Information about Sub-Saharan Africa África subsahariana
Chapter

5 Debt Overhang and Economic Growth in Sub-Saharan Africa

Editor(s):
Zubair Iqbal, and S. Kanbur
Published Date:
September 1997
Share
  • ShareShare
Information about Sub-Saharan Africa África subsahariana
Show Summary Details
Author(s)

There has been significant progress in the past several years toward establishing a common understanding and appreciation of the critical features and severity of the African debt problem. It is now widely accepted that the crux of sub-Saharan Africa’s debt problem is the excessive debt overhang, which has led many countries in the region to be classified essentially as insolvent. In contrast to the past when liquidity concerns were dominant, debt-stock-related solvency problems are now more pronounced. Debt overhang refers to the existence of a large debt that has adverse consequences for investment and growth because investors expect that current and future taxes will be increased to effect the transfer of resources abroad. This definition brings to bear three important concerns: impact on fiscal adjustment, current and future resources to enhance economic growth, and current and future resources and resource flows to enhance private and public investments.

The debt situation in sub-Saharan Africa has acquired significant proportions and attention. Many of the world’s poorest countries are to be found in sub-Saharan Africa. Of the 32 developing countries currently classified as heavily indebted poor countries, 26 are in sub-Saharan Africa. It is the only region where on average the stock of external debt now exceeds GNP. Debt servicing in 1994 on average took up 18.6 percent of total exports. Furthermore, debt service in these countries on average consumes more than 21 percent of government revenue collected each year. For the 26 HIPCs in the region, the situation is much more grave, with the debt-stock ratios standing typically at more than a third above the averages of sub-Saharan Africa and nearly threefold above developing countries’ averages.

The debt distress of sub-Saharan Africa is indicated more directly by rising arrears of servicing obligations. In spite of positive net inflows and revolving relief schemes (including refinancing), sub-Saharan Africa in the past six years on average could only service a third of its obligations (Mistry, 1996). Consequently, the growth momentum of the debt stock has been dominated by accrual of arrears and this is the essence of the problem of debt overhang.

What is perhaps more worrisome is the fact that the composition of the outstanding debt is rapidly changing toward preferred creditors, as are the debt-servicing requirements. These creditors have much less scope for relief arrangements largely for fear of deteriorations in their credit rating. Although bilateral and non-OECD debt still accounts for more than half of the total African debt stock, the prominence of multilateral debt has quickly gained ground over the last decade. The share of multilateral debt in the total stock for Africa has risen from 19.5 percent in 1980 to 26 percent in 1994. The share of actual debt servicing of this debt is much higher, at a ratio of 2 to 1, relative to its share in outstanding debt. Currently, multilateral debt-service burden is the largest across different types of creditors accounting for 46 percent compared with 22 percent for bilateral and 38 percent for private creditors (Mistry, 1996). The lower share of bilateral debt service in spite of their larger share of the debt stock is partly explained by relief schemes and higher tolerance toward arrears. Multilateral debt relief has been much more limited, largely through refinancing, and has had a very marginal impact on the magnitude of debt overhang in the region.

The consequences of the debt overhang problem for the region’s future are indeed very grave, and more so for HIPCs in the region. The expenditure-crowding-out effects of servicing the rapidly growing stock of debt have empirically been shown in the context of cross-country growth regressions. Rising debt-service ratios imply reduced availability of resources to support renewal of growth. And because growth cannot be restored the solvency problem deepens in a vicious circle. Indeed, rising debt-servicing requirements along with stagnant exports has meant either defaulting on payment or parting with scarce foreign exchange badly needed for imports required for production and investment. Real outward resource transfers were effected when servicing requirements were met in the context of rapidly depreciating local currencies.

There is sufficient ground to speculate that the internal transfer constraint to servicing debt is probably more severe than the external transfer constraints. This has been exacerbated by the steep depreciation of local currencies, which raised substantially the local costs of external debt servicing. In a number of adjusting countries, foreign reserve net positions have improved. In contrast, fiscal distress has worsened, manifesting itself in the forms of severely compressed development budgets and a shrinking fiscal base for providing essential public services. This happens to be the case particularly in countries that have registered significant success in reducing deficits. While the average primary deficit as a proportion of GDP declined from a peak of 4.4 percent in 1988 to about 2 percent in 1993, debt-service payments of the predominantly public or publicly guaranteed debt registered a steep rise over the same period from 3.9 percent of GDP in 1986 to 5.9 percent in 1993. Capital expenditure shares in the government budget declined from 3.2 to 2.3 percent. In real terms, development expenditures have registered a steep decline. Restoration of a functional state (necessary for supporting the revival of growth) is severely constrained by a shrinking resource base for financing public services. The scope for restructuring expenditures is quickly being exhausted. A rising proportion of improved revenue collected is being channeled to servicing debt.

More recently, it has become evident that a greater concern regarding debt overhang is its negative consequences on sustaining a virtuous circle of reforms and adjustment—particularly in relation to reestablishing fiscal sustainability and renewing growth in the short run. The uncertainty that debt overhang induces can undermine the effectiveness and hence sustainability of an otherwise credible reform program. Furthermore, debt-service burdens are viewed by potential investors as a threat to sustaining reforms and as a potential cause of a higher inflation tax to meet debt-service requirements (Ndulu, 1995; Elbadawi, 1996). In this situation a coordination problem has emerged, due to which the response by the would-be investors has been to exercise their option of waiting until the front-loading of investment returns is sufficient to compensate them for the risk of investing or repatriating capital (Dornbusch, 1990; Serven, 1996), Available evidence shows that response of investment to reform measures has been slow (Soyibo, 1996; Elbadawi and Ndulu, 1995). Where such response has occurred, it has tended to be dominated by short-term investment in trading activities with quick returns rather than long-term, high-risk, irreversible investments in production (Serven, 1996). Furthermore, anecdotal evidence suggests that where flight capital returned, it has tended to be held in liquid assets such as treasury bills and foreign-currency-denominated deposits in domestic banks rather than in irreversible capital assets.

The above suggests that stabilization by itself may not be enough to trigger the “good equilibrium” that is consistent with a virtuous cycle—from stabilization to growth. There is therefore a need for an external mechanism to stem the coordination failure and break the tendency of the market to wait. For sub-Saharan Africa, this external mechanism should be in the form of substantial debt relief. At the same time, to avoid a relapse to the current grave situation, sustained credible reforms are a necessary complementary measure to restore solvency via growth—that is, to restore confidence in the economies of sub-Saharan Africa and in turn induce the required investment and efficiency for growth.

With the above background in mind, this paper focuses on assessing the impacts of external debt overhang on growth and investment. The paper uses a simple simulation model that establishes the levels of sustainable debt ratios consistent with growth revival to levels commensurate with the requirements for reversing the deep economic crisis in sub-Saharan Africa. The overarching theme is that restoration of solvency for HIPCs rests squarely on raising growth by reducing the debt burden (along with other factors). The first section explains the relationship between external debt and growth. It singles out three channels of transmission from debt to growth: the effect of debt overhang on investment; liquidity constraints related to debt servicing; and an indirect channel via the effects on public sector expenditures and deficits. Empirical estimations of these relationships are then applied to a simple simulation model of external debt sustainability in the context of growth. The purpose is to establish—relative to a broader development perspective—sustainable levels of debt-stock ratios and debt-service ratios relative to exports and government revenue collection. The chapter concludes by drawing out implications of the foregoing for debt relief schemes and restoration of solvency on a sustainable basis.

Impact of External Debt on Growth and Investment

It is now more widely understood that the crux of sub-Saharan Africa’s debt problem is excessive debt overhang, which has led many countries in the region to be classified as insolvent. The severity of the debt crises has impacted negatively on growth in per capita incomes and private investment rates in the region.

This in turn has prevented fiscal adjustment from taking effect, thus threatening credible reform programs and future growth prospects. Ample evidence from behavioral empirical models suggests that debt overhang has had a significant and deleterious impact on growth and investments.

The empirical analysis in this section brings to bear this proposition in an attempt to explain the effects of external debt on the growth of per capita incomes and private investment rates. The analysis is carried out for a cross section of developing countries.

The magnitude of indebtedness is shown by the indicators in Tables 1 and 2. From these figures, the ratio of debt stock to GNP has risen from 31 percent in 1980 to more than 70 percent in the 1990s. The ratio of debt stock to exports of goods and services has risen to about 270 percent in 1995 from 91 percent in 1980. The indicators in Table 1 should be compared with those in Table 2 for the severely indebted low-income countries. For severely indebted countries, the increases in indebtedness have been spectacular (Table 2). For example, debt stock to GNP rose to 145 percent in 1994 from 31 percent in 1980.

Table 1.Indicators of Indebtedness in Sub-Saharan Africa
EDT/GNPEDT/XINT/XTDS/X
198030.690.96.29.7
198867.2242.910.020.7
198969.1237.79.517.9
199070.8225.79.117.8
199170.6239.48.916.4
199269.8235.67.815.7
199373.2251.96.314.9
199478.7265.76.914.0
199574.1269.87.214.7
Source: World Bank (1996), where EDT/GNP is the ratio of debt stock to GNP, EDT/X is the ratio of debt stock to exports of goods and services, INT/X is the ratio of total interest payments to exports, and TDS/X is the ratio of total debt-service-payment obligations to exports of goods and services.
Source: World Bank (1996), where EDT/GNP is the ratio of debt stock to GNP, EDT/X is the ratio of debt stock to exports of goods and services, INT/X is the ratio of total interest payments to exports, and TDS/X is the ratio of total debt-service-payment obligations to exports of goods and services.
Table 2.Severely Indebted Low-Income Countries
EDT/GNPEDT/XINT/XTDS/X
198031.4106.36.011.2
1988104.1489.013.929.1
1989127.5515.411.924.9
1990140.5457.111.523.0
1991136.1498.612.223.0
1992137.9498.810.222.2
1993139.9530.08.317.4
1994145.2529.49.420.0
1995128.4487.97.921.0

In 1995, increases in these indicators were somewhat checked, except for total debt-service obligations, which declined.

Empirical Implementation

The analysis focuses on cross-section regressions for 99 developing countries spanning sub-Saharan Africa, Latin America, Asia, and the Middle East. These cross-section regressions offer a uniform statistical assessment of growth and private investment rates in this wide array of countries. The growth equation is estimated to capture the channels in which indebtedness is working against growth in per capita incomes. These channels are both direct and indirect. There are three such channels in which the indebtedness in sub-Saharan Africa is working against growth, the direct channel works through current debt inflows as a ratio of GDP, which stimulate growth, while past debt accumulation (debt overhang) impacts negatively on growth. These two channels produce a debt Laffer curve, which shows that there is a limit at which debt accumulation stimulates growth, in line with resource gap models. When this limit is reached, further debt accumulation impacts negatively on growth.1 The debt Laffer curve conventionally used usually refers to the relationship between the amount of debt repayment and the size of the debt. However, in this paper, the Laffer curve is used to mean the possible negative effect of debt on growth when the level of indebtedness is very high. Thus we have growth inducement effects at low levels and growth is retarded at high levels of indebtedness.

These results confirm the overborrowing proposition postulated by Greene and Khan (1990), The third direct channel works through a liquidity constraint, where debt-service-payment obligations reduce export earnings and thus impact negatively on growth. The final channel is an indirect one and works through the impact of the above channels on public sector expenditures, which impacts negatively on growth. Besides these variables, we include in the regression model other policy, fundamental and shock variables.

The original formulation of the debt overhang hypothesis centered on the adverse effects of debt on investment in physical capital. How debt overhang might affect and discourage private investments depends on how the respective government is expected to raise fiscal revenue necessary to finance external debt-service obligations (an inflation tax and excessive government expenditure will contribute to increased domestic inflation that also discourages private investment). The other channels that compound the problem are crowding-out effects, lack of access to international financial markets due to solvency indications, and the effects of the stock of debt on the general level of uncertainty in the economy. These effects combine to discourage private investment and thus have a negative impact on national output growth. This discussion suggests that we can estimate a cross-section regression for the period 1960–94 of the form:

The variables are defined and motivated as follows: GDPCAP is per capita GDP growth, the dependent variable. EDTGDP is the ratio of the current stock of debt to GDP, which should be positive to confirm the first channel. EDTGDPL2 reflects debt overhang, past debt accumulation, which is squared to capture the highly indebted countries, that is, the intensity of indebtedness in relation to growth. The coefficient should be negative. A high debt-stock accumulation reduces current financial flows for investment due to repayment and service obligations and thus cannot stimulate growth of domestic incomes.

Debt service as a ratio of export earnings is reflected by DSX, whose coefficient should be negative to confirm and reinforce the liquidity constraint channel. DEFGDP and DEFGDPL are the ratios of fiscal deficit to GDP, current and lagged, respectively, which again should have negative coefficients. PUINV is public investment as a ratio of GDP. The coefficient should be positive to confirm the positive behavioral relationship between investment rates and income. The rate of inflation, INFL, should stimulate growth at low and containable levels but should impact negatively on growth at high and crisis levels. To capture the impact of high and crisis levels of inflation, a dummy variable, depicting inflation crisis, is added: INFL40, where it takes a value of 1 if inflation exceeds 40 percent for two or more consecutive years. Furthermore, the rate of inflation reflects macroeconomic sustainability and stability and thus may impact negatively or positively on growth depending on whether crisis levels have been reached. Thus this may be considered a policy variable.

The other policy variable included in the model is real exchange rate misalignment, RERMIS (calculated from purchasing power parity relations and misalignment, defined as the percentage deviation from the mean). Misalignment may create distortions in the economies and thus impact negatively on growth. External shocks are reflected by terms of trade variability, CVTOT, and the presence of internal dissent, REVOLS, reflects internal shocks. The coefficients should show a negative impact on growth. Population growth, RPOP, reflects the impact of population pressure on domestic resources; its coefficient is ambiguous. LSCHOOL reflects human capital development, which should impact positively on growth. Finally, we include a variable that takes care of initial incomes, LKGDP, which should capture the convergence effects (see Easterly and Levine, 1994), even though in our results it shows divergence, though not significant. The data are annual for the period 1960–94.

Growth Equation: Empirical Results

The results presented here confirm the three channels working in the growth equation. We start with a simplified model to show the existence of a kind of Laffer curve between growth and debt overhang, add to the model the effects of public sector deficit and the effect of debt-service obligations on growth, and finally a general model of growth that takes into account the variables postulated above. The results are shown in Tables 3, 4, and 5. The fixed and random effects models are shown. The fixed effects model assumes that the countries in the sample have a common slope, but that each has its own intercept that may be correlated with the regressors.

Table 3.Growth and Debt Overhang: Panel Estimates
Fixed Effects ModelRandom Effects Model
VariableCoefficientt-ratioCoefficientt-ratio
EDTGDP29.7038.7529.0039.20
EDTGDPL2–1.50–6.98–1.54–6.58
Constant–10.34–7.59
R2 = 0.74R2 = 0.73
s.e. = 12.96s.e. = 13.15
Note: Hausman test X2(1) = 0.027 [0.8802].
Note: Hausman test X2(1) = 0.027 [0.8802].
Table 4.Growth, Debt, and Fiscal Deficits: Panel Estimates
Fixed Effects ModelRandom Effects Model
VariableCoefficientt-ratioCoefficientt-ratio
EDTGDP23.0926.2123.5329.72
EDTGDPL2–2.89–7.60–2.45–8.18
DSX–0.129–4.25–0.145–7.94
DEFGDP–0.190–1.93–0.157–3.71
DEFGDPL–0.0024–3.34–0.0027–8.69
TIME0.2456.27–0.0019–1.43
Constant–18.05–5.33
R2 = 0.76R2 = 0.75
s.e. = 9.51s.e. = 9.67
Note: Hausman test X2 (4) = 56.24 [0.00].
Note: Hausman test X2 (4) = 56.24 [0.00].
Table 5.Growth Equation: Panel Estimates
Fixed Effects ModelRandom Effects Model
VariableCoefficientt-ratioCoefficientt-ratio
EDTGDP5.484.165.387.44
EDTGDPL2–2.97–13.61–2.77–16.31
DEFGDP–0.307–6.76–0.312–14.43
DEFGDPL–0.0010–1.45–0.0010–3.66
PUINV0.25410.040.23427.29
DSX–0.046–2.93–0.057–5.41
INFL–3.192–1.52–3.76–4.26
INFL40–1.406–1.533–1.22–1.510
CVTOT–0.029–5.023–0.019–1.56
RPOP0.00030.73–0.0011–1.20
LRGDP0.00380.8260.0061.29
RERMIS–0.0152–3.65–0.012–2.32
LSCHOOL0.1352.550.01572.44
FINDEP2.421.342.9573.711
LIFE0.00221.130.0010.545
WARCIV–0.0318–2.65–0.004–0.290
TIME0.0291.23–0.004–.3.630
SSA–0.003–0.6400.00130.049
REVOLS0.012.250.0142.68
Constant–23.99–20.98
R2 =0 .95R2 = 0.96
s.e.= 8.65s.e. = 8.52
Note: Hausman test X2 (9) = 76.496 [0.000].
Note: Hausman test X2 (9) = 76.496 [0.000].

The random effects model, however, has the advantage of being able to estimate variables that are constant over time, and hence no information is lost. That is, the model uses all the information on all the individual units and all the variables, even those that do not vary over time. A Hausman specification test is used to compare these two models. A large value of the test is in favor of the fixed effects model; the null and low probability values (in brackets) signal the rejection of the null.

EDTGDP is the ratio of debt stock to GDP, and EDTGDPL2 is the debt stock lagged one period and squared, as defined above. The Hausman specification test shows a high probability value, and tests the random effects model against the fixed effects model. In this case the results are in favor of the fixed effects model. The interesting result of this regression model is that there is a Laffer curve reflecting the debt overhang problem. From this and subsequent estimations, we can solve for a growth-maximizing level of the stock of external debt. Table 4 adds more information to this model. Where the same relationship is somewhat maintained, however, the coefficient values change drastically. We do not anticipate the coefficients to be invariant to changes in the information set, but drastic changes are suggestive of data problems and instabilities in the regression equations. The results should thus be interpreted with caution. In this table, we add the trend term, TIME, to capture the trend effects in the model. When more information is added to the model, the Hausman test moves in favor of the random effects model. The results, even though tentative, are again clear cut. Debt-service obligations and public sector deficits retard growth, confirming the earlier channels outlined.

These results thus tentatively confirm and reinforce the previous preoccupation with resolving liquidity problems linked to servicing the existing external debt stock. The problem is compounded by the effects of debt overhang on fiscal deficits. This further illustrates the precariousness of fiscal adjustment in the highly indebted countries. Thus the crowding out effects of development expenditures and the uncertainty-inducing effects of debt overhang may undermine the effectiveness and credibility of reform programs and investment. This may perhaps explain both the poor growth performance of developing countries and, more so, of the highly indebted countries starting in the 1980s, and also their difficulties with the reform process. Ironically, even with these negative impacts from debt overhang, reform policies and structural adjustment programs have added to the already highly indebted countries’ stock of debt in the 1980s and 1990s. This places in jeopardy both the credibility of the reform process and future growth prospects.

From these results, we see that debt stock spurs growth, debt accumulation deters growth, and current and lagged levels of fiscal deficit impact negatively on growth. The trend is not significant in the random effects model, while the other variables in the model increase their level of significance in the random effects model. This confirms why the specification test favors the random effects model.

These results are consistent with Cohen (1993), who argues, using an investment equation, that there is a “debt Laffer curve problem.” It illustrates that the Laffer curve problem is present in the growth equation as well, but via the investment channel. In addition, since investments are supposed to induce positive effects on growth in behavioral models, this Laffer curve problem is further reinforced and explains the slow and negative growth of per capita incomes since the debt crises in the 1980s. The results further illustrate that the effects on growth are beyond those provided via the investment channel. In a sense, growth allows accumulation via the accelerator channel and the multiplier process. If growth is retarded, then accumulation is also affected, and the multiplier process works in reverse. This could be a plausible explanation of this channel. This also explains the results in Table 5, where debt variables appear together with investment in the same equation, compounding the channels.

In line with these results, policy and other variables are added to the model, as the above discussion suggested. The results of the general growth equation are shown in Table 5. The general model now includes policy variables for inflation (INFL) and for inflation persistently above 40 percent and for real exchange rate misalignment (INFL40—a dummy). Inflation shows that it is negatively related to growth, and this is an indirect effect of macroeconomic instability. A high and persistent inflation dummy also depicts the same results but is statistically not highly significant. Misalignment in the real exchange rate retards growth. We have included public investment (PUINV), which should contribute positively to growth. External shocks will have negative effects on growth, which is reflected by a coefficient of variation in the terms of trade (CVTOT) that shows it is negatively related to growth.

Internal shocks, reflected by a dummy for revolutions (REVOLS), surprisingly show a positive effect on growth, even though we would anticipate the shocks to retard growth since they reflect political instability. Population growth puts pressure on the available resources, and it should be anticipated to be negatively related to growth. In our results, however, it is negatively related to per capita growth in the random effects model, but the results are not significant. Thus population has no effect, but the sign switches to negative from the fixed to the random effects model. Equally, LIFE, which is a measure of life expectancy, has no effect on growth. Finally, we have a variable that reflects the role of human capital development on growth, LSCHOOL, and that is positively related to growth. The dummy for sub-Saharan African countries is not significant and a trend term is significant in the random effects model. The other variable that reflects the impact of internal dissent on growth is a variable for civil wars, WARCIV. It should be negatively related to growth, but surprisingly it becomes insignificant in the random effects model. Perhaps its impact is taken over by the presence of a constant, but again this shows the precariousness of the data available.

The results are in favor of a random effects model, and it achieves a high coefficient of determination, which reflects the fact that variables in the model help to explain the factors behind growth of per capita incomes in sub-Saharan Africa (even though some of the variables have the wrong expected sign while others are not significant). These results thus confirm the effect on growth emanating from indebtedness, fiscal effects, policy variables, and internal and external shocks. These results compare well with those of Easterly and Levine (1994), even though they did not include the external debt effects. We thus come to the conclusion that debt overhang has retarded growth in sub-Saharan African countries, and these results can be used to derive the sustainable levels of debt accumulation, beyond which further debt accumulation starts to affect growth.

The Laffer curve is shown in Figure 1, where gy is growth in per capita incomes and d is the ratio of debt stock to GDP. The turning point shows the limit at which debt accumulation starts to have negative effects on growth.2

Even though the above results have been presented and discussed, a few caveats are in order. First, the coefficients are not robust and changes in the specification produce drastic changes in the coefficient values. It would not be anticipated that if the information set is increased the coefficient will remain invariant, but substantial changes are subject to instability. Second, the data used are suspect; there are various gaps in the data set and thus the model is again subject to instability. Finally, some of the variables that could have been included turned up with wrong signs and supersignificant t-ratios and disturbed the whole regression equation. They were thus subsequently dropped from the analysis (a pitfall of empircal research). The results should be treated as an average outcome that uncovers the effects of accumulated external debt on growth.

A Private Investment Model

The effects of debt on investment have been heavily researched. This is in line with the resource gap literature. The preoccupation before the debt crisis was that debt flows would close resource gaps and thus stimulate investments and economic growth. Since the debt crisis, the converse has been happening, due to outflows of resources to meet debt repayments and service obligations. This has come to be referred to as the debt overhang problem; see Cohen (1993). Some researchers (e.g., Mukhopedhyay, 1995), have argued that most estimations of the private investment model do not take care to specify whether private investments are demand determined or credit constrained. In this paper we estimate a demand-determined private investment model. It should, however, be noted that there is an indirect channel to the credit constraint since increases in debt stock, debt-service obligations, and fiscal deficits have reduced both the demand for private investments and the supply of credit in developing countries. This is the debt over-hang explanation and is thus a demand-side determinant of private investment rates (thus the debt-stock-to-GDP ratio enters the model, lagged one step). The dependent variable is the ratio of private investment to GDP (IPY). The variables are as earlier defined; they include real exchange rate misalignment, RERMIS; this variable reflects the effects of exchange rate policy and exchange rate management on the private investment rates. The rate of inflation, INFL, reflects uncertainty and macroeconomic instability. High inflation rates are an indicator of macroeconomic instability and thus make the environment for investors uncertain (Mukhopedhyay, 1995), Fiscal policy, DEFGDP, and public investment, PUINV, are included to capture the effect of crowding out private sector investment, while public investment supplements private investment. A statement of the investment equation follows:

Figure 1.Debt and Growth Relationship

Debt variables are included just like in the growth equation. Growth in per capita incomes should be positively related to private investment rates; inflation should be positive, but high inflation should deter investment. Population growth is ambiguous; it may be positive or negative. In the results shown here, the coefficient is positive and thus shows that an expanding population offers new investment opportunities (Cohen, 1993). In addition, we have a trend term and a sub-Saharan African regional dummy. The current debt stock should have a positive coefficient showing that current debt flows should stimulate private investments while past debt accumulation should be negatively related to private investments. This is the negative impact of debt overhang on private investments; together with the debt-service ratio, DSX, the effect of indebtedness is to reduce private investments. We again start from a simple model where private investment is explained by per capita growth and the ratio of debt stock to GDP and the regressors increase sequentially. The motive is first to establish the explanatory power of growth and debt overhang on private investment rates. The results are shown in Tables 6 and 7.

Table 6.Private Investment Model: Panel Estimation Results
Fixed Effects ModelRandom Effects Model
VariableCoefficientt-ratioCoefficientt-ratio
GDPCAP0.00581.640.5841.135
EDTCDP0.0414.920.0415.94
EDTGDPL2–0.016–4.56–0.016–4.51
Constant–26.46–5.98
R2 = 0.940R2 = 0.936
s.e. = 9.302s.e. = 9.44
Note: Hausman test X2(1) = 0.0124 [0.9112].
Note: Hausman test X2(1) = 0.0124 [0.9112].
Table 7.Private Investment Model
Variable CoefficientsRandom Effects Model
VariableCoefficientt-ratioCoefficientt-ratio
GDPCAP0.00361.110.00441.19
EDTGDPL0.9658.230.96729.18
EDTGDPL20.0334.76–0.022–5.70
DEFGDPL0.00281.72–0.0059–1.57
PUINV0.97319.480.98935.96
INFL0.97015.220.97530.16
INFL40–1.09–1.75–0.329–0.552
TOTSHK–0.024–4.72–0.023–6.57
TIME0.0634.640.00292.002
SSA0.4353.020.00590.412
Constant0.4518.85
R2 = 0.91R2 = 0.90
s.e. = 7.78s.e. = 7.90
Note: Hausman test X2(1) = 3.8257 [0.0505].
Note: Hausman test X2(1) = 3.8257 [0.0505].

The ratio of debt stock to GDP is positively significant, and the debt overhang variable is negatively significant. Lagged GDP per capita growth is not a powerful explanatory variable for private investment. In Table 7, more information is added.

The results show that DEFGDPL, public sector deficit to GDP lagged one step, dampens private investment, but not significantly also changes sign from the fixed effects model and becomes negative in the random effects model. The ratio of public investment to GDP, PUINV, is positively related to private investment.

The rate of inflation (INFL) is positively related to IPY and the inflation dummy (INFL40), which reflects uncertainty and inflation persistently above 40 percent, is negative but not significant. Inflation and high persistent inflation are supposed to reflect macroeconomic instability. Surprisingly, inflation is positively related to IPY, perhaps reflecting the fact that it has not reached crisis levels to discourage investments. The crisis level of inflation should be captured by the dummy INFL40, and is significant in the fixed effects model but not in the random effects model. External shocks are reflected by terms of trade shocks (TOTSHK) and have a negative impact. We have added a trend term to control for the trend in influence in the model, while the sub-Saharan African dummy is positive and significant in the fixed effects model and not significant in the random effects model. Perhaps the constant makes all the difference and makes the random effects model powerful over the fixed effects model. We now add initial income level (LRGDP), population growth (RPOP), real exchange rate misalignment (RERMIS), debt-service ratio (DSX), and internal dissent variables, REVOLS, all of which affect private investment (except REVOLS, which is not significant).

The results of the estimation are shown in Table 8. In particular, debt-service obligations reduce available credit and thus create a disincentive for investors. Furthermore, it reduces the importing capacity, which is essential for resource-poor developing countries. In addition, the policy outturn is affected by debt accumulation, which again impacts on the real exchange rate, fiscal deficits, and the rate of inflation. All these factors combine to work against private investment growth. Table 8 shows the enlarged investment model, which is in line with results obtained by other researchers. But one surprising result is that RERMIS positively impacts on investment.

Table 8.Private Investment Model: Panel Estimation Results
Fixed Effects ModelRandom Effects Model
VariableCoefficientt–valueCoefficientt- value
GDPCAP0.00411.280.00561.51
EDTCDP0.9632.860.96229.07
EDTGDPL2–0.030–4.73–0.021–5.56
DEFGDPL–0.0024–1.580.0051–1.36
PUINV0.96715.940.98433.22
INFL0.97039.020.96829.95
INFL40–1.002–1.63–0.193–0.322
TOTSHK–0.0230–4.71–0.023–6.11
RPOP0.2471.240.06010.762
LRGDP–0.0021–0.494–0.0019–0.58
RERMIS0.01563.650.0122.98
SSA6.0040.3260.00530.36
DSX–0.00125–3.69–0.011–2.49
REVOLS–0.0044–0.7550.0019–0.47
TIME0.0544.490.0050.24
Constant11.903.34
R2 = 0.95R2 = 0.95
s.e. = 7.76s.e. = 7.88
Note: Hausman test X2(5) = 4.321 [0.3003].
Note: Hausman test X2(5) = 4.321 [0.3003].

The estimation results confirm that debt flows stimulate growth, and that past debt accumulation (debt overhang) impacts negatively on growth and private investment rates. Debt-service obligations reduce export earnings and thus impact negatively on the growth of per capita income and private investment rates. These results tentatively confirm the channels through which debt overhang deters growth and investment in developing countries, placing in jeopardy reform programs, which are undertaken with the help of massive resource flows that increase future debt payment obligations. The results, subject to the caveats on the difficulties of using the data, show that debt overhang works indirectly to affect other policy variables and reduces the economies’ flexibility in absorbing or adjusting to internal and external shocks. Thus, unless the problem is addressed and solutions are sought, future growth prospects and investments do not look promising in the face of this high indebtedness and the dependence of the reform process on aid flows.

A Simple Model of External Debt Sustainability

The HIPC Debt Initiative gives fairly specific criteria for assessing external debt sustainability, on the basis of the following three guidelines:

(1) The ratio of the net present value of debt to exports (the NPV debt-exports ratio) should be expected to fall within a range of 200–250 percent, or below, by the completion point;

(2) The ratio of debt service to exports should be expected to fall within a range of 20–25 percent, or below, by the completion point;

(3) Within these prescribed ranges, debt sustainability would be determined in conjunction with various measures of vulnerability, including the burden of external debt service on the government budget, the diversity of the country’s export base, its reserve coverage, its resource balance, and any other relevant factors.

The HIPC Initiative is an important step toward creating a level playing field for emerging reforming African countries. However, given the current extent of the tragedy of African development, the adequacy of this Initiative, in our view, merits more careful scrutiny. The tragedy of Africa’s development during the past two decades is very vividly reflected in its disappointing growth performance and its great and deepening poverty.3 Sub-Saharan Africa has been the only developing region of the world to have registered negative per capita growth rates (on average) over the last two decades. Furthermore, estimates of the incidence of poverty in the region range from 66 percent (Ali, 1995) downward. According to the World Bank, to deal with such unusually alarming levels of poverty, even the most successful performers among the African countries may have to considerably enhance their growth performance to about 5 percent per capita growth rates per year.

To test the adequacy of the external debt sustainability criteria of the multilateral debt initiative from the broader African development perspective, we built a simple aggregative model and used it to generate sustainable debt indicators consistent with the development target of 5 percent real per capita GDP growth rate.

Structure of the Model

The model is built around the two behavioral equations (per capita output growth and private investment) discussed in the previous section, and an identity for fiscal policy consistency. The two behavioral equations emphasize the links of growth and private investment to the external debt situation (debt burden and debt-service ratio), fiscal balance, and public investment. In the case of the growth equation, a debt-growth Laffer curve is obtained as well, the policy implications of which were analyzed in the previous section. Aggregate fiscal deficit is made up of the noninterest deficit (revenue minus government expenditure) plus interest payment on domestic and foreign debt. The fiscal policy identity shows explicitly the sources of the overall fiscal deficit.4 The other component of the model is a simple behavioral relationship between public investment and the ratio of external debt service to public sector revenue. The remaining equations are definitional identities of debt indicators. The complete set of variables for the model’s equation is contained in Appendix Table 1.

As the above brief description suggests, our model is very aggregative: with one aggregate output on the supply side, two agents (public and private) on the investment-demand side, and one consistency equation for fiscal policy. Even though the model is conceived as a two-gap model (public sector and saving-investment/balance of payments gaps), only the public sector gap is explicitly modeled. In addition, the model also does not include a goods market equation for endogenizing relative prices. Both exclusions entail some limitations on the part of the model. The absence of a balance of payments equation implies that the private sector constraint is not accounted for, while the exclusion of the goods market equation means that relative price effects are not captured. However, given the overwhelming dominance of public sector constraints in the context of the external debt problems for sub-Saharan African HIPCs, this simple model provides a useful benchmark for focusing attention on the main issues.

Model Closure

The model closure (selection of exogenous and endogenous variables) is mainly driven by the structure of the model and the issues of interest for the analysis. The model closure adopted for solving the model follows the “normative” mode (see Appendix), where a target for growth is fixed (5 percent per capita growth) and the model is solved for debt and fiscal deficit indicators consistent with this target growth rate as well as other assumptions about relative prices (real interest rates, real exchange rate), inflation, and terms of trade shocks.

The model solution can be summarized by two key relationships linking both sustainable external debt-GDP and deficit-GDP ratios to output growth and the ratio of public investment to GDP. If we make the logical assumption that HIPCs fall on the “bad” part of the Laffer curve, sustainable debt is negatively associated with both variables, while sustainable deficit is positively linked to the two variables. Appendix Figure 1 summarizes the simple comparative statistics for the effect of a higher growth target (or public investment ratio) on sustainable external debt and public sector deficits.

Debt Situations and Sustainability

An analysis of debt indicators, economic policy stance, and economic performance for the sub-Saharan African HIPCs reveals the depth of the crises faced by these countries, even by African standards. Table 9 contains data on indicators of external debt, domestic debt, and other relevant macroeconomic variables for sub-Saharan Africa in the 1990s. In addition to sub-Saharan Africa, the table also contains median and average indicators for African HIPCs and non-HIPCs, as well as for three HIPC groupings used by the IMF and the World Bank, Martin and Mistry (1996), and Mistry (1996). While the median annual per capita growth rate for African non-HIPCs (in 1990–94) was disappointing at -0.23 percent, it was even more alarming for the HIPCs at -1.8 percent. Naturally the external debt burden (ratios of stocks of external debt to GDP and to exports, and ratios of debt service to exports and to fiscal revenue) in 1990–94 is much higher in HIPCs compared with non-HIPCs. However, it is surprising that external debt-based HIPCs are also the most indebted in terms of domestic debt. The median stock of domestic debt to GDP is 13.7 percent for HIPCs compared with 10,4 percent for non-HIPCs. The same comparison carries over for other macroeconomic variables, where HIPCs as a group have a higher fiscal deficit ratio and lower public and aggregate investment ratios.

Table 9.Median Debt Ratios and Related Variables, 1990–94(in percent)
Ratio or

Variable
Sub-

Saharan

Africa
Heavily

Indebted

Poor

Countries
Non-

Heavily

Indebted

Poor

Countries
IMF-World BankMartin AnalysisMistry Analysis
UPSUPSUR
GDP growth–1.37–1.77–0.23–1.37–2.94–2.10–0.02–1.53–2.01
External debt to GDP70.3087.5051.90199.8072.10127.7068.60125.9068.60
Domestic debt to GDP11.4013.7010.4013.7022.6011.7022.6013.70
External debt to X441.00648.40303.001749.00665.801023.00412.60953.70411.60
External debt service to X14.3619.579.5135.4419.5728.7511.3029.3611.37
External debt service to R15.2418.0513.0513.7915.692.5210.7015.9220.66
Overall deficit to GDP–3.15–4.44–0.81–5.15–3.94–4.95–3.94–4.95
Public investment to GDP8.466.919.924.754.444.358.634.757.99
Private investment to GDP8.698.479.9816.248.5810.468.639.639.16
Source: IMF and World Bank data.Note: X = exports, U = countries with unsustainable debt burdens, PS = possibly stressed countries, and R = countries that need rescheduling (see text).
Source: IMF and World Bank data.Note: X = exports, U = countries with unsustainable debt burdens, PS = possibly stressed countries, and R = countries that need rescheduling (see text).

The IMF-World Bank and Martin groupings of HIPCs contain further disaggregation of the group into unsustainable multilateral debt burden countries (U) and possibly stressed countries (PS); while Mistry classified HIPCs into U and those that would need rescheduling of residual stock on a soft/long-term basis (R). Comparison of the three methods for assessing indebtedness reveals that the IMF-World Bank criteria are the most conservative among the three. According to the IMF-World Bank criteria, the group of countries classified as U (PS or R) has a median external debt-exports ratio of 1,749 percent (666 percent) in 1990–94, compared with 1,023 percent (413 percent) for Martin and 954 percent (412 percent) for Mistry. Since the IMF-World Bank approach is the operational methodology used for the implementation of the multilateral debt initiative, it is important to evaluate it relative to the Other two approaches from the broader development perspectives of sub-Saharan Africa, as we discussed above.

Table 10 contains five derived indicators of external debt sustainability, based on the macroeconomic model described above: ratios of stock of external debt to exports; external debt service to exports; external debt service to fiscal revenue; aggregate fiscal deficit to GDP; and private investment to GDP. Focusing on the first two indicators, the comparison of sustainable ratios to the thresholds set by the multilateral debt initiative (see above) suggests that the initiative appears adequate as a framework for addressing HIPCs’ external debt problem. However, an analysis of the median debt ratios for HIPCs (U and PS) according to the IMF-World Bank groupings shows unequivocally that the implementation of the initiative is not only conservative relative to other approaches, but also that it threatens to render the multilateral debt initiative to be inadequate for restoring and sustaining growth levels compatible with reversing the economic tragedy currently faced by sub-Saharan African HIPCs. Using configurations of the sustainable debt indicators derived above, groupings of unsustainable sub-Saharan African HIPCs could be generated, where the criteria of classification is whether or not the debt burden is consistent with the 5 percent annual per capita growth rate, deemed to be the requirement for addressing the current African economic crisis (see Table 11). Consistent with the above evidence, the table shows a much larger group of unsustainable HIPCs.

Table 10.Actual and Derived Indicators of External Debt Sustainability for Sub-Saharan Africa(In percent)
Actual
Dependent

Variable
Heavily

indebted

poor

countries
Non-

heavily

indebted

poor

countries
IMF-World BankDerived
UPS
EXDEBTX648.40303.001,749.00665.80330.60
EXDSX19.579.5135.4419.578.80
EXDSR18.0513.0513.7915.6910.00
PRINV8.479.9516.248.5815.00
Deficit/GDP–4.44–0.81–5.15–6.00
Table 11.Criteria for Classification of African Heavily Indebted Poor Countries
Variable
CountryRatio of external

debt to GDP
Ratio of external debt

service to exports
Ratio of external debt

service to revenuea
Benin
Burkina Faso
Burundi*
Cameroon
Cape Verde
Central African Republic
Comoros
Congo
Côte d’lvoire
Equatorial Guinea
Ethiopia
Gabon
Gambia, The
Ghana
Guinea
Guinea-Bissau*
Kenya
Lesotho
Madagascar
Malawi
Mali
Mauritania
Mauritius
Mozambique*
Niger
Nigeria
Rwanda
São Tomé and Princípe*
Senegal
Sierra Leone
Somalia
Sudan*
Tanzania
Uganda
Zaïre*
Zambia*
Zimbabwe
Note:

Model-based classification.

IMF–World Bank classification.

Classification according to the criteria is done only for countries with available data on revenue.

Note:

Model-based classification.

IMF–World Bank classification.

Classification according to the criteria is done only for countries with available data on revenue.

Conclusions

This paper has attempted to assess the adequacy of the multilateral debt initiative for HIPCs from the perspective of the wider development agenda for sub-Saharan Africa, aimed at reversing the current tragedy of economic development there. The evidence of this paper suggests that the debt burden faced by the African HIPCs has very strongly and negatively affected economic growth since the second half of the 1980s; threatened the sustainability of reforms; and prevented the development of a capable and functional state, due to the fiscal crisis that ensued. The multilateral debt initiative remains the most serious and comprehensive effort so far to address this critical type of indebtedness for sub-Saharan Africa. However, simulation results based on the model of this paper suggest that the implementation of this initiative—as reflected by the classification of countries into “unsustainable” HIPCs, “possibly stressed” HIPCs, and non-HIPCs—appears to be very conservative, not only relative to other approaches, but also relative to the broader development perspective for sub-Saharan Africa. Hence, despite the adequacy of the guidelines, the multilateral debt initiative may, in effect, end up being inadequate to propel sub-Saharan Africa to the minimum growth path required to reverse its current economic decline. Moreover, when the model-based criteria are applied for classifying the debt situation of African countries, the set of unsustainable HIPCs is found to be much larger than indicated by the IMF-World Bank classification.

The paper’s empirical analysis of debt, investment, and growth was based on cross-section regressions for 99 developing countries spanning sub-Saharan Africa, Latin America, Asia, and the Middle East. The evidence obtained from these regressions corroborates the literature regarding the conventional direct and indirect channels through which indebtedness impinges on growth and investment. Current debt inflows stimulate growth, while past debt accumulation (debt overhang) impacts negatively on growth. These two channels produce a Laffer curve, which shows that there is a limit at which debt accumulation stimulates growth, in line with resource gap models. When this limit is reached, further debt accumulation impacts negatively on growth. The third direct channel works through a liquidity constraint where debt-service-payment obligations reduce export earnings and thus impact negatively on growth. The final channel is an indirect one and works through the effects of the above channels on public sector expenditures, which impact negatively on growth. These results on the association between debt, investment, and growth are estimated while controlling for other relevant growth and investment fundamentals.

Admittedly, our simulation model is very simple and highly aggregative. Also some of the assumptions are subject to criticism, and needless to say, the model simulation may not be sufficiently robust against different sets of assumptions. Nevertheless, this exercise provides a useful analytical benchmark for assessing this important initiative, relative to the requirements for restoring growth that could make a meaningful impact on poverty and other development problems of sub-Saharan Africa.

Appendix: A Simple Model of Debt Sustainability

Per the discussion above, the model structure stated in Appendix Table 1 includes the basic building block, the model closure, and assumptions.

Model Closure

The model closure (selection of exogenous and endogenous variables) is mainly driven by the structure of the model and the issues of interest for the analysis. There are two types of closure mode available. “Positive” closure treats the public sector aggregate demand component and public sector liability as exogenous and solves the model for all macroeconomic variables such as output growth, prices, real exchange rate, and interest rates. “Normative” closure inverts the model: by fixing exogenous targets for output, interest rates, and relative prices, it solves the model for public sector liability stocks (or flows) and public sector aggregate demand components or revenue items.

The model closure adopted for solving the model follows the “normative” mode, where a target for growth is fixed (5 percent per capita) and the model is solved for debt and fiscal deficit indicators consistent with this target growth rate as well as other assumptions about relative prices (real interest rates, real exchange rate), inflation, and terms of trade shocks.

Some Comparative Statics

Equations 1 and 2 of section B of Appendix Table 1 give the solution for ratios of external debt to GDP and of the deficit to GDP as functions of output growth and the ratio of public investment to GDP. If we make the logical assumption that HIPCs fall on the “bad” part of the Laffer curve, sustainable debt is negatively associated with both variables, while sustainable deficit is positively linked to the two variables. Appendix Figure 1 summarizes the simple comparative static of higher growth target (or public investment ratio) for sustainable external debt and public sector deficits. An increase of the growth target from g0 to g1 requires a reduction of the sustainable external debt ratio from b0 to b1 but this would allow for a higher consistent fiscal deficit, from d0 to d1, and would also require a higher public investment ratio, from (Ig/y)0 to (Ig/y)1

Appendix Table 1.Simple Model of Debt Sustainability
ABasic Model
1.gy =5.38EXDEBT2.97EXDEBT20.313DEF+2.54PUINV0.046EXDSX0.029CVTOT+ADF(1990)
2.PRINV=0.0041gy+0.963EXDEBT0.030EXDEBT20.0024DEF0.023TOSHK0.15EXDSX+ADF(1990)
3.PUINV=γ1TDS*XER+ADE(1990)
4.DEF=(RG) + r×DOMDEBT+r*(EXDEPT*NFA*)e+ADF(1990)
5.EXDSX=TDS*XEX
6EXDSR=TDS*XER
B.Model Closure
1.EXDEBT=f1(gy, PUINV, ...)2.DEF=f2(gy, PUINV, ...)3.PRINV=f3(gy, PUINV, ...)4.EXDSX=f4(gy, PUINV, ...)5.EXDSR=f5(gy, PUINV, ...)
C.Main Assumptions
1.gy= 0.056. e x NFA* = 0.0001
2.PUINV= 0.107.e=E(local/US$)CPI=0.89
3.r= -1.89348. R =0.1033
4.r*= 0.489. G =0.2132
5.DOMDEBT= 0.1141
D.Definitions
gyReal per capita CDP growth
EXDEBTTotal public and publicly guaranteed debt to GDP
DOMDEBTDomestic debt to GDP
DEFOverall deficit to GDP
EXDSXPublic and publicly guaranteed debt service to exports
EXDSXPublic and publicly guaranteed debt service to revenue
PUINVPublic investment to CDP
PRINVPrivate investment to GDP
RGovernment revenue excluding grants to GDP
GGovernment expenditure to GDP
EXDEBT*Public and publicly guaranteed debt to CDP (in dollars)
NFA*Net foreign assets to CDP (in dollars)
ENominal exchange rate (local/dollars)
eReal exchange rate
r*Foreign real interest rate
rDomestic real interest rate

Appendix Figure 1.Some Comparative Statics

References

    AliA.1995“The Challenges of Poverty Alleviation in Sub-Saharan Africa” paper presented at the World Congress of the International Economic AssociationTunisDecember.

    • Search Google Scholar
    • Export Citation

    AnandR. and S.van Wijnbergen1989“Inflation and the Financing of Government Expenditure: An Introductory Analysis with an Application to Turkey,”World Bank Economic ReviewVol. 3No. 1 pp. 1738.

    • Search Google Scholar
    • Export Citation

    AsanteYawforthcoming“Determinants of Private Investment Behaviour in-Ghana”AERC Discussion Paper (Nairobi: African Economic Research Consortium).

    • Search Google Scholar
    • Export Citation

    BuiterW.1983a“Measurement of Public Sector Deficit and Its Implications for Policy Evaluation and Design,”Staff PapersInternational Monetary FundVol. 30No. 2 pp. 30649.

    • Search Google Scholar
    • Export Citation

    BuiterW.1983b“The Theory of Optimum Deficits and Debt,”NBER Working Paper No. 1232 (Cambridge, Massachusetts: National Bureau of Economic Research).

    • Search Google Scholar
    • Export Citation

    BuiterW.1985“A Guide to Public Sector Debt and Deficits,”Economic PolicyVol. 1 pp. 1379.

    CohenD.1988“The Management of the Developing Countries’ Debt: Guidelines and Applications to Brazil,”World Bank Economic ReviewVol. 2 pp. 77103.

    • Search Google Scholar
    • Export Citation

    CohenD.1993“Low Investment and Large LDC Debt in the 1980s,”American Economic ReviewVol. 83No. 3 pp. 43749.

    DornbuschR.1990“Policies to Move from Stabilization to Growth,”in Proceedings of the World Bank Annual Conference on Development Economicsed. by S.FischerD.de Tray and S.Shah (Washington: World Bank).

    • Search Google Scholar
    • Export Citation

    EasterlyW. and RossLevine1994“Africa’s Growth Tragedy” (unpublished; Washington: World Bank).

    EasterlyW. and RossLevine1996“The Tragedy of African Growth,”Policy Research Working Paper 1503 (Washington: World Bank).

    ElbadawiI.1996“Consolidating Macroeconomic Stabilization and Restoring Growth in Sub-Saharan Africa” in Policy Perspectives on African Development Strategiesed. by B.Ndulu and N.van de Walle (Washington: Overseas Development Council).

    • Search Google Scholar
    • Export Citation

    ElbadawiI. and B.Ndulu1995“Growth and Development in Sub-Saharan Africa: Evidence on Key Factors” paper presented at the World Congress of the International Economic AssociationTunisDecember.

    • Search Google Scholar
    • Export Citation

    GreeneJ. E. and M.S. Khan1990The African Debt Crisis AERC Special Paper 3 (Nairobi: Initiative Publishers).

    GuidottiP. E. and M.S. Kumar1991Domestic Public Debt of Externally Indebted Countries IMF Occasional Paper No. 80 (Washington: International Monetary Fund).

    • Search Google Scholar
    • Export Citation

    LairdS. and J.Nogues1989“Trade Policies and the Highly Indebted Counies,”World Bank Economic ReviewVol. 3No. 2 pp. 24161.

    • Search Google Scholar
    • Export Citation

    LipumbaN.H.I. and P.E.M.Noni1990“Foreign Debt Management and economic Growth in Tanzania,”AERC Discussion Paper (Nairobi: African Economic Research Consortium).

    • Search Google Scholar
    • Export Citation

    MartinM. and P.Mistry1996Financing Imports for Development in Low-Income Africa Research Report (Oxford: External Finance for Africa).

    • Search Google Scholar
    • Export Citation

    MistryP.1996Resolving Africa’s Multilateral Debt Problem: A Response to the IMF and the World Bank (The Hague: Forum on Debt and Development).

    • Search Google Scholar
    • Export Citation

    MlamboK. and M.C.Mhlopheforthcoming“Investment Behaviour under Uncertainty: An Analysis of the Determinants of Investment in Zimabwe,”AERC Discussion Paper (Nairobi: African Economic Research Consortium).

    • Search Google Scholar
    • Export Citation

    MukhopedhyayH.1995“Private Investment and External Debt: The Debt Overhang Hypothesis Revisited” Institute of Economic Development (Boston: Boston University).

    • Search Google Scholar
    • Export Citation

    NduluB.1995“Foreign Resource Flows and Financing of Development in Sub-Saharan Africa,”in International Monetary and Financial Issues for the 1990sProceedings of the Conference Sponsored by the Group of Twenty-Four on the Occasion of the Fiftieth Anniversary of the Bretton Woods Conference.

    • Search Google Scholar
    • Export Citation

    RutayisireL.1987“Measurement of Government Budget Deficit and Fiscal Stance in a Less Developed Economy: The Case of Tanzania, 1966–84,”World DevelopmentVol. 15No. 10/11 pp. 133751.

    • Search Google Scholar
    • Export Citation

    SachsJ. and X.Warner1996“Sources of Growth in African Economies” Harvard Institute for International Development (unpublished; Cambridge, Massachusetts: Harvard University).

    • Search Google Scholar
    • Export Citation

    SalihS.1995“Impact of Africa’s Growing Debt on Its Growth,”Research for Action Paper (Helsinki: World Institute for Development and Economic Research).

    • Search Google Scholar
    • Export Citation

    ServenL.1996“Irreversibility, Uncertainty and Private Investment: Analytical Issues and Some Lessons for Africa,”paper presented at the AERC Research WorkshopNairobiMay.

    • Search Google Scholar
    • Export Citation

    SowaN.1994“Fiscal Deficits, Output Growth and Inflation Targets in Ghana,”World DevelopmentVol. 22No 8 pp. 110517.

    SoyiboA.1966“Banking Sector Reforms in Africa: Effects on Saving Investment and Financial Development” paper presented at the AERC Senior Policy Seminar AbidjanNovember.

    • Search Google Scholar
    • Export Citation

    van WijnbergenS.1991“Debt Relief and Economic Growth in Mexico,”World Bank Economic ReviewVol. 5No. 3 pp. 43755.

    World Bank1996African Development Indicators (Washington).

The authors acknowledge the research assistance of Ngina Stanley and Sheila Nyanjui. The authors also acknowledge in-depth, helpful comments made by Mohsin Khan and other participants in the conference that is the subject of this volume.
1These results need to be interpreted with caution, however. The quality of the data and data gaps in most of the countries and in various variables are worrisome. The results are thus tentative, uncovering only general and average trends.
2This growth-maximizing level of the stock of external debt is given by solving the simple maximization of growth relative to EDTGDP on the basis of the estimates in Table 3. The derived growth-maximizing level is equal to the coefficient of EDTGDP divided by 2 multiplied by the coefficient of EDTGDPL2 (5.48 ÷ 2 × 2.97 = 0.97).
3For recent evidence on Africa’s growth performance, see Easterly and Levine (1996), Elbadawi and Ndulu (1995), Salih (1995), and Sachs and Warner (1996). Ali (1995) provides more recent analyses of poverty issues in sub-Saharan Africa.
4The deficit could be financed by three sources: running down reserves, accumulation of further foreign and domestic debt, and printing of money. The analytical framework on the identity between sources of deficits and sources of finance is due to Anand and van Wijnbergen (1989); also see Buiter (1983a, 1983b, 1985). For an application of this analysis to some African countries, see Rutayisire (1987) and Sowa (1994).

    Other Resources Citing This Publication