IMF Working Paper: Higher shares of social spending do not guarantee better social conditions in sub-Saharan Africa
Some social outlays appear to be rising but show considerable volatility
1Excluding Mauritius, South Africa, and Seychelles, where considerably higher levels of social spending distort typical trends for sub-Saharan African countries.
Data: World Bank, World Development Indicators, 2001; and IMF, Government Finance Statistics, 2001, and World Economic Outlook database
In the mid-1990s, international development assistance to sub-Saharan Africa began to address social priorities more explicitly. While continuing to emphasize policies designed to generate growth and correct financial imbalances, the IMF, too, has acknowledged poverty reduction as the central objective of policy, to which growth is the key. Particularly since the establishment of its Poverty Reduction and Growth Facility in 1999, the IMF has begun encouraging African countries to pay closer attention to the composition of government spending and to expand or at least preserve shares of social spending in their budgets from year to year. Although usually a step in the right direction, this approach cannot guarantee that actual spending will be adequate to improve social conditions in a meaningful way.
“A Comparative Analysis of Government Social Spending Indicators and Their Correlation with Social Outcomes in Sub-Saharan Africa” supports the case for increased scrutiny of social spending in the region by highlighting actual connections between social spending indicators and social outcomes using the most comprehensive data available for the region. The working paper also suggests some policy responses that are relevant not only for governments in the region but also for the international donor community in its role in supporting social programs.
How is social spending measured?
Identifying public spending that is truly social is a difficult exercise because there is no gauge of social merit that applies to every individual situation. Based on a functional classification of government expenditure, social spending is usually understood as funding for public health care, education, housing, and other social services. For most analytical purposes, social spending is therefore often proxied by budgetary outlays on health care and education, which tend to be the most representative and easily identifiable categories of such spending.
Social spending is usually measured in one of three ways: in U.S. dollars per capita, as a percentage of GDP, or as a percentage of total government expenditure. Each measure has both merits and shortcomings in its ability to gauge the adequacy of social spending. Per capita spending allows international comparisons of absolute spending; percentages of GDP give an idea of spending relative to the size of the economy; and shares of government spending—a more discretionary indicator—give some sense of policy direction and potential. But the policy assumption underlying the use of all these measures is that the higher the social spending, the better the social outcomes.
Higher spending, better outcomes?
Key health and education indicators in sub-Saharan Africa have generally improved over the past 20 years. For example, infant mortality declined from about 115 deaths per 1,000 live births in 1980 to about 93 deaths in 1999. In some areas, improvements have been more dramatic: illiteracy among youths aged 15 to 24 declined from 45 percent in 1980 to almost 23 percent in 1999; and the share of infants under 12 months who were immunized rose from 23 percent in 1980 to almost 59 percent in 1999. But there have been setbacks as well. For example, life expectancy increased from almost 48 years in 1980 to about 50 in 1990 but then dropped to about 47 in 1999. This and other adverse developments in health and education indicators vary significantly between countries and are often associated with armed conflicts or HIV/AIDS.
Although the generally improving trends reflect the increasing access to social services in the region, it is not evident that access can be attributed primarily to more generous government social spending policies. If it could be, one would expect to see a parallel rising trend in social spending. Although the evidence is not as clear-cut, the trends in health and education spending appear to be rising over time (see chart, page 90), but with more or less volatility depending on the indicator. Interestingly, at the end of the survey period, all the indicators of social spending rose sharply, suggesting that the heightened emphasis on social spending in development assistance since the mid-1990s may indeed have led to greater social spending in sub-Saharan African countries. It is also possible that, as suggested by the higher frequency of data reporting in the more recent years, the greater interest in social expenditures may have led to their more comprehensive tracking and reporting.
Are social spending, outcomes linked?
To evaluate the relationship between social spending and social indicators in sub-Saharan Africa, the study relied on comprehensive regional data for 48 countries during 1980–99, looking for correlations between 10 social indicators (5 each for health and education) and each of the three measures of relevant government spending (in either health or education). The objective was to determine the relative raw explanatory power of the three specifications in terms of each of the three measures.
The results confirmed some correlation between social outcomes and social government spending in U.S. dollars per capita and in percent of GDP, but not in shares of government spending. Spending indicators were better at explaining education outcomes than health outcomes, but, with the exception of female secondary enrollment, none of the spending indicators was by itself very powerful in explaining social outcomes.
This should not come as a surprise given that social conditions depend on significantly more than current levels of social spending. The fact is that prevailing socioeconomic levels will determine social indicators in the immediate future more than any changes in social spending. Indeed, it may take a generation or longer for higher spending to have a significant impact on social indicators that cover individuals too old or handicapped to change their health or education status.
To improve the results, the study therefore controlled for national welfare, as proxied by income per capita in U.S. dollars. The addition of this variable confirmed the explanatory strength of the two specifications of government social spending previously shown to be more significant—that is, U.S. dollars per capita and percent of GDP. It did not statistically improve the specifications in terms of shares of government spending, which remained all but insignificant.
What are the policy implications?
The econometric and survey results suggest that absolute levels of social spending matter the most for social outcomes and that budgetary allocations may be misleading. The lack of evidence that higher shares of government spending go hand in hand with better social indicators suggests that increasing such shares is helpful only to the extent that it results in an actual increase in absolute spending. This can be expected in a stable context of rising annual government budgets, which cannot be taken for granted in sub-Saharan Africa, where budgets are frequently affected by, for example, swings in commodity prices or variations in external assistance. Thus, policymakers should not treat rising shares of government spending as evidence that more resources are being channeled to priority social sectors or as a sign that social indicators can be expected to improve.
The recognition that absolute social spending allocations are paramount in determining social outcomes also has major implications for international assistance to sub-Saharan Africa, where many public social programs depend directly on financing from bilateral donors, multilateral agencies, or nongovernmental organizations. This dependency is all the more crucial since the decline in per capita external assistance to the region in line with global reductions in aid budgets. According to the World Bank, total net official development assistance fell, on a per capita basis, from $36 in 1990 to $20 in 1999. Although spending has not fallen correspondingly, those reductions may have taken a toll, or may do so with a lag, on regional social indicators.
Against this background, donors deciding nominal assistance levels and their regional distribution should take into account alternative measures of social spending and their likely impact on social indicators. More generally, by considering alternative measures of social spending, government officials, donors, and other policymakers might gain useful insights that allow them to choose between competing social programs. For example, by understanding better how social indicators respond to changes in spending, they may be able to identify implementation constraints, improve delivery strategies, or, in the case of the IMF, enhance program design. For sub-Saharan Africa, any such increased insight into how social conditions can be ameliorated can only help with the decisions that may, eventually, allow the region to overcome its harsh social realities.
Copies of IMF Working Paper 02/176, “A Comparative Analysis of Government Social Spending Indicators and Their Correlation with Social Outcomes in Sub-Saharan Africa,” by Paulo Silva Lopes, are available for $15.00 each from IMF Publication Services. See below for ordering information. The full text is also available on the IMF’s website (http://www.imf.org).