economic crisis was accelerating in donor countries, is encouraging. The Global Fund to Fight AIDS, Tuberculosis, and Malaria (the Global Fund) disburses quickly once allocations are decided, but recipient country spending has been slow. So the issue is sluggish disbursement and a new concern for efficiency of resource use (see annex 2.2 for a detailed discussion). Almost 40 percent of the Global Fund resources remain undisbursed, a possible source of additional resources if there is a shortfall or delay in funding flows. Almost half the allocations to Sub-Saharan Africa are undisbursed (figure 2.9). The $900 million allocated in late 2009 under Round 9 is also unlikely to have been disbursed yet.18

FIGURE 2.9Undisbursed HIV/AIDS grants from the Global Fund to Fight AIDS, Tuberculosis and Malaria, Rounds 1–7

Source: Lewis 2009.

The recent buildup of social safety nets

Safety nets have been a crucial part of the response to the crises in the hardest-hit countries. Many countries that responded most effectively already had safety nets, which governments were able to quickly modify and expand. Evidence on the distribution of safety net programs shows that programs vary considerably across regions. For example, food-based programs are more common in Africa than in other regions (figure 2.10).19

FIGURE 2.10Food-related safety net programs are more common in Africa than elsewhere

Source: Wodon and Zaman 2010.

Note: Based on a March 2008 survey of 120 World Bank country teams.

Another sign of the importance of safety nets in responding to the crisis is the dramatic increase in World Bank lending for safety nets after the crisis struck—topping $3 billion in 29 countries in fiscal 2009. Elevated activity is expected to continue in 2010-11, particularly in low-income countries and fragile and postconflict settings (table 2.4).20 The regional distribution of lending activities reflects the dominance of Latin America, which had the greatest number of effective safety nets in place before the crisis that could be scaled up (table 2.5). Less funding to Africa and South Asia reflects the fact that existing safety nets were smaller and less able to absorb funds. Thus, where capacity was in place, lending could be quickly leveraged. Some countries have been reluctant to introduce safety net programs because of the costs involved, although reducing across-the-board subsidies while augmenting targeted safety nets can help reduce poverty without a significant drain on revenues (box 2.4).

TABLE 2.4World Bank lending for safety nets before and since the food, fuel, and financial crises, 2006–11US$ billions
International Bank for Reconstruction and DevelopmentInternational Development Association
2006–08 (precrisis)0.570.620.031.23
2009–11 (postcrisis)4.481.380.035.89
Source: Data from the World Bank Business Warehouse project database for projects classified as social safety nets (54).Note: Data do not include interventions funded under the Global Force Crisis Response Program.
TABLE 2.5World Bank portfolio allocations to social safety nets, by region, 2009–10
RegionAmount (US$ millions)Number of projects
Latin America and the Caribbean2,91721
Europe and Central Asia92621
East Asia and Pacific6189
Sub-Saharan Africa57423
South Asia3739
Middle East and North Africa198
Source: Data from the World Bank Business Warehouse project database for projects classified as social safety nets.Note: Data do not include interventions funded under the Global Force Crisis Response Program.

Several countries expanded existing or planned safety net programs in response to the crisis.

  • The Republic of Yemen, hard hit by the global food crisis (drought has forced imports of more than three-quarters of its food), expanded safety net programs with support from the World Bank and the European Union. The cash-for-work program was extended to an additional 22,000-26,000 households in communities most affected by higher food prices, the share of cash transfers to the poorest beneficiaries was increased, and 40,000-50,000 more households were added to the cash transfer program.

  • The food and fuel price shocks in 2008, the global economic crisis, and a recent typhoon have sharply increased poverty in the Philippines. The government had begun planning for a pilot conditional cash transfer program (Pantawid Pamilyang Pilipino Program, or 4Ps) in 2007. It was launched in February 2008 for 6,000 households. As the crisis unfolded, the government accelerated and augmented the program, rolling it out to 376,000 households by March 2009. In mid-2009, the government announced plans to expand the program to as many as 1 million households by the end of 2009.

  • Before the crisis, the government of Brazil had established a highly successful conditional cash transfer program, Bolsa Familia, to protect poor families. When the crisis hit, the government expanded the program to more than 12 million families, using a new methodology of poverty maps and an income volatility index, and raised the benefit level 10 percent to compensate for higher food prices. The program was expanded in regions where poverty reduction has been slow—in urban municipalities and in the mid-south region—reaching 1.3 million families in those areas in 2009. Another 600,000 families within poverty belts or in specific vulnerable groups are expected to join the program in 2010.

  • In response to the food, fuel, and financial crises, Chile announced in April 2009 the strengthening of multiple safety net programs. Family allowances of about $45 were distributed to 1.4 million families, including all families in the Chile Solidario program (around 300,000), families in the Family Subsidy program, and families whose monthly income was $555 or less. In all, some 5.6 million people in the bottom 40 percent of the income distribution will benefit, at a cost of $62 million.

  • The government of Ethiopia established the Productive Safety Net Program in 2005 to pay for participation in labor-intensive public works and provide direct support to elderly or incapacitated household members. The program has been expanded since, providing immediate assistance to 1.5 million households when the food and fuel crises struck, and providing additional transfers to 4.4 million people as the crisis deepened. Evaluations find a positive impact on use of health services and caloric availability and reductions in negative coping behaviors, such as child labor and withdrawal from school.

BOX 2.4Using safety nets to lower the cost of reducing poverty

Some countries have hesitated to establish safety nets because of the cost. Safety net expenditures in developing countries average 1–2 percent of GDP. Expenditures on programs that are to scale and that have been evaluated as delivering significant positive impacts, such as Mexico’s Oportunidades and Brazil’s Bolsa Familia, average 0.4 percent of GDP. Ethiopia’s largest safety net program, the Productive Safety Net, costs about 1.7 percent of GDP.

The introduction of a well-targeted safety net can provide the political space to reduce or eliminate expensive and poorly targeted general price subsidies, freeing up resources to fund the targeted programs. The potential for such reallocations is considerable because many countries have large and costly price subsidies. More than a third of countries recently surveyed by the International Monetary Fund raised subsidies an average of 1 percent of GDP in response to higher food and fuel prices. Several examples illustrate successful country experiences in switching from universal subsidies to targeted safety nets.

In the late 1990s Mexico progressively moved funding from price and in-kind food subsidy programs to the Oportunidades conditional cash transfer program, probably the most positively evaluated safety net program in a developing country. Fifteen years later, as food prices rose dramatically, the government was able to protect the poor by issuing a one-time top-up benefit to those already in the program. The response was easy, fast, and affordable because of the earlier investment.

In 2005 Indonesia cut its fuel subsidies by $10 billion, using a quarter of the released funds for a targeted cash transfer that more than compensated poor recipients for their losses. Another quarter of the savings went to basic health and education programs for the poor.

In 2008 the Philippines found itself short of effective policy instruments to protect the poor against escalating rice prices. A key part of its multipronged response package, which cost some 1.3 percent of GDP, was a program of loosely targeted and distortive rice subsidies. Realizing that this approach is expensive and regressive, the government is working on better safety net options—unifying administration under a new umbrella program, scaling up a proxy means test for targeting households, reforming and expanding the school feeding program, and accelerating rollout and scaling up of a conditional cash transfer program.

Source: Data for 87 countries for which data on safety net expenditure were available from World Bank public expenditure reviews, safety net assessments, social protection strategy notes, and other studies. Data coverage is low for Sub-Saharan Africa, where government spending on safety nets may be low, but where donor funding may compensate considerably. See also IMF 2005.

Safety nets are important not only in cushioning the effects of the crisis but also as part of a broader poverty reduction strategy interacting with social insurance; health, education, and financial services; the provision of utilities and roads; and other policies for reducing poverty and managing risk. Many challenges remain, however. Safety net programs in low-income countries are often slight and fragmented and cover only a small percentage of poor and vulnerable populations. There are real concerns over whether they are affordable and administratively feasible or desirable, considering the negative incentives they might create. Thus a part of policy reforms in developing countries should be understanding what kind of safety net program best serves various social assistance activities, what the implementation challenges are, and how to develop programs for maximum effectiveness.21

Informal safety nets and remittances

Households manage risk through informal safety nets (such as crop diversification), informal savings and credit associations, burial societies, labor exchange arrangements, migration, and emigration. Informal safety nets are generally more effective against idiosyncratic shocks that affect only one or a few households than against systemic shocks that affect whole communities. Thus it is not clear whether such risk-mitigation strategies were any more effective during the recent crisis than they had been previously.

MAP 2.2An infant in a developing country is ten times more likely to die than a newborn in a developed country

Source: World Development Indicators.

Remittances have played an important countercyclical role in crises that affected individual developing countries.22 Because remittances are unaffected by idiosyncratic shocks or even local or national systemic shocks, they are an important part of the household safety net for many poor households. More than tripling since earlier in the past decade, remittances constitute important monetary flows to developing countries; they reached $338 billion in 2008 before the full impact of the financial crisis was felt.23

The global nature of the current crisis has likely reduced the support that remittances can provide. Remittance flows to developing countries are estimated at $317 billion for 2009, a 6.1 percent decline from 2008. Analysis of the first nine months of 2009 shows that the financial crisis has affected remittance flows unevenly. Remittances to Latin America and the Caribbean have suffered large declines (down 13 percent in Mexico, for example), mainly because of the early effects of the crisis in the United States and Spain. Similarly, remittances to the Middle East and North Africa have declined more than expected, plunging 20 percent in the Arab Republic of Egypt and Morocco. The situation is even more serious in Europe and Central Asia, where many countries are among the top recipients of migrant remittances as a percentage of GDP. Tajikistan, where remittances make up 50 percent of GDP, experienced a decline of more than 30 percent in the first half of 2009. Many other countries in the region have experienced similar declines. By contrast, in Sub-Saharan Africa the decline has been less steep and in some countries, such as Uganda, flows have increased. In South Asia, remittances have remained strong, even increasing in some cases (up 24 percent in Pakistan, 16 percent in Bangladesh, and 13 percent in Nepal). In East Asia and the Pacific, flows were also stronger than expected (up 4 percent in the Philippines).

A study by the World Food Programme found that families that rely on remittances from abroad were among groups most affected by the current financial crisis.24 In Armenia, where remittances make up 20 percent of GDP and are the main source of income for 25 percent of households, the impact was felt immediately, with remittances slumping 30 percent in the first quarter of 2009. In Nicaragua, a country highly dependent on remittances and vulnerable to economic downturns in the U.S. economy, food consumption patterns are changing and families are spending less on health and education.

This crisis is not about domestic policy failure

Improvements in developing countries’ policies since the 1990s may blunt the impact of the crisis on human development. Crises in low-income countries have often been driven by poor governance, civil conflicts, or severely distorted macroeconomic policies. (Internal shocks accounted for 89 percent of output volatility in low-income countries from the early 1960s to mid-1990s.) The failure of domestic institutions has been an important reason for the severity of past crises on human development, macroeconomic variables, and the quality of institutions.

But some indirect evidence suggests that this situation may be changing and that the impacts of the current crisis on human development could be less severe than in previous crises. Since the 1990s output volatility in low-income countries has lessened, and the influence of external shocks has intensified (box 2.5). To the extent that lower volatility and a reduced importance for internal shocks indicate improved policies, governments should be better placed to protect their people from the most severe impacts of the crisis.

Assessing the quality of policies and institutions over time is difficult, but external evidence does indicate an improvement in many developing countries since the 1990s. Inflation rates have declined substantially, fewer countries have unsustainable debt positions, more countries have access to private capital markets and have attracted substantial foreign direct investment, financial intermediation has risen as a share of output, trade barriers have come down, black market exchange rate premiums have shrunk, and civil conflict has subsided in many countries. The pace of policy reform has varied. In Latin America weak currencies, banking sectors, and poor fiscal management tended to amplify the impact of past crises, whereas improvements in the policy and institutional framework have cushioned the impact of the current crisis—a first for the region. By contrast, in Europe and Central Asia middle-income countries that were unable to halt large increases in private sector credit growth were the hardest hit by the current crisis. They had higher growth rates before the crisis but also larger declines after the crisis, and on balance they experienced lower average growth rates than countries with more modest increases in private sector credit growth.

Comparing the economic performance of countries according to the quality of their policies and institutions shows the importance of policy reform. Although there is no perfect measure of the quality of the policy and institutional environment in developing countries, the World Bank’s Country Policy and Institutional Assessment (CPIA) provides a consistent framework for assessing country performance (on a scale from 1, worst, to 6, best).25 Countries with better policies or initial fiscal positions have generally done better in the current crisis (see chapter 3). And before the crisis (2001-07), developing countries with 2008 CPIA scores of 3.2 or better grew faster than countries below this cutoff (figure 2.11). Per capita GDP growth averaged 3.9 percent for countries with good policies and 1.9 percent for fragile states with poorer policies. Countries with better policies also had lower inflation, at 5.2 percent a year, compared with 6.6 percent for countries with poorer policies. The pattern is the same for countries in Sub-Saharan Africa.26 Before the current crisis, countries with better policies tended to have better outcomes for MDG indicators such as under-five mortality, gender equality in primary and secondary education, primary school completion, and access to an improved water source. Several empirical studies also showed that better policies and institutions improve the marginal contribution of growth to progress on human development indicators.27

FIGURE 2.11Economic performance and MDG outcomes are better with good policy

Source: World Bank staff calculations.

Note: CPIA is the World Bank’s Country Policy and Institutional Assessment framework for assessing country performance; ratings range from 1 (worst) to 6 (best). Countries with a CPIA score of 3.2 or better have better policies than countries that score under 3.2.

The impact of the current crisis is still worrisome

The crisis has generated predictions of rising mortality rates and closed schools as governments reduce services in response to falling output and public revenues. These fears are grounded in the experience of past crises, when, as noted earlier and in box 2.6, poverty, hunger, health outcomes, and access to education deteriorated sharply. Despite policy improvements and efforts to sustain social spending and ramp up safety nets, preliminary indications of the impact of the crisis on human development point to serious problems. An important reason is the size of the shock—it is the largest global downturn since the Great Depression. Thus while developing countries’ efforts have been important in mitigating the impact of the crisis, the crisis nevertheless has been a severe setback to poverty reduction.

BOX 2.5Are external shocks becoming more important than internal shocks for developing countries?

Historically, developing countries have endured greater macroeconomic volatility than have industrial economies. A simple look at the data shows that output volatility (measured as the standard deviation of real GDP growth) has been two to three times greater in developing countries than in industrial countries in the last 20 years.

Because developing countries are highly dependent on primary commodities and foreign capital and have greater exposure to natural disasters, policy makers often blame external shocks, such as terms-of-trade fluctuations, natural disasters, and aid volatility, for countries’ uneven macroeconomic performance.

However, research shows that external shocks account for only a small fraction of the variance in real per capita GDP in low- and middle-income countries. Among low-income countries, external shocks, including terms-of-trade fluctuations, global economic growth, international financial conditions, natural disasters, and aid volatility, explain no more than 11 percent of output volatility. The 89 percent residual is probably related to internal conditions, such as the volatility of macroeconomic management. Among middle-income countries, external shocks account for about 20 percent of output volatility.

Since the 1990s, however, many developing countries have undergone structural transformations that may have calmed internal volatility and increased the importance of external factors. Research shows that external shocks have become more important for developing countries in several regions during the past two decades (see the figure at the right). In African countries this shift has resulted not from an increase in the volatility of external shocks, or in countries’ vulnerability to them, but rather from the taming of internal sources of volatility. In these countries—among the most volatile—standard indicators of democratic accountability, economic management, and control of corruption have improved since the early 1990s. Many middle-income countries have also strengthened their fiscal position by reducing deficits and accumulating reserves; tamed inflation through independent central banks; and promoted local bond markets after the Asian and Russian financial crises of 1997–98.

External shocks have become more important since the 1990s

Source: Raddatz 2007, 2008a, 2008b.Note: The data for Latin America and the Caribbean, East Asia and Pacific, and other middle-income countries correspond to 1974–1985 (earlier) and 1986–2004 (later). The bars for Africa correspond to 1963–89 (earlier) and 1990–2003 (later). External shocks explaining long-run output volatility include the state of the world business cycle, international financial conditions, terms of trade, natural disasters, and aid flows.

The recent deterioration in human development indicators began with the food and fuel price shocks of 2007. In some countries food prices almost doubled with no adjustment in earnings.28 In Mozambique incomes were almost halved and food consumption fell by a fifth; children’s weight for age and body mass index were reduced with no change in height for age, indicating that the price rise has seriously compromised nutrition. The effects spilled over into the efficacy of HIV/AIDS treatment, with lower-income households showing slower improvements than households with higher incomes and better access to adequate nutrition, which reinforces the beneficial effects of antiretroviral therapies. Recent analysis finds that the 2008 global food price spike may have increased global undernourishment by some 6.8 percent, or 63 million people, relative to 2007.29 Moreover, the analysis shows that the sharp slowdown in global growth in 2009 might have contributed to 41 million more undernourished people compared with the number there would have been without the economic crisis.

The problems were compounded by the global economic crisis. A poverty monitoring study of 13 countries suggests that in countries like the Central African Republic and Ghana, parents were forced to take their children out of school and that in other countries they scrambled to finance their children’s continuing attendance.30 In Serbia, Roma children dropped out of school because of a lack of clean clothes and soap. Poor households in Cambodia and the Philippines reported cutting overall consumption in response to income shocks to protect children’s school attendance. Although little information exists on the differential impact of the crisis on women and men, recent surveys of East Asia do not show that women have been disproportionately affected (box 2.7).

Recent surveys in Armenia, Montenegro, and Turkey give a sense of how declines in income induced by the crisis are reducing household consumption.

BOX 2.6Human development suffered severely during crises in developing countries

Household studies from past crises suggest that the impact on human development can be serious. In the lowest-income countries, poverty rises, people eat lower-quality food, school enrollments fall, health care use drops, and infant mortality rises. Even modest reductions in food consumption for children between birth and age two can have lasting deleterious effects on cognitive and physical development. In South Africa and Zimbabwe, the nutritional deprivation of young children led to lower height for age and shorter stature in adulthood.

Analysis of the effects of downturns on infant mortality in Sub-Saharan Africa shows that a 1-percent reduction in per capita GDP is associated with a rise in the infant mortality rate of 0.34-0.48 per 1,000 live births, or 34-39 percent of the average annual change in the infant mortality rate. Infant girls are more likely than boys to die during downturns, and both rural and less educated women are at higher risk of losing their infants.

For middle-income countries the picture is less consistent. In Latin America school attendance has increased during crises, possibly because children are not needed for economic activity, but infant mortality appears to have risen. In Indonesia the crisis of the late 1990s had little measurable impact on schooling or health, possibly because the country was better off and perhaps because education and health services were better protected. But the impacts of recession are far more severe for child health than for education, even in middle-income countries.

Source: Wodon and Zaman 2010; Ferreira and Schady 2009; Dinkelman 2008; Alderman, Hoddinott, and Kinsey 2006; and Gottret and others 2009.

Food consumption in Armenian households has fallen 41 percent, and health care spending is down 47 percent. Some 50-60 percent of households in the four lowest income quintiles have cut back on health care services and drug purchases. Household reductions in food consumption are inversely related to income, with 20 percent of the wealthiest households cutting back (noteworthy in itself) and more than 55 percent of the poorest 20 percent doing so. Even bigger cuts are seen in spending on entertainment and expensive foods. There is some evidence that these cutbacks have helped protect education spending.

In Montenegro unemployment figures suggest that cutbacks affect almost a quarter of households. Safety nets cover only 18 percent of the poorest 20 percent of households, and informal private transfers are disappearing as remittances shrink and informal safety nets unravel. Private investments in education, health insurance, and preventive health care have fallen, reducing resilience to further shocks (figure 2.12). Overall, 9 percent of households reduced preventive care visits, but 25 percent of poor households did, and the same percentage of poor households canceled health insurance. In education the wealthiest households cut back the most—20 percent compared with 11 percent for the lowest-income households.

FIGURE 2.12Spending cutbacks in crisis-affected households are jeopardizing future welfare in Armenia, Montenegro, and Turkey

a. Armenia

Source: Armenia Integrated Living Conditions Survey 2009. See Ersado, forthcoming.

b. Montenegro

Source: Montenegro Crisis Monitoring Survey 2009. See Hirshleifer and Azam, forthcoming.

c. Turkey

Source: TEPAV, UNICEF, and World Bank 2009; Turkey Welfare Monitoring Survey.

In Turkey, the poorest households have experienced the largest reductions in wages and self-employment income. Some 91 percent of the poorest 20 percent of households lost income, but even the wealthiest 20 percent experienced some income loss. Safety nets cover only 20 percent of the poorest households, requiring the rest to sell assets, draw down savings, and find other informal sources of support. Among the poorest households, 75 percent have reduced children’s food consumption, 29 percent have curtailed health care use, and 14 percent have cut back on education spending. Even middle-class households have trimmed spending, especially in education.

The data now available on the impact of the crisis on human development are still much too limited to draw any conclusions on the overall impact. But there is certainly evidence of suffering as a result of the severe global downturn. Even if the deterioration in human development indicators has not been as severe as in previous crises (as speculated above), the human suffering will be considerable.

BOX 2.7Gender differences in impacts of the crisis: Evidence from East Asia

Although the effects of the crisis have clear gender dimensions, it is not clear that women in East Asia have been disproportionately affected. Gender-specific impacts would be expected because of the gender division of labor in the labor market and in the home, gender disparities in access to productive resources, and gender dimensions of household resource allocation. But precise impacts are unclear because they depend on multiple factors including the size of the shock, the economic structure of the country, the nature of government responses, and the speed of economic recovery. Identifying the gender impacts of the crisis is thus an empirical issue.

Empirical analysis is complicated by a lack of data. High-frequency data on the social impacts of the crisis is generally not available, and the lack is particularly intense for gender-disaggregated data. Thus multifaceted approaches are needed, such as rapid qualitative assessments (including focus group discussions), ex ante simulations using precrisis household survey data, analysis of labor force survey data as available, and triangulation across data sources.

Data indicate that unemployment in East Asia has barely changed during the crisis, for men or women, but that women’s participation has tended to rise. In some countries unemployment has fallen more for women than for men, while increases in labor force participation have been more marked for women than for men (see the figures below), particularly in poorer countries, where female labor has shifted from unpaid work to self-employment. Both quantitative and qualitative data indicate longer working hours as men and women take on additional jobs to compensate for falling earnings from primary jobs.

Labor force participation by gender in selected East Asian countries, 2007–09

Source: World Bank staff calculations.

The impact of labor market shocks is driven by several factors, of which gender is only one. Both quantitative and qualitative findings suggest that simple interpretations of labor market data may be misleading. As an example, well-publicized data show layoffs from enterprises producing garments and other products for export to shrinking markets, sectors where female employment tends to dominate. Less well-documented is the contraction in hours and earnings in sectors serving domestic markets, where purchasing power is closely linked to the health of the export sector. These sectors may be dominated by men. Women laid off from formal sector work may be better off than men facing highly restricted earnings in informal sector jobs. Quantitative and qualitative evidence from Cambodia suggests that more male workers in the construction sector have been affected by the crisis than female workers in the garment sector. Moreover, male construction workers are more likely to be poor and have fewer economic fallbacks than female garment workers.

There is no consistent cross-country pattern in differences in hours of paid work by gender. In some countries, such as Cambodia, both men and women have greatly increased their hours of paid work. In other countries, such as Indonesia, women have overtaken men in hours of paid work in the past two years. And in other countries, such as the Philippines, men and women appear to work the same number of paid hours. However, focus group discussions suggest that women’s total work burden (paid plus unpaid domestic work) has increased over the past year. In urban Thailand, women explain that their time on unpaid domestic work has declined a little but not enough to offset rising labor market hours. In rural Cambodia and the Philippines, research teams noted that an increased dependence on common property resources, including firewood, has increased women’s time on domestic chores.

The welfare impacts of the crisis, by gender, also appear to be nuanced. Microsimulations of the poverty impacts of the crisis in Cambodia suggest that male-headed households were more affected in urban areas, while female-headed households were more affected in rural areas (see the figure below). For urban male-headed households, this finding likely reflects the impacts of the crisis on male jobs in construction and tourism. The effects for rural female-headed households appear to reflect the loss of remittance income in addition to more direct crisis impacts on household earnings. Findings from rapid assessments in rural Cambodia indicate that female-headed households commonly cut back consumption sharply and increased their indebtedness to cope with loss of income as remittances from urban areas fell. Male migrant workers—often migrant spouses—reported being unable to return home as often as before because of increased transportation costs and reduced earnings, meaning less male labor on the farm during peak periods.

Impacts of the global financial crisis on male- and female-headed households in Cambodia

Although many people in middle-income countries are above the threshold of the poverty MDG, they are also the hardest hit by adjustments in wage earnings and employment.31

Early evidence in 41 middle-income countries indicates that the impact on the labor market has been severe, especially in wealthier middle-income countries of the Europe and Central Asia region. Although the number of jobs and their growth have been negatively affected, the impact has been mostly on the quality and earnings of employment (figure 2.13).

FIGURE 2.13The crisis sharply reduced wage earnings in middle-income countries

  • Three-quarters of the labor market adjustment stems from slower growth in take-home pay, only one-quarter from less job creation.

  • Earnings in most middle-income countries are falling mainly because people are working fewer hours. Hourly wages have changed little except in Europe and Central Asia, where they have declined.

  • The crisis severely affected labor markets, with few countries spared. It caused a sharp slowdown in wage-bill growth, which fell by an average of 8 percentage points. The exceptions were Argentina, China, and the former Yugoslav Republic of Macedonia, where wage-bill growth accelerated.

  • Employment has shifted away from industrial employment into services, where jobs tend to be of lower productivity and offer lower wages.

  • For a given decline in GDP growth, the labor market impact was more severe in upper-middle-income countries and in countries with fixed exchange rates.

The large impact in Europe and Central Asia resulted mainly from sharp drops in GDP growth. But fixed exchange rates worsen the labor market impact. On average, countries with fixed currency regimes witnessed a decline in employment of 1.7 percentage points, compared with only 0.4 in countries with floating rates. The slowdown in the wage-bill growth was also less severe for the countries with moderate levels of development.

The nature of recent labor market adjustments in these countries suggests that effective policy packages should also focus on supporting earnings and household income, not just generating employment. Responses taken in developed European countries—such as partial unemployment insurance, expanded cash transfers to poor workers, and temporary wage subsidies—may be priority interventions in those countries where hours and earnings adjustments dominated.


Because crises have very negative effects on human development indicators, good policies and institutions are essential in developing countries to avert downturns in the first place, dampen their negative effects when they do occur, and reduce the potential for reversal of reforms. Policy failures, particularly in low-income countries affected by corruption and violent conflict, have been a major reason for the sharp deterioration in human development indicators in past crises.

There are some reasons for hope that the current crisis may be different for low-income countries. A great deal of social spending has been protected so far. Policies and institutions had improved before the crisis. And external shocks, not domestic policy failures, were the main causes of the current crisis. Nonetheless, the impacts on progress toward the MDGs are already worrisome.

While recovery of the global economy appears to be stronger than expected, small reductions in growth could still have lasting negative consequences for poverty and human development. The contraction was so sharp that a long period of strong growth is needed to undo the damage inflicted on development outcomes. The next chapter examines the growth outlook and macroeconomic challenges, including the fiscal tensions created by temporary stimulus measures and protection of social spending.

Annex 2.1: Human and economic indicators during growth cycles

This annex presents more detailed information on the asymmetric impact of growth decelerations on human development indicators, macroeconomic variables, and institutional quality in developing; countries. Tables 2A.–2A.3 show the average level of each indicator during growth accelerations, growth decelerations, other periods, and across all times. Tests for differences in the means of these variables between growth accelerations, decelerations, and all country-year observations show that they are all statistically significant at the 1 percent level.

The conclusion that these indicators tend to deteriorate more in bad times than they improve in good times does not stem from composition effects. It is important to examine these effects because the averages for each period (accelerations, decelerations, and other) do not reflect the same number of observations or equal participation by different income groups—there are more accelerations than decelerations, and low-income countries have greater representation in the sample means during bad times because of the higher frequency of decelerations in these countries (see main text). Because human development indicators are generally lower in low-income than in middle-income countries, the greater frequency of low-income country observations drops the averages for decelerations, which could account for the asymmetric relationship. However, even after controlling for the sample composition effects by comparing the sample means of countries undergoing growth decelerations and accelerations with their own sample means when not in growth decelerations (the column “otherwise” in the three tables), decelerations still have an asymmetric effect.32 Furthermore, the averages for periods not in acceleration or deceleration (normal times) are close to the averages for the entire sample (the last column in each table), providing evidence that the economic cycles are being correctly identified.

TABLE 2A.1Differences between sample averages: Human development and gender indicators
VariableGrowth accelerationGrowth decelerationOtherwise not in acceleration or deceleration)Sample period
Life expectancy at birth, women (years)72.163.469.470.0
Life expectancy at birth, men (years)66.658.164.264.7
Life expectancy at birth, total (years)69.260.766.767.3
Infant mortality rate (per 1000 live births)27.759.739.935.9
Child mortality under-five rate (per 1,000)42.496.359.354.3
Primary completion rate, girls (% of relevant age group)83.249.876.378.6
Primary completion rate, boys (% of relevant age group)84.559.680.281.4
Primary completion rate, total (% of relevant age group)84.454.878.180.2
Ratio of girls to boys, primary enrollment95.686.192.593.6
Ratio of girls to boys, secondary enrollment96.980.794.895.3
Ratio of women to men, tertiary enrollment107.365.1106.2105.4
Source: World Bank staff calculations based on data from World Development Indicators.Note: Tests for differences in the means of these variables between growth accelerations, decelerations, and all country-year observations show that they are all statistically significant at the 1-percent level.
TABLE 2A.2Differences between sample averages: Sub-Saharan Africa
VariableGrowth accelerationGrowth decelerationOtherwise (not in acceleration or deceleration)Sample period
Life expectancy at birth, girls (years)55.252.353.454.0
Life expectancy at birth, boys (years)52.248.950.150.8
Life expectancy at birth, total (years)53.750.551.752.3
Infant mortality rate (per 1.000 live births)80.7106.697.391.9
Child mortality under-five rate (per 1,000)133.5161.3154.3146.2
Primary completion rate, girls (% of relevant age group)55.133.842.147.4
Primary completion rate, boys (% of relevant age group)59.848.450.955.0
Primary completion rate, total (% of relevant age group)57.441.046.551.1
Ratio of girls to boys, primary enrollment89.777.982.485.0
Ratio of girls to boys, secondary enrollment82.363.676.177.7
Ratio of women to men, tertiary enrollment60.232.564.458.7
Source: World Bank staff calculations based on data from World Development Indicators.Note: Tests for differences in the means of these variables between growth accelerations, decelerations, and all country-year observations show that they are all statistically significant at the 1-percent level.
TABLE 2A.3Differences between sample averages: Economic and institutional indicators
VariableGrowth accelerationGrowth decelerationOtherwise (not in acceleration or deceleration)Sample period
Final consumption (% GDP)81.4588.7883.7483.30
Government consumption (% GDP)15.4116.6816.6116.10
Gross capital formation (% GDP)23.7618.5723.3523.10
Gross domestic savings (% GDP)18.5811.2316.2616.70
Gross fixed capital formation private sector (% GDP)16.3510.4313.7514.40
Imports (% GDP)45.8037.4543.8544.10
Exports (% GDP)40.4330.0536.5237.60
Trade (% GDP)86.2367.5080.3781.70
Foreign direct investment. net inflows (% GDP)4.482.073.564.00
Private capital flows, total (% GDP)2.991.402.032.40
CPI inflation (%)14.88251.3237.9043.90
Institutions (–2.5 to 2.5)
Political stability-0.16-0.650.03-0.10
Voice and accountability-0.07-0.470.09-0.02
Regulatory framework-0.03-0.820.150.01
Rule of law-0.14-0.900.12-0.07
Government effectiveness-0.04-0.960.140.00
Frequency of conflicts (Sub-Saharan Africa)0.130.23
Aid to poor countries (Sub-Saharan Africa)
ODA (% GDP)13.8012.10
ODA per capita (US$)69.5041.80
Source: World Bank staff calculations. Data for Sub-Saharan Africa from Arbache, Go, and Page (2008). Indicators on institutions are from the World Bank Institute’s Worldwide Governance Indicators database, which relies on 33 sources, including surveys of enterprises and citizens, and expert polls, gathered from 30 organizations around the world; they each range from -2.5 (worst) to 2.5 (best).Note: Tests for differences in the means of these variables between growth accelerations, decelerations, and all country-year observations show that they are all statistically significant at the 1 percent level. ODA = official development assistance.
Annex 2.2 The special case of HIV/AIDS spending

Large increases in funding have made HIV/ AIDS (human immunodeficiency virus/acquired immune deficiency syndrome) one of the most important items on the development agenda. In less than a decade the international community has mobilized talent and financing to address HIV/AIDS with new institutions and long-term financial commitments to countries suffering from an established and growing epidemic. This attention and financing have produced data that outstrip that available for health care generally, allowing a more thorough examination of trends. The 2008-09 recession is the first global crisis to affect international support for HIV/AIDS spending, and the responses are instructive.

Roughly 33 million people have HIV/AIDS, but only a third of those are on antiretroviral therapy that will extend their life. There is no cure for AIDS. Discontinuities in treatment create resistance to the basic (“first line”) antiretroviral treatment, which can lead to broader drug resistance. The alternative “second line” treatment is 10-20 times more expensive. Thus antiretroviral therapy is central to meeting the MDG 6A to combat HIV/AIDS. Equally important to treating those who have contracted HIV/AIDS is strengthening prevention—the only way to stem the pandemic.

Likely short-term effects of the crisis

Funding for HIV/AIDS has risen sharply over the past decade. During 2001-05, aid commitments for HIV/AIDS programs rose almost 30 percent ($4.75 billion), fueled by the establishment of the Global Fund and by philanthropic efforts by the Clinton Foundation, the Bill & Melinda Gates Foundation, and others. New sources of funding have come onstream since 2005 with the U.S. President’s Emergency Plan for AIDS Relief (PEPFAR) and UNITAID, which disburses much of its resources through the Global Fund.

In 2008 public and private entities allocated $15.8 billion for global HIV/AIDS programs, $6.7 billion of it from bilateral and European Union contributions.33 Pledges to the Global Fund rose from $2.5 billion in 2007 to $3.0 billion in 2008, and then declined to $2.6 billion in 2009. In the last funding cycle (Round 9), demand from countries also fell.34 The U.S. PEPFAR program increased its contributions from $4.5 billion in 2007 to $6.2 billion in 2008 and has subsequently increased its annual budgets. The 2010 fiscal year allocation is just shy of $7 billion, suggesting that U.S. support is continuing.35

Fueled largely by increased donor resources, public health spending in the high-prevalence countries of eastern and southern Africa has risen rapidly in absolute and per capita terms (see map 1.2). As a share of GDP, the increases have gone disproportionately to people with HIV/AIDS.36 Government spending in countries that formerly had high HIV/AIDS prevalence, like Brazil and Thailand, has financed both prevention and treatment. Other countries, such as Ghana, have legally binding commitments ensuring treatment for people with AIDS.

Of 77 countries recently surveyed, most indicated that they had adequate funding from governments, donors, and other sources to finance their current HIV/AIDS programs, but they raised concerns about the future.37 Prevention was identified as the likely victim if funding fell. A further concern was the increased cost of imported drugs and supplies resulting from currency devaluations in some countries.38 The Clinton Foundation recently obtained price concessions from manufacturers that could compensate for the exchange rate penalty.

The impact of the current downturn is not entirely clear, but the uptick in donor spending in 2008 and 2009, when the economic crisis was accelerating in donor countries, is encouraging. The Global Fund disburses quickly once allocations are decided, but recipient country spending has been slow. Almost 40 percent of the Global Fund resources remain undisbursed, a possible source of additional resources if there is a shortfall or delay in funding flows. Almost half the allocations to Sub-Saharan Africa are undisbursed (see figure 2.9). The $900 million allocated in late 2009 under Round 9 is unlikely to have been disbursed yet.39

Although countries may appear to have “adequate” funding for HIV/AIDS, the situation is more nuanced: Some donor funds cannot be applied flexibly, leaving countries with important gaps even when they appear to be highly funded in aggregate terms. This is where the unearmarked flexibility of the International Development Association becomes critical.

The highest-prevalence regions of Africa receive the bulk of external funding (figure 2A.1), but financing per current AIDS patient paints a different picture (figure 2A.2). Although there is a general correlation between the number of patients and funding across countries, financing available for each patient still lags in the highest-prevalence regions of Africa.

FIGURE 2A.1Projected Global Fund to Fight AIDS, Tuberculosis, and Malaria and U.S. PEPFAR HIV/AIDS grants as of April 2009

Source: Lewis 2009.

FIGURE 2A.2Projected Global Fund to Fight AIDS, Tuberculosis, and Malaria and U.S. PEPFAR HIV/AIDS grants per AIDS patient as of April 2009

Source: Lewis 2009.

Greater efficiency is imperative because the agenda has broadened and the pace of infection has not slowed. Targeting high-risk groups and improving management and efficiency in delivery can raise quality and efficiency. The Bahamas plan greater use of generic drugs, better patient adherence to treatment protocols, and a sharper focus on the cost effectiveness of purchases and service delivery. While not costless, such improvements will boost effectiveness and reduce waste, which are equivalent to reducing costs. They also raise the quality of services including health care services.

And what of prevention?

Most international resources are earmarked for treatment. But the only way to stem the need for treatment and save lives is to expand prevention initiatives.

An in-depth evaluation of the U.S. PEPFAR program concluded that it reduced deaths by 5 percent but had no effect on prevention.40 The recent multimillion dollar evaluation of the Global Fund noted the organization’s neglect of prevention.41 A more modest assessment of the programs of the World Bank, Global Fund, and PEPFAR also concluded that prevention was the weak link.42 The challenge is that for every HIV/AIDS patient placed on treatment, two or three newly infected people will need treatment for life.43

Countries that have prioritized prevention—Brazil, Rwanda, and Thailand—have seen prevalence decline or remain low, despite spiraling levels in the early 1990s. Prevalence rates in these countries contrast with those in Botswana and Swaziland, which have struggled to initiate effective prevention programs as prevalence reached epidemic proportions. The long-term trends reflect lack of attention to prevention 5-10 years ago. But current prevention efforts remain inadequate, and the crisis could further curtail such efforts if constrained budgets force cutbacks in prevention. It is a dynamic problem; new infections occur daily, and so a continuous, uninterrupted response is required. It may take 7-10 years for a person to become symptomatic, but even people without evident symptoms can pass on the virus and infect others. Actions now will reduce the rate at which people with the virus can pass it on, underscoring the importance of antiretroviral therapy as a prevention measure.

The Bill & Melinda Gates Foundation and others are financing extensive efforts in prevention technologies, and considerable ongoing research is exploring how to discourage risky behaviors. But equal attention must go to actually promoting behavior change and rolling out promising approaches where prevention lags. Because programs for prevention are dwarfed by those for treatment, the balance deserves some recalibration to spare those not yet infected. While neither simple nor easy, a push to expand prevention is warranted if there is to be progress on Goal 6A: halting the spread of HIV/AIDS by 2015.


    Alderman, H., J.Hoddinott, and B.Kinsey.2006. “Long-Term Consequences of Early Childhood Malnutrition.”Oxford Economic Papers 58 (3): 45074.

    Arbache, J., D.Go, and V.Korman.2010. “Does Growth Volatility Matter for Development Outcomes? An Empirical Investigation Using Global Data.”Background paper for Global Monitoring Report 2010. World Bank, Washington, DC.

    Arbache, J., D.Go, and J.Page,eds. 2008. Africa at a Turning Point? Growth, Aid, and External Shocks. Washington DC: World Bank.

    Arbache, J., and J.Page.2007. “More Growth or Fewer Collapses? A New Look at Long-Run Growth in Sub-Saharan Africa.”Policy Research Working Paper 4384. World Bank, Washington, DC.

    Arbache, J., and J.Page.2010. “How Fragile Is Africa’s Recent Growth?”Journal of African Economies 19 (1): 125.

    Bendavid, E., and J.Bhattacharya.2009. “The President’s Emergency Plan for AIDS Relief in Africa: An Evaluation of Outcomes.”Annals of Internal Medicine 150 (10): 68895.

    Blomquist, J., J. P.Cordoba, M.Verhoeven, C.Bouillon, and P.Moser.2002. “Social Safety Nets in Response to Crisis: Lessons and Guidelines from Asia and Latin America.” In Towards Asia’s Sustainable Development: The Role of Social Protection, pp. 297332. Paris: OECD.

    Bruni, A.Mason, Pabon, and C.Turk.Forthcoming. “Gender Dimensions of the Crisis in East Asia and the Pacific.”World Bank, Washington, DC.

    Case, A., and C.Paxson.2006. “Stature and Status: Height, Ability, and Labor Market Outcomes.”NBER Working Paper 12466. National Bureau of Economic Research, Cambridge, MA.

    Dang, H., S.Knack, and H.Rogers.2009. “International Aid and Financial Crises in Donor Countries.”Policy Research Working Paper 5162. World Bank, Washington, DC.

    De Walque, D., H.Kazianga, M.Over, and J.Vaillant.2010. “The Impact of the Food Crisis on HIV/AIDS Treatment and Health Outcomes in Mozambique.”Paper presented at World Bank Development Economics Development Research Group Workshop, January13, Washington, DC.

    Dinkelman, T.2008. “The Long-Term Effects of Being Born in a Drought: Evidence from the Cape Area Panel Study 2002-06.”Paper presented at the Northeastern Universities Development Consortium Conference, November 8-9, Boston University, Boston, MA.

    Ersado, L.Forthcoming. “The Global Economic Crisis in Armenia: Policy Responses and Household Coping Strategies.”World Bank Working Paper. Washington, DC.

    Ferreira, F., and N.Schady.2009. “Aggregate Economic Shocks, Child Schooling, and Child Health.”World Bank Research Observer 24 (2): 14181.

    Filmer, D., and L.Pritchett.1999. “The Impact of Public Spending on Health: Does Money Matter?”Social Science and Medicine 49 (10): 130923.

    Gottret, P., V.Gupta, S.Sparkes, A.Tandon, V.Moran, and P.Berman.2009. “Protecting Pro-poor Health Services during Financial Crises: Lessons from Experience.”In Innovations in Health System Finance in Developing and Transitional Economies: Advances in Health Economics and Health Services Research, vol. 21, ed.DovChernichovsky and KaraHanson.Bingley, U.K.: Emerald Group Publishing.

    Grosh, M., C.del Ninno, C.Tesliuc, and A.Ouerghi.2008. For Protection and Promotion: The Design and Implementation of Effective Safety Nets. Washington, DC: World Bank.

    Hausmann, R., L.Pritchett, and D.Rodrik.2005. “Growth Accelerations.”Journal of Economic Growth 10 (4): 30329.

    Hirshleifer, S., and M.Azam.Forthcoming. “Impact of the Economic Crisis in Montenegro: Evidence from Crisis Response Survey.”ECSHD Working Paper. World Bank, Europe and Central Asia Region, Human Development Unit, Washington, DC.

    IMF (International Monetary Fund). 2005. “Petroleum Product Subsidies: Costly, Inequitable, and Rising.”IMF Staff Position Note SPN/10/05. Washington, DC.

    IMF (International Monetary Fund). 2009. “Fiscal Policy in Sub-Saharan Africa in Response to the Impact of the Global Crisis.”IMF Staff Position Note SPN/09/10. Washington, DC.

    Kaiser Family Foundation. 2009. Updates of various information, including “The U.S. President’s Emergency Plan for AIDS Relief (PEPFAR),” Fact sheet on U.S. Global Health Policy. Washington, DC (November) (

    Khanna, G., D.Newhouse, and P.Paci.Forthcoming. “Fewer Jobs or Smaller Paychecks? Labor Market Impacts of the Financial Crisis in Middle Income Countries.”World Bank, Washington, DC.

    Laing, R., and I.Buysse.2010. “Impact of the Global Crisis on the Global Pharmaceutical Sector.”Paper presented (January7) at University of Utrecht, the Netherlands.

    Lewis, M..2009. “Likely Impacts of the Crisis on HIV/AIDS Programs.”Paper presented to HIV/AIDS Economics Reference Group Meeting, April21, Washington, DC.

    Lewis, M., and M.Verhoeven.2010. “Financial Crises and Social Spending: The Impact of the 2007–2009 Crisis.”Background paper for Global Monitoring Report 2010. World Bank, Washington, DC.

    Ooman, N., M.Bernstein, and S.Rosenzweig.2007. Following the Funding for HIV/AIDS. Washington, DC: Center for Global Development.

    Pritchett, L.2000. “Understanding Patterns of Economic Growth: Searching for Hills among Plateaus, Mountains, and Plains.”World Bank Economic Review 14: 22150.

    Raddatz, C.2007. “Are External Shocks Responsible for the Instability of Output in Low-Income Countries?”Journal of Development Economics 84 (1): 15587.

    Raddatz, C.2008a. “External Shocks and Macro-economic Volatility in Latin America.”World Bank, Washington, DC.

    Raddatz, C.2008b. “Have External Shocks Become More Important for Output Fluctuations in African Countries?” In Africa at a Turning Point? Growth, Aid, and External Shocks, ed.J.Arbache, D.Go, and J.Page.Washington, DC: World Bank.

    Rajkumar, A., and V.Swaroop.2008. “Public Spending and Outcomes: Does Governance Matter?”Journal of Development Economics 86 (1): 96111.

    Ratha, D., S.Mohapatra, and A.Silwal.2009. “Migration and Remittance Trends 2009: A Better-than-Expected Outcome So Far, but Significant Risks Ahead.”Migration and Development Brief 11.World Bank, Washington, DC.

    Revenga, A., M.Over, E.Masaki, W.Peerapatanapokin, J.Gold, V. T.Sathien, and S.Thanprasertsuk.2006. “The Economics of Effective AIDS Treatment Evaluating Policy Options for Thailand.”World Bank, Washington, DC.

    Sabarwal, S., N.Sinha, and M.Buvinic.2009. “Aggregate Economic Shocks and Gender Differences: A Review of the Evidence.”World Bank, Washington, DC.

    Sherry, J., S.Mookherji, and L.Ryan.2009. The Five-Year Evaluation of the Global Fund to Fight AIDS, Tuberculosis and Malaria. Calverton, MD: ICF Macro International.

    TEPAV (Economic Policy Research Foundation), UNICEF (United Nations Children’s Fund), and World Bank. 2009. “The Economic Crisis and the Welfare of Families: Results from the Five Urban Centers in Turkey.”Ankara.

    Tiwari, S., and H.Zaman.2010. “The Impact of Economic Shocks on Global Undernourishment.”Policy Research Working Paper 5215. World Bank, Washington, DC.

    Turk, C.2009. “Rapid, Qualitative Assessments of the Impacts of the Economic Crisis: Overview of Findings from Eight Countries.”Internal discussion paper. World Bank Social Development Department, Washington, DC.

    Turk, C., and A.Mason.2009. “Impacts of the Economic Crisis in East Asia: Findings from Qualitative Monitoring in Five Countries.”Paper presented at the 3rd China-ASEAN Forum on Social Development and Poverty Reduction and 4th ASEAN +3 High Level Seminar on Povery Reduction, September28–30, Hanoi. (

    UNAIDS (Joint United Nations Programme on HIV/AIDS). 2009. “25th Meeting of the UNAIDS Programme Coordinating Board.”Geneva (December8–10).

    Wagstaff, A., and M.Claeson.2004. Rising to the Challenges: The Millennium Development Goals for Health. Washington, DC: World Bank.

    Wodon, Q., and H.Zaman.2010. “Higher Food Prices in Sub-Saharan Africa: Poverty Impact and Policy Responses.”World Bank Research Observer 25 (1): 15776.

    World Bank. 2006. Global Economic Prospects: Economic Implications of Remittances and Migration. Washington, DC: World Bank.

    World Bank. Forthcoming. “El Salvador Public Expenditure Review.”World Bank, Washington, DC.

    World Bank. Various years. World Development Indicators. Washington, DC: World Bank.

    World Food Programme. 2009. “Households Go Hungry As Financial Crisis Bites.”Rome (June11).

Arbache, Go, and Korman 2010. Although the aggregate figures show girls’ education is affected by growth cycles, there is still a lack of microstudies that show girls are disproportionately more likely to be pulled out of schools during covariate shocks. However, some of these more adverse effects may be occurring in conflict or disastrous situations with institutional breakdowns so that microstudies are not available.

For the entire sample of developing countries, 47 percent of the 4,415 country-year observations are classified as growth accelerations while 11 percent are classified as growth decelerations. The remaining 42 percent of observations are for years in which countries experienced neither growth acceleration nor deceleration episodes.

To some extent this pattern may be endogenous, because average income per capita tends to rise in countries with more frequent growth accelerations and fall in countries with more frequent collapses.

Arbache, Go, and Page 2008. The inflation figure would have been higher had Zimbabwe been included; it was excluded from the analysis because of missing data for other variables.

The analysis is taken from Lewis and Verhoeven (2010) and relies on data from the International Monetary Fund (IMF), the World Bank, United National Educational, Scientific, and Cultural Organization (UNESCO, education spending), and the World Health Organizationv (WHO) National Health Accounts (health spending).

The absence of a consistent time series in education spending data required the integration of data from UNESCO, the IMF, and the World Bank. This is in contrast to the consistent and much higher quality data from WHO National Health Accounts.

Gottret and others 2009.

Ferreira and Schady 2009.

High-Level Seminar on Africa Fiscal Policy for Growth in Light of the Global Crisis, Maputo, December 2009, sponsored by the World Bank and various governments.

Wodon and Zaman 2010.

These data do not include safety net and nutrition interventions under the World Bank’s Global Food Crisis Response Program, which has funded an estimated $380 million for safety net interventions in 21 countries, including grant funding for small, targeted projects in 17 low-income IDA-eligible countries (totaling $95 million).

See Grosh and others (2008) for detailed discussion of the design and implementation challenges of effective safety nets.

World Food Programme 2009. The study analyzes 126 countries and focuses on five case studies.

Countries are rated according to performance in 16 areas grouped into four clusters: economic management, structural policies, policies for social inclusion and equity, and public sector management and institutions.

Arbache, Go, and Page 2008.

Khanna, Newhouse, and Paci, forthcoming.

See Arbache, Go, and Korman (2010) for more discussion.

The figure includes international and domestic philanthropic contributions, World Bank financing, government expenditures, and household spending, but it excludes other multilateral and private sector funding.

The Global Fund has initiated a new $2.6 billion funding window for the next two years, which it estimates is insufficient. These requirements are not addressed here because the focus is on financing HIV/AIDS prevention and treatment.

Kaiser Family Foundation (2009) and provide updates of spending.

Case and Paxson 2009.

A survey of UNAIDS and WHO country offices by the World Bank, UNAIDS, and WHO (2009) asked about possible issues as the crisis evolved and the likely impact on HIV/AIDS programs over the next 6–12 months.

UNAIDS 2009.

    Other Resources Citing This Publication