The IMF and the Poor

IMF-Supported Programs and the Poor:
The Experiences of Low-Income Countries

In 1987, the IMF established the Enhanced Structural Adjustment Facility (ESAF) to provide resources to low-income countries for longer periods on concessional terms (see Box). Like its precursor, the Structural Adjustment Facility (SAF), the ESAF was created in response to a need to better address the macroeconomic and structural problems of low-income countries. In a review of 36 countries (International Monetary Fund, 1997a; see Table 1) that have implemented structural adjustment policies under SAF/ESAF-supported programs during 1986-95, the IMF found that these countries have made substantial progress in creating the conditions for a stable macroeconomic environment and sustainable growth and in improving the composition of their public expenditures. An even more recent analysis of data for 66 countries (Gupta, Clements, Verhoeven, and Tiongson, 1998; see Table 2), of which 32 had SAF/ESAF programs, shows an increase in social spending since the mid-1980s. 2 What has been the impact of such IMF-supported programs on the poor? The following section reviews the experiences of these countries and changes in key variables that affect the poor––inflation, growth, and the composition of public expenditure.

Box. SAF/ESAF: A Concessional Facility to Assist Poorer Countries
The IMF's Executive Board established the Enhanced Structural Adjustment Facility (ESAF) in 1987 to better address the macroeconomic and structural problems faced by low-income countries. It offers loans with lower interest rates and for longer terms than the typical IMF market-related arrangements. The principal objectives are to promote balance of payments viability and foster sustainable long-term growth. Although the objectives and features of the ESAF are similar to those of its predecessor, the Structural Adjustment Facility (SAF), set up in 1986, the ESAF was expected to be more ambitious with regard to macroeconomic policy and structural reform measures. The IMF no longer makes disbursements under the SAF.

ESAF loans are disbursed semiannually (as against quarterly for regular IMF stand-by arrangements), initially upon approval of an annual arrangement and subsequently on the observance of performance criteria and after completion of a mid-term review. ESAF loans are repaid in 10 equal semiannual installments, beginning 5½ years and ending 10 years after the date of each disbursement. The interest on ESAF loans is 0.5 percent a year. By contrast, charges for stand-by arrangements are linked to the IMF's SDR market-determined interest rate, and repayments are made within 3¼ to 5 years of each drawing. A three-year access under the ESAF is up to 190 percent of a member's quota. The access limits of stand-by arrangements are 100 percent of quota annually and 300 percent cumulatively.

An eligible member seeking to use ESAF resources develops, with the assistance of the IMF and the World Bank, a policy framework paper (PFP) for a three-year adjustment program. The PFP, updated annually, sets out the authorities' macroeconomic and structural policy objectives and the measures that they intend to adopt during the three years. The PFP also lays out the associated external financing needs of the program, a process that is meant to catalyze and help coordinate financial and technical assistance from donors in support of the adjustment program.

Table 1. Sample of 36 SAF/ESAF Countries

BoliviaKyrgyz RepublicPakistan
Burkina FasoLao People's Dem. Rep.Senegal
BurundiLesothoSierra Leone
CambodiaMadagascarSri Lanka
Côte d'IvoireMalawiTanzania
Equatorial GuineaMaliTogo
Gambia, TheMauritaniaUganda

Table 2. Sample of 66 Countries with IMF-Supported Programs

Albania*El SalvadorNiger*
BarbadosGuatemalaPapua New Guinea
Burkina Faso*Kenya* São Tomé and Príncipe*
Burundi*KoreaSierra Leone*
Cambodia*Kyrgyz Republic*Sri Lanka*
Congo, Republic of*MexicoTurkey
Costa RicaMoldovaUkraine
Côte d'Ivoire Mongolia*Uruguay
Czech RepublicMozambique*Venezuela

*Countries with SAF/ESAF-supported programs.


Overall, IMF-supported programs have been most successful in ending episodes of high inflation––when the rate is over 40 percent––which strongly correlates with slow growth. However, countries have been less successful in achieving low (single-digit) inflation (International Monetary Fund, 1997a). The countries with high initial inflation rates experienced a sharp reduction in inflation, and the vast majority of them managed to exit these high-inflation situations. Countries with intermediate inflation (the largest group) showed a mixed record. Many countries saw some disinflation, although in some cases it amounted to a reversal of a run-up of inflation before the program began. By the third year of the program, only about one-fourth of the countries with intermediate initial inflation had managed to reduce it to low rates. In one-third of the countries with low initial inflation rates, inflation rose to the intermediate level of about 16 percent. However, the rest of the group, on average, experienced slightly reduced inflation.


During the early 1980s, developing countries, on average, experienced virtual stagnation in per capita real GDP. In the 36 program countries reviewed, the picture was even bleaker, as real GDP declined by 1.4 percent a year on average in the first half of the 1980s. In the 10 years thereafter (1986-95), growth picked up in the developing countries as a group, but was faster in the sample of program countries, leading to the elimination by 1995 of the gap between their growth rates and the average of other developing countries. However, this improvement varied among regions. Western Hemisphere countries experienced the sharpest turnaround in growth rates, whereas those of African countries with a SAF or an ESAF in place turned negative in the early 1990s, so that per capita GDP growth rates, on average, remained close to zero.

A comparison of growth in the three years after the start of such programs, relative to their growth in the three years preceding the programs, shows the following picture: all 36 program countries reviewed, on average, were able to increase real per capita GDP growth. The turnaround was most dramatic in African countries outside the CFA franc zone 3 and in countries in the Western Hemisphere.

The results of a standard empirical model (International Monetary Fund, 1997a) used to examine the determinants of economic growth suggest that part of the marked narrowing of the growth gap between developing countries with programs and those without is attributable to stronger macroeconomic policies. These policies were designed to reduce both inflation and budget deficits and were reinforced by structural reforms in key areas. Other factors were more benign exogenous influences, particularly trends in the terms of trade and the weather. However, policies fell short in several areas––most notably by not opening up to international trade and not reducing the size of government.

Composition of Expenditures

Social expenditures

On the basis of evidence for the 66 countries with IMF-supported programs for which data are available during 1986-96, spending on education and health, on average, fared reasonably well. Changes in such spending were larger in the 32 countries supported by a SAF or an ESAF program. While changes in social indicators varied among countries, the indicators did, in general, improve. The data reported here do not include education and health spending by the private sector, which can be sizable.4

Many countries included in the sample sought real increases in education and health expenditures as well as improvements in expenditure efficiency. A comparison of the last year for which data are available and the preprogram year (defined as the year preceding the first IMF-supported program for each country) shows that real public spending on education increased by 44 percent (43 percent in SAF/ESAF countries). Real per capita spending increased on average by 0.9 percent a year in the 66 countries and 2.8 percent a year in countries with a SAF/ESAF program (Figures 1 and 2). However, this increase in education spending varied substantially across regions. African countries increased spending by less than the others in the sample: education spending rose only 2 percent in real terms a year, compared with 11 percent a year in other SAF/ESAF countries. Because of the small increase in such spending and the high population growth in Africa, per capita education spending actually declined by 0.7 percent a year, on average, compared with a rise of 9 percent a year in the other SAF/ESAF countries.5 In the 66 countries, real public health expenditures increased by 61 percent (79 percent in SAF/ESAF countries). Health expenditures increased in real terms in all but eight countries with a SAF or an ESAF program, with per capita spending increasing, on average, by 2.8 percent a year (Figures 1 and 2). As with education expenditures, African countries with SAF or ESAF programs increased their spending by less, with their real per capita outlays on health increasing by only 2.5 percent a year, compared with 3.3 percent elsewhere in the sample of SAF/ESAF countries. For both education and health, their shares in total expenditures increased, indicating that these sectors became higher priorities under IMF-supported programs.

Figure 1. Average Annual Change in Real Per Capita Education
and Health Spending Under SAF/ESAF-Supported Programs


Sources: Country authorities; and IMF staff estimates.

Figure 2. Government Social Spending Under
IMF-Supported Programs, 1986-96

(Mean changes)

Sources: Country authorities; and IMF staff estimates.
Note: The mean changes are between the preprogram year and the latest year for which data are available.

While many factors can potentially affect the link between government expenditures and social indicators, the increased spending on education and health care coincided with improvements in both education and health indicators. The illiteracy rate has declined by 2.4 percent a year since the start of the first IMF-supported program, with comparable improvement in SAF/ESAF countries. Gross primary and secondary enrollment rates rose more sharply in SAF/ESAF countries (about 12 percent on average between the last year for which data are available and the preprogram year) compared with the overall sample of 66 countries (about 6-9 percent on average for the same period). Again, gains in literacy were comparatively lower in the African SAF/ESAF countries, with the increase in gross primary enrollment rates being typically lower than in other countries (Figure 3).

Figure 3. Average Annual Change in Primary Education
Enrollment and Infant Mortality Rates Under
SAF/ESAF-Supported Programs

Sources: World Bank, World Development Indicators database; and UNESCO Statistical
, various issues.
Note: An improvement in infant mortality rates is indicated by an increase.

In the health sector, access to health care increased on average to 79 percent of the population from 69 percent, with a sharper rise (60-74 percent) in SAF/ESAF countries. Immunization rates and access to safe water and sanitation also increased, with some SAF/ESAF countries experiencing greater improvements. At the same time, life expectancy increased by 0.3 percent a year (0.2 percent in SAF/ESAF countries), and infant mortality fell by 1.8 percent a year (1.7 percent in SAF/ESAF countries). Improvements in health in the African countries were smaller than in other countries in the sample.

As mentioned above, the poverty-reducing impact of social expenditures depends largely on the intrasectoral allocation of these outlays. In the earlier review of SAF/ESAF countries for 1986-95, Abed and others (1998) showed that the distribution of benefits from social spending disproportionately favored higher-income groups. The ability of upper-income groups to capture a disproportionate share of the benefits of this spending may have reflected an urban bias in the provision of social services.6 In the education sector, the poorest 20 percent of the population in a group of eight countries received 13 percent of the benefits on average, compared with 32 percent for the richest 20 percent. To improve the benefit incidence, government should spend more on primary education and impose user charges for tertiary education, in combination with a system of financial aid for poorer students. For the five SAF/ESAF countries for which health data are available, the poorest 20 percent received an average of  just 12 percent of the benefits of total health care spending, compared with 30 percent for the richest 20 percent (see Abed and others, 1998). Again, countries could improve the benefit incidence by emphasizing primary and preventive health care services and focusing less on curative and hospital care. The findings of Abed and others (1998) are supported by a more recent analysis by Gupta, Clements, and Tiongson (forthcoming). In a sample of 46 countries, the average share of education spending allocated to tertiary education was found to be 21 percent, with Asian countries allocating the lowest share. While it is difficult to determine the appropriate sectoral shares of education expenditures, it is noteworthy that Asian countries with IMF-supported programs have experienced the sharpest improvements in education indicators. This same study found that, in a sample of 33 countries, 60 percent of health spending is absorbed by high-cost curative care, with sub-Saharan African countries allocating the largest share. This finding may partly explain why improvements in overall health indicators among program countries have been more modest in African countries.

The Managing Director of the IMF has recently called on IMF staff to improve the collection of data on health and education expenditure and to monitor developments in basic social indicators in developing and transition economies. The Executive Board has requested that the staff pay particular attention to social spending and social indicators in the most heavily indebted poor countries (see International Monetary Fund, 1997c), which are receiving assistance under an initiative instituted by the IMF and the World Bank.

Military expenditures

Excessive military spending can crowd out private investment or more productive public expenditures, with a potentially adverse effect on growth (Knight, Loayza, and Villanueva, 1996). Recent research also confirms that in countries where large increases in military expenditures have occurred, the budget deficit has increased and public investment has declined (Gupta, Schiff, and Clements, 1996). The IMF therefore often advises countries to review military expenditures to identify potential fiscal savings.

Progress has been encouraging in recent years (Gupta, Clements, and Ruggiero, 1997; Gupta, McDonald, and Ruggiero, 1998). In developing countries with IMF-supported programs, the declines in worldwide military outlays have been even more rapid, although one should be cautious in attributing these declines purely to the programs. In these countries, military expenditures fell, on average, by 2.5 percentage points of GDP between 1990 and 1997, compared with 1.3 percentage points in developing countries without an IMF-supported program. This larger decline in program countries is due mainly to the relatively more rapid reduction in military expenditures in the transition economies (former centrally planned countries making the transition to market economies). The decline in military expenditures has allowed social expenditures to expand. For the 52 countries for which consistent data are available for 1990-96, military spending fell by an average of 3.1 percentage points of total expenditures, while social expenditures increased, on average, by 1.2 percentage points.

Other aspects of expenditure composition

The review of the 36 countries' performance during 1986-95 showed that SAF/ESAF-supported programs aimed, on average, to roughly maintain the level of total expenditures as a share of GDP (see Table 3 and Figure 4) while shifting expenditures from current to capital. Relative to the three-year preprogram average, an increase in capital expenditures and net lending of about 1.4 percentage points of GDP was targeted. This increase in capital expenditures was to be facilitated by an average reduction in current spending of 2.2 percentage points, with budgetary savings anticipated to come from reductions in excessive public sector employment and in inefficient subsidies and transfers. In practice, these countries did make significant progress in changing the composition of expenditures in favor of capital outlays, although by less than programmed. Compared with the three-year preprogram average and the last year for which data are available, the share of outlays devoted to capital and net lending rose by about 2.6 percentage points, whereas the portion of expenditures absorbed by wages and salaries as well as by subsidies and transfers declined (Table 3 and Figure 4).

Table 3.Summary of Expenditures by Economic Classification in SAF/ESAF Countries

Last Year
Minus Last Year
Three-Year Average Last Year Three-Year Minus
Preprogram Preprogram Program (1994 or Preprogram Preprogram Sample
Average Year Target 1995) Average Year Size1
(Percent of GDP)2
Total expenditures and net
29.0 27.6 28.2 26.2 -2.8 -1.4 36
   Current expenditures 20.9 19.7 18.7 17.9 -3.0 -1.8 36
      Goods and services 13.6 12.6 11.9 11.1 -2.5 -1.5 28
          Wages and salaries 7.1 6.5 6.1 6.2 -0.9 -0.3 33
          Other 6.3 5.8 6.0 4.9 -1.4 -0.9 28
       Interest 4.0 3.9 4.0 3.6 -0.4 -0.3 35
       Subsidies and transfers 4.4 3.9 3.2 3.2 -1.2 -0.7 29
       Other 3.9 3.9 5.1 3.6 0.3 -0.3 21
    Capital expenditures and        net lending 7.9 7.7 9.3 8.1 0.2 0.4 36
(Percent of total expenditures and net lending)
Total expenditures and net
    Current expenditures
70.5 69.6 65.4 67.8 -2.7 -1.8 36
       Goods and services 44.4 43.4 40.5 40.9 -3.5 -2.5 28
          Wages and salaries 24.9 23.7 22.4 23.5 -1.4 -0.2 33
          Other 19.7 19.9 19.5 18.0 -1.7 -1.9 29
       Interest 12.3 12.5 13.5 13.5 1.2 1.0 35
       Subsidies and transfers 12.4 12.6 9.6 11.4 -1.0 -1.2 31
       Other 16.1 15.9 17.9 15.3 -0.8 -0.6 20
    Capital expenditures and        net lending 28.9 29.7 34.0 31.5 2.6 1.8 36

Sources: Country authorities; and IMF staff estimates.
1 Number of countries for which data are available for a given expenditure category. If the sample size varies for different columns, then the maximum figure is given.
2 The sum of the expenditure components may differ from the totals because of differences in sample size.

Figure 4. Government Spending Under SAF/ESAF-Supported Programs
(Percent of GDP)

Sources: Table 3.
1 Either 1994, 1995, or 1996.

Social safety nets

Although data problems make it difficult to distinguish subsidies designed to help the poor from those designed for other policy purposes, several countries improved the efficiency of outlays on subsidies and transfers by reducing generalized subsidies and increasing spending on targeted social safety nets during 1986-95. Partly as a result of such reductions (particularly in transition economies), spending on subsidies and transfers by the group of 36 SAF/ESAF countries declined by about 1 percent of total expenditures between the last year for which data were available and the three-year preprogram average. Some countries temporarily froze prices on key staples after major exchange rate devaluations, whereas others instituted programs to transfer income to the poor during the reform period.

When civil service reform involved cutting staff, most countries included severance packages for the newly unemployed and made it easier for small enterprises to obtain credit, often with the support of donor financing. Public works or "food-for-work" programs were also implemented to provide income support to the unemployed, those adversely affected by reduced employment opportunities, and soldiers rendered unemployed by military demobilization programs. In some other countries, a part of the privatization proceeds was used to compensate workers who were made redundant in privatized enterprises.

Countries that have attempted to reform existing social assistance programs and implement new social safety nets have faced several difficulties, and the results have often been uneven. First, weak administrative structures and the lack of appropriate social policy instruments have constrained the implementation of cost-effective social safety nets, especially when the poor are located in geographically isolated areas. Furthermore, because of a lack of data, it has sometimes been difficult to assess a social safety net's effectiveness in reaching its intended beneficiaries. In some countries, political support for establishing and reforming social safety nets has at times been insufficient, so that social safety nets continue to have unintended beneficiaries. Finally, the weakening of the revenue base has dried up sources of financing for social benefits in some transition countries (Chu and Gupta, 1996).

2 The findings in the sample of 66 countries largely confirm an earlier analysis of developments in social spending in 23 countries supported by SAF or ESAF programs (see Abed and others, 1998).
3 The 14 countries that make up the CFA franc zone are Benin, Burkina Faso, Cameroon, Central African Republic, Chad, Republic of Congo, Côte d'Ivoire, Equatorial Guinea, Gabon, Guinea-Bissau, Mali, Niger, Senegal, and Togo.
4 In 1990, for example, the private sector accounted on average for 45 percent of health outlays in Africa and for 60 percent of such outlays in Asia (see Psacharapoulos and Nguyen, 1997)
5 The performance in Africa was influenced by the drop in spending in CFA franc zone countries in the aftermath of the 1994 currency devaluation and by the repatriation of expatriate teachers. In some African countries, such as Burkina Faso, Ghana, and Lesotho, real per capita education expenditures increased significantly.
6 See Schwartz and Ter-Minassian (1995) for a recent review of the literature on the distributive incidence of public expenditures.