Chapter

6 An Econometric Analysis of External Debt and Economic Growth in Sub-Saharan African Countries

Editor(s):
Mohsin Khan, and Simeon Ajayi
Published Date:
May 2000
Share
  • ShareShare
Show Summary Details
Author(s)
Milton A. Iyoha

Socioeconomic conditions in African countries deteriorated sharply during the 1980s, a decade that is widely regarded as Africa’s “lost decade” of development opportunities. Available empirical evidence shows that per capita income in sub-Saharan Africa declined at an average annual rate of 2.2 percent; per capita private consumption fell by 14.8 percent; export volume was stagnant while import volume plummeted at an average annual rate of 4.3 percent; and the terms of trade fell by 9.1 percent (see Tables 13 and Figure 1).

Table 1.Sub-Saharan Africa: GNP Per Capita, 1970–94
U.S. Dollars1
1970160
1971180
1972180
1973200
1974250
1975290
1976340
1977380
1978400
1979470
1980540
1981570
1982550
1983500
1984450
1985450
1986390
1987350
1988340
1989340
1990340
1991340
1992321
1993306
1994460
Sources: World Bank (1992a, 1992c, 1996b).

Calculated according to the World Bank Atlas methodology.

Sources: World Bank (1992a, 1992c, 1996b).

Calculated according to the World Bank Atlas methodology.

Table 2.Sub-Saharan Africa: GDP, 1970–94
In Millions of U.S. DollarsGrowth Rate (In percent)
1970138,9138.3
1971150,4427.7
1972162,0272.7
1973165,0002.6
1974170,7277.6
1975183,7031.2
1976185,9087.0
1977198,9214.0
1978206,878–0.2
1979206,4652.4
1980211,4203.2
1981212,4260.7
1982213,9132.7
1983219,6890.05
1984219,7990.2
1985223,7633.4
1986228,8213.6
1987232,5770.9
1988241,7654.4
1989249,4663.1
1990251,9691.1
1991254,2052.5
1992253,8631.8
1993256,1540.9
1994258,4602.2
Sources: World Bank (1992c, 1995).
Sources: World Bank (1992c, 1995).
Table 3.Comparative Economic Indicators(In percent except as noted)
Sub-Saharan AfricaLow Income EconomiesHigh Income Economies
Growth of per capita GNP, 1980–89–2.22.32.5
Change in per capita private consumption, 1980–90–14.831.324.0
Gross domestic saving as a share of GDP, 199016.028.022.0
Gross domestic investment as a share of GDP, 199019.031.022.0
Growth of export volume, 1980–90 (annual)0.25.44.3
Growth of import volume, 1980–90 (annual)–4.32.85.3
Change in terms of trade, 1985–90–9.1–6.53.1
Gross international reserves, 1990 (in months of import coverage)2.33.43.1
Source: International Monetary Fund, World Economic Outlook, various issues.
Source: International Monetary Fund, World Economic Outlook, various issues.

Figure 1.Sub-Saharan Africa: GNP Per Capita, 1970–94

(In U.S. dollars)

Sources: World Bank (1992a, 1992c, 1996b).

Using data on GDP instead of GNP improves the picture marginally, but the situation is still nothing short of catastrophic. An analysis of Tables 4 and 5 confirms that, between 1981 and 1990, the average annual growth rate of real GDP in sub-Saharan Africa was 1.7 percent. However, given the region’s high population growth rate, the average annual growth rate of real GDP per capita between 1981 and 1990 was –0.9 percent. Among all developing regions, sub-Saharan Africa’s performance on the latter measure was the poorest, with the exception of the Middle East and North Africa region. Indeed, sub-Saharan Africa’s negative growth rate contrasts sharply with East Asia’s impressive real per capita GDP growth rate of 6.3 percent and China’s fantastic growth rate of 8.2 percent during the decade.

Table 4.Growth of Real GDP by World Region, 1966–95(Annual averages, in percent)
1966–731974–801981–901991–9319941995
World total5.13.43.21.22.92.8
Industrial countries4.83.03.21.33.02.5
Developing countries16.95.03.34.64.64.9
East Asia7.96.87.68.79.39.2
China8.56.39.912.312.210.2
South Asia3.74.05.73.24.75.5
Sub-Saharan Africa4.73.41.70.62.23.8
Latin America and the Caribbean6.44.81.73.23.90.9
Middle East and North Africa8.54.70.23.40.32.5
Source: International Labor Organization (1996).

Excludes Eastern Europe and the former Soviet Union.

Source: International Labor Organization (1996).

Excludes Eastern Europe and the former Soviet Union.

Table 5.Growth of Real Per Capita GDP by World Region, 1981–95(Annual averages, in percent)
1981–901991–9319941995
World total1.2–0.41.21.4
Industrial countries2.20.72.41.8
Developing countries1.22.62.73.2
East Asia6.37.27.68.0
China8.211.010.59.2
South Asia3.11.12.73.6
Sub-Saharan Africa–0.9–2.3–0.71.1
Latin America and the Caribbean–0.51.32.0–0.7
Middle East and North Africa–3.00.7–2.20.1

Nor did the economic performance of sub-Saharan Africa improve in the early 1990s. Table 5 confirms that, between 1991 and 1993, the region’s growth rate of real per capita GDP was negative, averaging –2.3 percent annually. In 1994, real per capita GDP remained negative at –0.7 percent. Only in 1995 did the growth rate become positive, reaching 1.1 percent. This, of course, still fell well short of the 8.0 percent growth rate posted by East Asia and the 9.2 percent growth rate of real GDP per capita achieved by China.

The bottom line is that the 600 million people inhabiting the roughly 50 countries in sub-Saharan Africa are among the poorest in the world. The World Bank classifies 74 percent of these countries as low-income economies, and the United Nations Development Program classifies 79 percent of them as low-human-development countries. Finally, of the 41 countries in the world classified as Heavily Indebted Poor Countries by the World Bank and the IMF, 33 (80 percent) are in sub-Saharan Africa. Thus, the great majority of countries in sub-Saharan Africa not only are poor (by any standard of evaluation) but also report low or negative rates of growth of per capita income. The conclusion is therefore inescapable that the income gap between sub-Saharan African countries and the rest of the world is continuing to widen.

A close examination of Table 5 shows that the relative position of sub-Saharan African countries started to deteriorate significantly in the 1980s, when the region found itself in a quagmire of economic problems. The acute economic crisis of the last 15 years—which has led to extremely poor growth performance—is attributable to a host of factors, both internal and external. The internal factors are connected with domestic macroeconomic policy weaknesses that have led to inflation, unemployment, rising fiscal deficits, and capital flight. The effects of domestic macroeconomic policy mistakes have been exacerbated by structural weaknesses in the economies, and in particular by a narrow and technologically backward production base and heavy dependence on exports of a narrow range of primary products. The external factors reflect an increasingly hostile international economic environment characterized by low and falling primary commodity prices, declining terms of trade, soaring global interest rates, rising protectionism in the industrial countries, dwindling capital flows into African countries that have resulted in mounting current account and balance of payments deficits, and an escalating external debt stock. The total external debt stock,1 which was $84 billion in 1980, had jumped to $165 billion in 1988 and $190 billion in 1990. By 1995 the external debt stock amounted to $223 billion (World Bank data). Between 1980 and 1995 the debt stock increased by $139 million, for an average annual rate of 6.7 percent. Associated with the rising external debt stock has been a crushing debt-service burden. From a level of $6.4 billion in 1980, total debt service on long-term debt2 rose to $12.3 billion in 1990 and then dipped to $8.8 billion in 1995, largely because of reschedulings and some amount of debt forgiveness. The debt-service ratio (the ratio of actual debt-service payments to exports of goods and services), which was a mere 5.4 percent in 1970, had jumped to 21.3 percent in 1985 before falling to 14.7 percent in 1995 (Table 6).

Table 6.Sub-Saharan Africa: Debt Indicators, 1970–95(In percent except as noted)
EDT/XGSDOD/XGSEDT/GNPDOD/GNPTDS/XGSRES/MGS (In months)RES (In millions of U.S. dollars)
197091.063.220.514.35.42.52,028
1971105.873.524.216.86.22.21,899
1972113.076.220.013.57.02.32,139
1973106.970.022.614.88.82.53,027
197477.650.423.815.54.54.78,167
197593.761.624.316.05.83.47,908
197698.164.724.516.25.83.17,696
197798.664.325.616.74.82.47,475
1978138.492.428.919.37.11.65,407
1979129.885.230.720.27.13.010,215
1980119.778.630.620.17.83.422,249
1981143.4108.428.721.710.31.77,870
1982199.7152.932.024.515.01.45,335
1983259.2197.242.232.120.31.54,713
1984203.3155.747.936.718.31.55,127
1985227.9173.355.141.921.31.76,519
1986223.7183.256.246.022.01.87,075
1987244.1199.764.352.618.41.98,090
1988242.9198.967.255.220.71.510,378
1989237.7195.469.156.817.91.712,129
1990225.7183.970.857.717.82.015,597
1991239.4195.870.657.716.42.318,054
1992235.6188.769.855.915.71.814,280
1993251.9195.273.256.714.91.915,310
1994265.7206.478.761.114.02.120,107
1995269.8210.574.157.814.7
Source: Author’s calculations based on data from World Bank, World Debt Tables, various issues.Note: Indicators are defined as follows: EDT = total external debt stock; XGS = exports of goods and services; DOD = total debt outstanding and disbursed; TDS = total debt service; RES = international reserves; MGS = imports of goods and services.
Source: Author’s calculations based on data from World Bank, World Debt Tables, various issues.Note: Indicators are defined as follows: EDT = total external debt stock; XGS = exports of goods and services; DOD = total debt outstanding and disbursed; TDS = total debt service; RES = international reserves; MGS = imports of goods and services.

One disturbing aspect of the macroeconomic management of sub-Saharan African economies in the 1980s was the failure of these economies to respond favorably to the ministrations of structural adjustment programs. These programs were recommended by the World Bank and the IMF as an antidote to the economic crisis brought on by the debt crisis, among other reasons. Structural adjustment programs were presented as programs that would restore stabilization in the short term and sustainable growth in the medium to long term. A majority of sub-Saharan African countries adopted structural adjustment programs in the 1980s, but unfortunately, in virtually all of them structural adjustment has meant a period of austerity, declining income and living standards, mounting unemployment, and increasing poverty (International Labor Organization, 1996, p. 1). Table 7 compares labor force with population in sub-Saharan Africa. Devaluation, which featured in virtually all structural adjustment programs, led to an increase in total debt and debt-service payments denominated in domestic currency. The deflation required by the structural adjustment programs led to a fall in domestic product and a reduction in national income available for consumption, provision of public services, and investment. A reduction in investment meant a fall in economic growth. Meanwhile, increased inflows of foreign capital expected on the adoption of structural adjustment programs failed to materialize, because of a lethal combination of political instability, poor macroeconomic policies, weak economic performance, and the debt overhang syndrome. In addition, heavy debt-service payments have tended to crowd out foreign investment (Tables 8 and 9). Indeed, aid and foreign investment were routinely diverted into debt-service payments mainly to pay the World Bank and the IMF, because debt owed to these multilateral institutions cannot be rescheduled or written off (Kapijimpanga, 1996). Besides, debt rescheduling by bilateral donors does not constitute true debt relief; rather it leads to a further buildup of the debt stock, as rescheduled debt often attracts higher interest rates, and the interest accumulated is capitalized (Tables 10 and 11). “So, rescheduling has not solved the African debt problem but rather intensified it” (Kapijimpanga, 1996, p. 14).

Table 7.Sub-Saharan Africa: Labor Force and Population, 1970–95(In millions)
Labor ForcePopulation
1970130.0287
1971133.5295
1972137.4303
1973141.2311
1974145.2320
1975149.3329
1976153.7339
1977157.8349
1978162.1359
1979166.5370
1980171.0380
1981175.8391
1982180.7403
1983185.8415
1984191.0427
1985196.4440
1986201.8453
1987207.5467
1988213.4481
1989220.0496
1990226.0511
1991232.0526
1992238.0542
1993244.4558
1994251.0575
1995257.8592
Sources: World Bank (1994b, 1995, 1996b).
Sources: World Bank (1994b, 1995, 1996b).
Table 8.Sub-Saharan Africa: Investment and Saving, 1970–94
Gross Domestic Investment Per Capita (In 1987 U.S. dollars)Investment (In percent of GDP)Saving (In percent of GDP)
19707016.715.5
19718018.815.1
19728019.516.8
19738019.818.4
197410019.621.9
197511023.316.6
197612025.920.5
197713026.019.5
197810023.515.1
19799021.016.7
198010021.919.7
198110022.013.0
19829019.911.5
19838015.810.8
19846012.812.3
19856013.113.1
19867016.712.5
19875015.912.6
19885015.711.5
19895015.913.4
19905016.115.5
19915015.912.7
19926016.014.0
19936015.714.8
199417.0
Sources: World Bank (1992c, 1995, 1996b).
Sources: World Bank (1992c, 1995, 1996b).
Table 9.Sub-Saharan Africa: Capital Stock and Marginal Product of Capital, 1970–93
Capital Stock1 (In billions of U.S. dollars)Marginal Product of Capital
1970457.70.22809
1971446.60.24140
1972441.30.25243
1973439.80.25159
1974439.50.25482
1975440.00.26677
1976449.80.26456
1977464.20.27571
1978481.10.28132
1979493.70.27380
1980500.00.27350
1981508.80.26829
1982517.40.26394
1983521.10.26348
1984516.80.25781
1985506.10.25548
1986497.50.25572
1987498.40.25334
1988498.00.25680
1989498.60.25855
1990500.80.25495
1991503.90.25007
1992506.50.24415
1993509.10.23988
Source: Author’s calculations.

A 5 percent depreciation rate is assumed.

Source: Author’s calculations.

A 5 percent depreciation rate is assumed.

Table 10.Sub-Saharan Africa: External Debt Stock, Debt Outstanding and Disbursed, and Debt Service, 1970–95(In millions of U.S. dollars)
External Debt StockDebt Outstanding and DisbursedDebt Service
19708,296.05,761.4522.9
19719,772.06,769.3516.7
197211,235.07,571.0648.8
197315,168.09,928.51,165.5
197419,340.012,558.11,046.5
197522,721.014,879.71,300.9
197626,996.017,807.51,476.2
197733,705.021,968.31,554.0
197843,655.029,142.72,103.3
197955,485.036,422.82,853.3
198084,049.058,208.03,977.7
198175,668.057,190.04,576.6
198281,946.062,393.05,235.0
198388,233.067,164.06,411.1
198488,866.068,046.07,885.1
198596,396.075,979.08,918.9
1986127,145.0104,116.08,975.0
1987162,629.0133,135.08,296.0
1988164,981.0135,476.010,974.0
1989171,236.0140,735.09,761.0
1990190,260.0155,088.012,264.0
1991194,779.0159,364.011,176.0
1992192,781.0154,439.010,862.0
1993197,886.0153,372.09,890.0
1994212,416.0165,048.09,315.0
1995223,298.0174,281.08,784.0
Sources: World Bank, World Debt Tables, various issues; and World Bank (1995).
Sources: World Bank, World Debt Tables, various issues; and World Bank (1995).
Table 11.Sub-Saharan Africa: Average Terms of New Commitments and Commercial Bank Lending Rate, 1970–93(In percent)
Average Terms of New Commitments1Commercial Bank Lending Rate
19703.66.7
19714.26.8
19724.47.0
19735.56.9
19745.27.1
19755.67.5
19765.58.6
19775.58.0
19786.610.1
19798.110.8
19807.312.0
198110.113.4
19828.513.7
19838.214.6
19846.314.5
19855.914.5
19865.113.5
19874.513.5
19884.013.6
19894.115.1
19904.316.0
19914.116.2
19923.316.8
19933.318.0
Sources: World Bank, World Debt Tables, various issues; and World Bank, Annual Report, various issues.

Averages are for all creditors.

Sources: World Bank, World Debt Tables, various issues; and World Bank, Annual Report, various issues.

Averages are for all creditors.

With the rapid buildup of external debt and poor economic performance, sub-Saharan Africa’s debt crisis has deepened, and the debt burden has become even more crushing. Indeed, relative to exports and economic activity (as measured by GNP),3 sub-Saharan Africa’s debt is the highest of any region in the world (Klein, 1987; Iyoha and Iyare, 1994; International Labor Organization, 1995). According to the International Labor Organization (1995, p. 3),

Africa’s external debt is the highest in the world as a proportion of GDP; some countries in the region are spending more than half of their export earnings to service foreign debts. The debts of many African countries are so large in relation to their foreign exchange earnings potential that it would be impossible to pay them off even if growth resumed and was sustained at unrealistically high levels. Largely as a consequence of debt servicing, flow of capital from Africa is significantly more than flow of new capital to the region.

As a result of a deterioration in the terms and conditions of sub-Saharan Africa’s debt in the 1980s, the region’s debt-export ratio (measured as the ratio of debt outstanding and disbursed to exports of goods and services) doubled between 1980 and 1985, from 78.6 percent to 173.3 percent, and almost doubled again between 1985 and 1991, to 329.4 percent (Figure 2). The ratio of actual to scheduled debt-service payments deteriorated during the 1980s—falling from 66 percent in 1983 to 63 percent in 1989. Between 1989 and 1995 the average ratio of actual to scheduled debt service was 58.9 percent (Table 12 and Figure 3). Further evidence of debt distress is found in the fact that, in 1986, 27 of the 44 countries in Africa had payments arrears. Also, between 1986 and 1987, 23 sub-Saharan African countries renegotiated their official bilateral debt through the Paris Club. In the mid-1980s, 30 countries were officially classified as “debt distressed” (Klein, 1987). In the 1990s, 33 countries have been classified as Heavily Indebted Poor Countries (World Bank, 1996a). Finally, the huge burden of sub-Saharan African debt constitutes a serious obstacle to employment creation and growth, as “investment resources for productive pursuits are consistently used to meet external debt service obligations” (International Labor Organization, 1995, p. 3).

Table 12.Sub-Saharan Africa: Indicators of Debt Distress, 1989–95(In millions of U.S. dollars except as noted)
Actual Debt ServiceScheduled Debt ServiceRatio of Actual to Scheduled Debt Service
198912,88020,55162.7
199014,99925,61858.5
199113,37620,99963.7
199212,84321,19160.6
199311,71021,22555.2
199411,21221,77651.5
199512,20720,15460.6
Average12,74721,64558.9

Figure 2.Sub-Saharan Africa: Debt Indicators, 1970–95

(In percent)

Source: Author’s calculations based on data from World Bank, World Debt Tables.

Figure 3.Sub-Saharan Africa: Ratio of Debt Service Paid to Debt Service Due, 1989–95

(In percent)

These two issues—debt and lack of growth—are clearly interrelated. Indeed, all indications are that the excessive stock of external debt is retarding the growth and hampering the socioeconomic development of sub-Saharan Africa. It seems apparent that, given the structural weakness of most of these economies, their low income, and their low saving and low investment, the current high levels of debt and debt servicing would militate against rapid economic growth and development. Today, analysts and international policy-makers appear to have reached a consensus that a satisfactory recovery of investment and output growth in indebted sub-Saharan African countries will remain difficult, perhaps unattainable, as long as they are saddled with a burden of debt servicing that requires a sizable net transfer of resources abroad. Stated simply, there is no way in which many of these countries can service their debt and still have adequate resources left for development finance. Thus, many now believe that a necessary condition for economic growth and development in sub-Saharan Africa is debt relief. This must go beyond debt reschedulings and even beyond the Naples terms of 1994. It could be a policy package that combines debt reduction with a significant amount of debt forgiveness.

This econometric study uses a simulation approach to investigate the impact of external debt on economic growth in sub-Saharan African countries. In particular, it undertakes policy simulations (using alternative debt stock reduction scenarios) to analyze the effect of debt on investment and output in these countries between 1987 and 1994.

Following this introduction, the next section offers a discussion of the scope, nature, and severity of the debt crisis in sub-Saharan Africa. The section that follows presents a specification, estimation, and discussion of an econometric model of external debt and economic growth for the region. The basic model consists of two stochastic equations explaining output and investment, and two identities (a capital accumulation identity and a debt accumulation identity). The simultaneous-equations model is estimated by the two-stage least-squares technique. The model is dynamically simulated in the penultimate section, which also includes a discussion of the results of a historical simulation carried out for the 1971–94 period and policy simulations for 1987–94. The policy simulations involved assessing the impact of alternative debt stock reduction scenarios (effective in 1986) on investment and output in the subsequent years. The last section contains policy recommendations, a summary, and some concluding remarks.

External Debt of Sub-Saharan African Countries: Scope, Nature, and Severity

The external debt crisis of sub-Saharan Africa is best understood as an integral part of the global debt crisis that emerged in 1982. This crisis arose as a result of

  • overborrowing by the developing countries and reckless lending by international commercial banks in the 1970s;

  • the collapse of world commodity prices (especially petroleum) in the early 1980s; and

  • the sharp increase in international interest (lending) rates in 1982.

The phenomenal increase in foreign borrowing that preceded the debt crisis was triggered by the oil price shocks of 1973 and 1979, which resulted in acute current account deficits in most non-oil-producing developing countries. These countries resorted to foreign borrowing to tide them over the problems raised by the internationally generated shocks to their balance of payments. At the same time, during the period following the oil price hike of November 1973, the international commercial banks were awash with “petrodollars,” which they were anxious to recycle. Thus the needs of the cash-strapped developing countries and the excessively liquid international banks seemed to complement each other, and loans were liberally, if not recklessly, approved. Things appeared to go on smoothly for some years as the debts of developing countries were rolled over as they fell due.

The era of easy availability of credit came to a sudden end with the global financial crisis of 1982. The crisis was precipitated by the collapse of oil prices and the sharp increase in interest rates that led to the Mexican debt default. The immediate effect of the crisis was that developing country debt could no longer be rolled over. Henceforth, debtor countries could only service their debts by increasing exports or reducing imports. Many debtor countries in the developing world were ill prepared for the required belt tightening. Nor were world trade conditions conducive to any significant increase in export earnings by developing countries. The crisis was further aggravated by the actions of the international banks, which belatedly scrambled to cover dangerously exposed positions. This was the genesis of the global debt crisis, which saw a real fear of a domino effect resulting in massive defaults by the highly indebted countries of Latin America. Such an eventuality would certainly have shaken the international financial system to its very foundations and could have led to a global depression.

The external debt of the sub-Saharan African countries witnessed a rapid buildup during the period immediately following the global debt crisis, with external debt obligations skyrocketing from $84 billion in 1980 to $223 billion in 1995 (Table 13). Thus, the region’s external debt stock increased by 165.5 percent in 15 years, for an annual average growth rate of 6.7 percent.

Table 13.Sub-Saharan Africa: Total External Debt, 1970–95
In Millions of U.S. DollarsChange from Previous Year (In percent)
19708,296
19719,77217.79
197211,23514.97
197315,16835.01
197419,34027.51
197522,72117.48
197626,99618.82
197733,70524.85
197843,65529.52
197955,48527.10
198084,04951.48
198175,668–9.97
198281,9468.30
198388,2337.67
198483,8660.72
198596,3968.47
1986127,14531.90
1987162,62927.91
1988164,9811.45
1989171,2363.79
1990190,26011.11
1991194,7792.38
1992192,781–1.03
1993197,8862.65
1994212,4167.34
1995223,2985.12

The sharp external debt buildup in the post-1982 period is attributable to several factors, including

  • continued decline in the terms of trade;

  • uncontrolled fluctuations in export earnings;

  • higher interest rates;

  • realignment of exchange rates; and

  • rescheduling and refinancing of external debt, which only served to increase the debt stock.

Thus, it would be erroneous to attribute the current debt crisis in sub-Saharan Africa to purely wasteful and unproductive expenditures by these countries. The fact is that a significant proportion of the increase in the region’s external debt since 1982 can be attributed to exogenous factors. Indeed, the doubling of sub-Saharan Africa’s external debt stock in the 1980s took place during a period when the countries borrowed virtually no additional money for productive investment, and the meager borrowings were directed mainly to debt service and occasionally to the maintenance of minimal import requirements.4 Many analysts have pointed out that the rapid buildup in external debt in the 1980s occurred notwithstanding the sharp cutback made in their expenditures and the tenacious implementation of socially debilitating IMF- and World Bank-inspired structural adjustment programs. Despite the widespread adoption of structural adjustment programs, there was a sharp fall in real net resource flows, especially from the IMF and the International Bank for Reconstruction and Development (World Bank) during the 1980s (Table 14). Moreover, the implementation of structural adjustment programs has led to the sacrifice of medium- to long-term growth objectives in these countries (Iyoha, 1991b; United Nations Economic Commission for Africa, 1989).

Table 14.Sub-Saharan Africa: IMF, World Bank, and External Transfers, 1980–89
19801983198419851986198719881989
IMF
Gross disbursements11,2171,6189527387356781,033865
Repayments and interest24877399931,1721,6891,5411,4951,593
Net transfer730879–41–434–954–863–462–728
IDA
Disbursements4246377788811,4001,6811,6971,700
Repayments and interest2144567994111128126
Net transfer4035937228021,3061,5701,5691,574
IBRD
Disbursements400708832647898998581835
Repayments and interest3284385276168651,0731,3061,226
Net transfer722703053133–75–725–391
Total net transfers by IMF, IDA, and IBRD1,2051,742986399385632382455
Other net transfers
Multilateral3707664442487650709672607
Bilateral31,6572,2951,9254721,2101,194630945
Private42,818270–1,667–2,648–1,132–213–434–428
Total debt-related net transfers5,6574,0921,727–8562,0673,1851,7122,307
Total net transfers, including grants and foreign direct investment5,8436,6064,4603,2136,1637,6267,9739,420
Source: Author’s calculations from data in World Bank (1991, pp. 130-33).

Purchases.

Repurchases and charges.

Excluding grants.

Publicly guaranteed and unguaranteed, excluding foreign direct investment.

Source: Author’s calculations from data in World Bank (1991, pp. 130-33).

Purchases.

Repurchases and charges.

Excluding grants.

Publicly guaranteed and unguaranteed, excluding foreign direct investment.

Several factors have brought the issue of external debt to the forefront of the problems facing the sub-Saharan African countries. They include

  • the rapid rise of the stock of external debt, especially since 1982;

  • the deterioration in the terms and conditions of debt, caused partly by changes in the composition of debt but mainly by the rise in interest rates;

  • the rapid increase in debt-servicing obligations arising from the first two factors;

  • the emergence of debt-servicing problems, including risk of default and loss of creditworthiness;

  • the stubborn problem of debt overhang; and

  • the adverse effects of all these factors on the growth prospects of the countries in the region.

It is now generally agreed that the combined effects of these factors and a sharp fall in real net resource flows since the early 1980s have had a debilitating impact on the economies of sub-Saharan African countries. Most of these economies have shrunk during this period and there has been a significant decline in the standard of living. Real per capita incomes were on average lower at the end of the 1980s than at the beginning (Figures 1 and 4). In addition, consequent on the collapse of investment, a large part of the region’s infrastructure has disappeared, and gains made in education, health, and nutrition during previous decades were dissipated.

Figure 4.Sub-Saharan Africa: Growth of Real GDP, 1970–95

(In percent)

Source: World Bank (1992c, 1995).

Since 1970 there has been a pronounced upward trend in the external debt stock of sub-Saharan Africa. Table 13 provided data on the level and growth of the region’s external debt from 1970 to 1995. From $8.3 billion in 1970, sub-Saharan Africa’s total external debt stock increased rapidly to $22.7 billion in 1975. During this period, external debt grew at an average annual rate of 22.55 percent. The growth in external debt was even more rapid during the next five years: between 1976 and 1980 the average annual rate of growth of total external debt amounted to 30.4 percent. The highest annual growth rate during this period was 51.5 percent, reported in 1980. With total external debt at a worrisome $84 billion in 1980, its rate of growth moderated. Thus, between 1981 and 1985 the average annual growth rate of external debt was a mere 3 percent. The next two years, however, witnessed a renewed upsurge in the growth of external debt, at rates of 32 percent in 1986 and 28 percent in 1987. In 1987 the total external debt stock stood at the alarming level of $162.6 billion. Since 1988 the annual growth rate of external debt has been 7 percent or less, except in 1990 when it jumped to 11.1 percent. However, despite the moderate growth rate in the 1990s, by 1995 sub-Saharan Africa’s total external debt was astronomical at over $223 billion. During the decade of the 1980s, despite a reduction in the tempo of its growth, the region’s external debt increased faster than that of the developing world as a whole (Iyoha and Iyare, 1994).

Compared with Latin America, Africa depends more on official borrowing and less on commercial borrowing from international banks. According to Klein (1987, p. 12),

Official debt dominates the external obligations of most African countries. Only five Sub-Saharan African countries—Congo, Côte d’Ivoire, Gabon, Nigeria, and Zimbabwe—owe more to commercial banks than to official creditors.

However, while available empirical evidence shows that there has been a shift in the source of North Africa’s external finance during the 1980s, that of sub-Saharan Africa has hardly changed. Official bilateral and multilateral loans still account for the bulk of the region’s external debt. According to the World Bank (1996a, p. 170),

Seventy percent of outstanding debt at the end of 1995 was owed to official creditors—90 percent if Nigeria and South Africa are excluded. The region is a major recipient of inflows, most on concessional terms, from multilateral institutions. Thirty percent of long-term debt outstanding at the end of 1995 is owed to the World Bank Group, the International Monetary Fund, and the African Development Bank and Fund.

Available World Bank data also show that sub-Saharan Africa, as a region, receives the largest share of official development assistance. Indeed in 1994 official development assistance equaled 12.4 percent of the region’s GNP. This percentage is greater than that of other low- and middle-income regions, which are invariably less than 2 percent. In quantitative terms, official development assistance flows to sub-Saharan Africa reached $16.9 billion in 1995, up from $15.4 billion in 1994 (World Bank, 1996a, p. 170). Nevertheless, since the outbreak of the debt crisis there has been a tendency for the middle-income countries in sub-Saharan Africa to depend slightly more on commercial (private) external finance than was hitherto the case. Table 15 gives data on the structure and composition of sub-Saharan Africa’s external debt by maturity and type of creditor, while Figures 5 and 6 provide graphical illustrations. In the second half of the 1980s, loans from the multilateral agencies, particularly the IMF, began to dry up. By 1989, net transfers from both the IMF and the World Bank had become negative. This means that sub-Saharan African countries were, on a net basis, transferring resources to the Bretton Woods institutions (Table 14). Since many sub-Saharan African countries are low-income countries, disbursements from the International Development Association have remained relatively buoyant, actually rising from $424 million in 1980 to $1.7 billion in 1989.

Table 15.Sub-Saharan Africa: Composition of External Debt, 1980–95
Long-Term DebtUse of IMF CreditShort-Term Debt
Total External DebtPPGPNGTotalShare of total debt (In percent)AmountShare of total debt (In percent)AmountShare of total debt (In percent)
198084.053.84.658.469.53.03.622.626.9
1988165.0130.74.8135.582.17.04.322.513.6
1989171.2135.75.0140.782.26.43.724.114.1
1990190.3149.85.3155.181.56.63.528.615.0
1991194.8154.05.4159.481.86.63.428.814.8
1992192.8149.35.1154.480.16.43.332.016.6
1993197.9148.35.1153.477.57.03.537.519.0
1994212.4159.55.6165.177.77.93.739.418.6
1995223.3167.56.8174.378.17.23.241.818.7
Source: World Bank, World Debt Tables, various issues.Note: PPG, public and publicly guaranteed debt; PNG, private nonguaranteed debt.
Source: World Bank, World Debt Tables, various issues.Note: PPG, public and publicly guaranteed debt; PNG, private nonguaranteed debt.

Figure 5.Sub-Saharan Africa: Maturity Composition of External Debt, 1988–95

(In percent)

Source: World Bank, World Debt Tables.

Figure 6.Sub-Saharan Africa: Creditor Composition of External Debt, 1988–95

(In percent)

Source: World Bank, World Debt Tables.

From 58.4 percent in 1980, the share of long-term debt rose to a peak of 82.2 percent in 1989 and thereafter declined slightly but steadily, reaching 81.8 percent in 1991. The share of short-term credit in total external debt fell from 26.9 percent in 1980 to 14.8 percent in 1991, while the share of IMF finance in the total debt declined from 3.6 percent in 1980 to only 3.4 percent in 1991. See Figure 5 for an illustration.

The severity of sub-Saharan Africa’s problems is best appreciated by undertaking a comprehensive analysis of debt indicators. Debt indicators or debt ratios are measures of the debt burden. They can be used as analytical tools or for policy purposes; they may also be used for descriptive or predictive purposes (Nowzad and others, 1981). As a descriptive measure, external debt indicators provide in an understandable and functional way a history of the terms, structure, and scale of past borrowing. They may therefore be used in intertemporal analysis to examine the level and rate of change in debt capacity and debt-service capability.

In the analysis that follows, primary emphasis is given to an examination of the two most important debt indicators, the debt ratio and the debt-service ratio. Table 6 presented data concerning the key debt indicators for sub-Saharan Africa during the 1970–95 period, while Figure 2 provided a graphic illustration. The debt ratio is conventionally measured by the ratio of total external debt to GDP, or of total external debt to export earnings of goods and services.5 The debt-service ratio is conventionally measured by the ratio of debt-service payments to exports of goods and services. Table 6 shows that the ratio of external debt to GNP increased by half, from 20.5 percent to 30.7 percent, between 1970 and 1979. In the 1980s, Africa’s “lost decade” of development opportunities, GNP declined. With external debt growing faster than GNP, the ratio of external debt to GNP increased sharply. Sub-Saharan Africa’s debt-income ratio doubled in five years, rising from 28.7 percent in 1981 to 56.2 percent in 1986. The debt-income ratio then moderated but continued to rise in the late 1980s, reaching 70.8 percent in 1990. The 1990s witnessed a continued increase in the debt-income ratio, which peaked at 78.7 percent in 1994 before dipping to 74.1 percent in 1995.

The story told by the ratio of debt outstanding and disbursed to GNP is similar, although slightly more serious. Whereas the external debt-GNP ratio increased by 261.5 percent between 1970 and 1995, the ratio of debt outstanding and disbursed to GNP increased by 304.2 percent during the same period. In the sub-Saharan African countries, debt grew more rapidly than exports during the period under study. Thus, the debt-export ratio rose sharply during the period. The ratio of external debt to exports increased by 196.5 percent during the period, from 91 percent in 1970 to 269.8 percent in 1995. The ratio of debt outstanding and disbursed to exports rose by 233.1 percent, from 63.2 percent in 1970 to 210.5 percent in 1995. The debt-service ratio rose from 5.4 percent in 1970 to 22.0 percent in 1986 before falling to 14.7 percent in 1995. The debt-service ratio doubled in the 1970s, rising from 5.4 percent in 1970 to 10.3 percent in 1981. It doubled again between 1981 and 1986, rising from 10.3 percent to 22 percent. It remained rather steady between 1986 and 1988 and has fallen ever since (as a result of debt restructuring, debt reschedulings, and forgiveness of debt or interest), reaching a level of 14.7 percent in 1995.

This decline in debt-service payments hides more than it reveals. For many countries in the region, which are poor, highly indebted, and deficient in export earnings, there is often a wide divergence between actual and scheduled debt-service payments. Thus, there is generally a significant difference between the ratio of actual debt service to export earnings and that of scheduled debt service to export earnings (Table 17). Yet it is the ratio of scheduled debt to export earnings that is the more accurate measure of the pressure of debt service on export earnings and on the economy. According to Kapijimpanga (1996, p. 12).

The external debt burden reflects itself in the inability of a country to meet its debt service obligations (scheduled debt service) in relation to its foreign currency earnings. In Eastern and Southern Africa, the respective figures for this ratio during 1994 were Ethiopia 53 percent; Kenya 51 percent; Madagascar 53 percent; Mozambique 91 percent; Tanzania 64 percent; Uganda 50 percent; Zambia 51 percent; and Zimbabwe 26 percent. It is evident from these figures that debt is a serious problem for many of these countries. Mozambique was supposed to pay 91 percent of its foreign exchange earnings to service its external debt. In reality, simply because of lack of resources, Mozambique only paid 23 percent during the year.

Thus, for sub-Saharan African countries at least, the actual ratio of debt service to exports grossly underestimates the true nature of the debt burden and the debt crisis.6

Without exception, there was a deterioration in all the debt indicators during the period under study. The debt burden increased while the debt-service capacity deteriorated. Klein (1987) has further pointed out that some low-income African countries have been unable to repay debt rescheduled with a standard ten-year maturity, even with a five-year grace period. Increasingly, many countries are being compelled to reschedule already rescheduled debt, and being forced to borrow to pay interest on past borrowing, thus escalating total debt and falling deeper into the debt trap.

The severity of the debt crisis in sub-Saharan Africa may be better appreciated when it is realized that, since 1982, external debt has been escalating at a time of massive decline in export revenues and dwindling capital inflows. According to Drouin (1989, p. 3), “over the last decade, the foreign debt burden of the region grew worse than that of any other category of country facing debt servicing problems.” Also, Gonçalves asserts (1996, p. 5), that

the devastating impact of debt on the fragile African economies can easily be illustrated by revelations that Uganda, for example, spends only US$3 per person on health compared to US$17 per person on debt payments. From 1990 to 1993, Zambia spent 35 times more on debt payments than it did on primary school education.

Indeed, relative to exports and economic activity, sub-Saharan Africa’s debt is among the highest of any region in the world (Iyoha and Iyare, 1994). Table 16 presents data on debt indicators for developing countries by region. It is easily verified that, in the early 1990s, sub-Saharan Africa’s debt-export ratio was higher than that of any region except Latin America and the Caribbean. By 1995, its debt-export ratio, at 269.8 percent, was the highest of any region.

Table 16.Comparative Debt Indicators by Developing Region, 1990–95(In percent)
Share of Total Debt, 1995Debt-Export Ratio
1990199119921993199419951
All developing countries100161.6175.3166.7168.6162.8150.0
Sub-Saharan Africa11225.7239.4235.6251.9265.7269.8
East Asia and the Pacific23106.6106.0101.6101.393.383.3
South Asia8315.9311.7319.4287.9271.6245.7
Europe and Central Asia18120.2157.8143.6151.5153.7144.6
Latin America and the Caribbean29277.4282.0276.2274.6258.6254.2
Middle East and North Africa11109.8129.2126.1134.5148.5136.9
Source: World Bank (1996a).Note: Data are based on nominal debt stock at end of year.

Preliminary.

Source: World Bank (1996a).Note: Data are based on nominal debt stock at end of year.

Preliminary.

The fiscal dimension of the debt burden also needs to be highlighted. Given that the composition of debt is heavily tilted to the public domain, responsibility for debt service also falls heavily on the public sector. The heavy debt-service payments have inevitably put great pressure on the budget, leading to rising fiscal deficits in the highly indebted countries. This leads to several problems. The first is that taxes need to be increased in order to raise the resources (in domestic currency) to service the debt. One of the consequences of the anticipated tax burden is to depress investment—the debt overhang effect. Second, it is necessary to transform the domestic resources into foreign exchange, in which debt service must be paid. The desperate demand for foreign exchange to service debt often results in aid resources being routinely diverted to finance debt-service payments. Third, the stiff demand of high debt-service payments on the budget results in forced reductions in public investment and reduced spending on education and health. The diversion of resources from public investment to debt service is related to the crowding out hypothesis. Thus, pressures of debt service have relevance for the fiscal sustainability of debt in addition to the sustainability of high investment, as demonstrated in the debt overhang and crowding out of investment. Certainly, the brutal pressure of debt service on the budget explains not only the escalating budget deficits but also the increasingly large discrepancy between actual and scheduled debt-service payments in many sub-Saharan African countries. Growth is bound to be retarded because of the depressing effect on investment of heavy debt-service payments and the reduction of growth-supporting government expenditures.

As matters now stand, sub-Saharan African countries are constrained to use approximately 20 percent of their export earnings for debt servicing. Since the foreign exchange earnings of many of these countries are inadequate to begin with, it is apparent that reducing the already insufficient amount by 20 percent leaves grossly inadequate resources for financing development.7 This is why a study by the United Nations Economic Commission for Africa (1991, p. 10) concluded that

It is increasingly clear that very little progress, if any, can be made in Africa without the resolution of the debt crisis; and that there is no way in which Africa can service its existing debt and still have resources left for development financing.

There is a growing consensus among analysts on the need to go beyond traditional debt relief mechanisms (which consist mainly of IMF and World Bank loans and Paris Club8 reschedulings). Indeed, it is increasingly clear that for highly indebted low-income countries in sub-Saharan Africa, even the full use of traditional debt relief mechanisms would continue to prove insufficient in helping them achieve debt sustainability. As a result, many now believe that it is necessary to go beyond traditional debt relief mechanisms if many sub-Saharan African countries are to achieve sustainable debt levels within reasonable time horizons. In particular, because of the inadequate relief offered by debt reschedulings, some analysts now consider debt forgiveness or debt cancellation as the most appropriate way of reducing the debt burden of heavily indebted low-income countries. According to Ogbe (1992, p. 29),

Debt forgiveness or cancellation is, no doubt, the most complete and effective strategy of debt relief. The principal debt is not only extinguished but also the steady accumulation of debt that comes from repeated debt rescheduling and the resulting capitalization of interest and arrears are eradicated. Moreover, the sizeable administrative and financial burdens associated with periodic debt rescheduling are also eliminated.

A Model of External Debt and Growth in Sub-Saharan Africa

This section presents a small macroeconometric model that permits simulation of the effect of external debt on economic growth in sub-Saharan African countries. The basic macroeconometric simulation model consists of four equations, of which two are stochastic and two are identities. The two stochastic equations relate to the production function (output equation) and an investment demand equation incorporating a debt overhang variable and a variable designed to capture the crowding-out effect of debt-service payments. The third equation is a capital accumulation identity, and the fourth equation is a debt accumulation identity.

The Output Equation

The inspiration for the output equation used in this study is neoclassical, tracing its roots to Solow (1957), who hypothesized that output depended on capital and labor inputs and on disembodied technical change. It also owes much to the modifications introduced by development economists, particularly Chenery and his associates (for example, Chenery and Strout, 1966), who emphasized the role of investment and the investment-income ratio. This combination is now becoming standard in the development literature, and variations of the model have been used by Oseghale and Amenkhienan (1987), Ram (1985), Mjema (1996), Iyoha (1995, 1997), Khan and Kumar (1993), and Pindyck and Solimano (1993). In this study it is hypothesized that output depends on labor and investment per capita (used as a proxy for the capital stock). Thus, the output equation to be estimated econometrically using time series data is

where GDP is gross domestic product, L the labor force, and PCI per capita gross domestic investment; In stands for the natural logarithm, et is the random error term (assumed to be Gaussian white noise), a1 > 0 is the elasticity of output with respect to labor, and a2 > 0 is the elasticity of output with respect to investment per head.

Equation (1) was estimated using the ordinary least-squares regression method for the period 1970–94. From preliminary calculations it was found that the one-period lagged value of PCI gave better and more consistent results than its contemporaneous value. It was therefore decided to use PCIt-1 in place of PCIt. The estimated output equation obtained is

R2 = 0.964 R2¯=0.96 F(2, 21) = 279

SEE = 0.03

Mean of dependent variable = 12.266

D-W statistic = 0.58

where t-values are reported in parentheses below the coefficients. Given the value of the R2, it can be concluded that the two independent variables (labor and per capita investment) together explain over 96 percent of the systematic variations in output during the period being studied. The F value of 279 passes the significance test at the 1 percent level. Thus, the hypothesis of a significant linear relationship between output and the two independent variables is validated. The signs of both coefficients are correct, and the t-values of the two independent variables are highly significant; both pass the two-tailed test of significance at the 1 percent level. The high significance of ln PCIt-1 suggests the existence of a distributed lag relationship. In other words, investment affects output with a lag. However, the Durbin-Watson statistic is very low, indicating positive first-order serial correlation.

In an attempt to correct for autocorrelation, we experimented with both first-order and second-order autoregressive schemes using both the Cochrane-Orcutt method and the Newton-Raphson iterative inverse interpolation method (Pesaran and Pesaran, 1991). The most satisfactory results were obtained with a first-order autoregressive scheme using the Newton-Raphson method. It converged after only seven iterations. The estimated regression set of equation is

R2 = 0.984 R2¯=0.98 F(3, 20) = 408

SEE = 0.02

Mean of dependent variable = 12.266

D-W statistic = 1.73

where, as before, t-values are reported below the coefficients. An examination of Equations (3) and (4) confirms that we have improved on Equation (2). The serial correlation has been eliminated, as indicated by a Durbin-Watson statistic of 1.73. The first-order autoregressive parameter, ρ^, lies between zero and unity in absolute value. Its t-value is significantly different from zero at the 1 percent level. With an R2 of 0.984, it is apparent that we are able to explain over 98 percent of the systematic variation in GDP by the two independent variables. Both variables are correctly signed. The labor force variable passes the significance test at the 1 percent level, but investment fails the significance test at the usual 5 percent level. It nevertheless passes the test at the 10 percent level. The F-statistic of 408 is significant at the 1 percent level. Thus the hypothesis of a significant linear relationship between GDP and the two regressors is validated. A plot of actual and fitted values of ln GDP is given in Figure 7.

Figure 7.Sub-Saharan Africa: Actual and Fitted Values of GDP, 1971–94

(In logarithmic values)

Source: Author’s calculations.

The Investment Equation

Gross domestic investment collapsed in sub-Saharan Africa in the 1980s, especially starting from 1983. According to World Bank data, from a level of $44 billion in 1981, gross domestic investment rapidly fell to a low of $23 billion in 1984. From that level it rose slowly, reaching $37.6 billion in 1991. According to the World Bank (1994b), the average annual growth rate of gross domestic investment was 5.1 percent over 1970–80 but was –3 percent between 1980 and 1992.

As may be expected, the investment-GDP ratio also declined sharply. In analyzing the movement of the investment-GDP ratio, the World Bank (1992a) shows that while the annual average ratio was 21.5 percent in 1975–79, it collapsed to 15.6 percent during 1980–85. Indeed, if the period 1983–86 is considered, it is found that the average investment-GDP ratio was only 13 percent. In fact, from a ratio of 26 percent of GDP in 1977, the investment ratio had fallen to a catastrophic 12.8 percent in 1984. No doubt a fall in gross domestic saving partly accounts for this. World Bank (1992a) statistics show that the ratio of gross domestic saving to GDP, which averaged 18.9 percent between 1975 and 1979, had fallen to 13.4 percent between 1980 and 1985. Many analysts believe that the poor investment and growth performance of many highly indebted developing countries (including those in sub-Saharan Africa) since the onset of the global debt crisis in 1982 can be attributed in part to the disincentive effect of their external debt burden. This phenomenon is often referred to as the debt overhang problem. The debt overhang hypothesis posits that the accumulated external debt of these countries acts as a tax on future output and thus discourages private investment. The theoretical case for the debt overhang effect has been made by several authors, including Dooley (1986), Krugman (1988), Sachs (1989), Froot (1989), and Calvo (1989). Some attempts have also been made to test the hypothesis empirically. Among these are Sachs (1989), Claessens (1990), Borensztein (1990, 1991), Cohen (1990), Warner (1992), Degefe (1992), Savvides (1992), Chhibber and Pahwa (1994), and Iyoha (1997). Researchers like Sachs (1989) and Krugman (1988) have also analyzed the crowding-out effect of debt-service payments. This arises from the fact that many highly indebted poor countries frequently divert resources, including foreign aid and other foreign exchange resources, to take care of pressing debt-service obligations, particularly debt owed to the multilateral financial institutions, which are deemed nonreschedulable.

Borensztein (1990) provided a major and interesting attempt to empirically test the debt overhang effect. Using data for the Philippines, he found that the debt overhang hypothesis was largely valid. Specifically, he found that the debt overhang had an adverse effect on private investment, which was strongest when private debt, rather than total debt, was used as a measure of debt overhang.

Following Sachs (1989), Krugman (1988), Borensztein (1990), Chhibber and Pahwa (1994), and Iyoha (1995, 1997), we specify an investment demand function that has its roots in neoclassical optimization theory. We also make allowance for the potential existence of both a debt overhang effect and a crowding-out effect of external debt. Finally, we allow for the possible existence of an investment accelerator effect. In the specification, investment per capita is hypothesized to depend negatively on domestic interest rates, positively on the marginal product of capital, negatively on the price of investment goods, positively on growth in real GDP, negatively on the external debt-income ratio, and negatively on the ratio of debt service to exports. Thus, the basic specification of the investment demand function is given by

The a priori signs for the coefficients are b1 < 0, b2 > 0, b3 < 0, b4 > 0, b5 < 0, and b6 < 0. PCI is per capita gross domestic investment; r is the interest rate (the commercial lending rate); MPK is the marginal product of capital;9PI is the price of investment goods; GDPGR is the growth rate of real output, which is expected to capture the investment accelerator effect; D/Y is the ratio of the external debt stock to GNP, which is the usual measure of debt overhang; DS/X is the ratio of total debt-service payments to exports of goods and services, which is expected to capture the crowding-out effect of external debt; and u is a stochastic error term assumed to be Gaussian white noise.

Some researchers, including Borensztein (1990), have suggested that the debt overhang effect should be particularly strong when considering private investment and private debt. This hypothesis could not be tested in this study for lack of the requisite data.10

From preliminary ordinary least-squares regression calculations, it was found that the partial regression coefficient of the price of investment goods (PI) in Equation (5) was not statistically significant, and the variable was therefore dropped in the final equation. The one-period-lagged values of the interest rate and the debt-GNP ratio were also found to give better and more consistent results than their contemporaneous values. It was therefore decided to use rt-1 in place of rt and D/Yt-1 in place of D/Yt. The amended equation was then estimated by the ordinary least-squares technique, and the following results were obtained:11

R2 = 0.88 R2¯=0.84 F(5,18) = 26

SEE = 9.49

Mean of dependent variable = 78.33

D-W statistic = 1.4.

In Equation (6), t-values are again given in parentheses below each coefficient. Judging from the value of R2, it can be concluded that the independent variables in this equation explain over 87 percent of the systematic variations in per capita gross domestic investment during the period under study. The F-value of 26 is significant at the 1 percent level, indicating that there is a significant linear relationship between the independent variables taken together and per capita investment. All the signs are correct except that of the interest rate. Of the four correctly signed independent variables, three (MPK, the debt-income ratio, and the debt-service ratio) pass the significance test at the 1 percent level. The fourth variable, the growth rate of real GDP, misses the two-tailed test of significance at the traditional 5 percent level. It nevertheless passes the significance test at the 8 percent level. Thus, there is some evidence of an investment accelerator effect, but it is not very strong. Since the debt overhang variable (D/Y) and the crowding-out effect variable (DS/X) are highly significant, there is clear evidence of both effects. Since their signs are negative, as postulated, their effect is to depress the level of investment. However, the D-W statistic shows evidence of serial correlation.

In an attempt to correct for autocorrelation, we experimented with both first-order and second-order autoregressive schemes using both the Cochrane-Orcutt method and the Newton-Raphson iterative method. The most satisfactory results were obtained with a second-order autoregressive scheme using the Newton-Raphson iterative method.

It converged after only nine iterations. The estimated regression equation was

R2 = 0.9 R2¯=0.86F(7, 16) = 20.7

SEE = 9.11

Mean of dependent variable = 78.33

D-W statistic = 2.09

where t-values are reported in parentheses below the coefficients. An examination of Equations (7) and (8) confirms that we have improved upon Equation (6). The serial correlation has been removed as indicated by a D-W statistic of 2.09. The second-order autoregressive parameters, ρ1^ and ρ2^, lie between zero and unity in absolute value. Their t-values are significantly different from zero at the 2 percent and the 12 percent level, respectively. With an R2 of 0.9, it is clear that we are able to explain 90 percent of the systematic variation in per capita gross domestic investment by the five independent variables. All the variables except the interest rate have the correct signs, but only three are significantly different from zero, at traditional levels of significance. MPK passes the significance test at the 1 percent level. The debt-service ratio is also significantly different from zero, using a two-tailed test, at the 1 percent level. The debt overhang variable (D/Y) passes the significance test at the 5 percent level. The investment accelerator variable (GDPGR) only passes the two-tailed significance test at the 10 percent level. The F-statistic of 20.7 is significant at the 1 percent level. Thus, the hypothesis of a significant linear relationship between per capita investment and the five independent variables is validated. A plot of actual and fitted values of per capita investment is given in Figure 8.

Figure 8.Sub-Saharan Africa: Actual and Fitted Values of Per Capita Investment, 1971–94

(In 1987 U.S. dollars)

Source: Author’s calculations.

The results offer a confirmation of the debt overhang hypothesis for the sub-Saharan African countries. The debt overhang variable, proxied by the ratio of external debt to GNP, is negative and highly significant. The elasticity12 of investment with respect to the debt overhang variable is –0.337. Thus, a 10 percent decrease in the debt-GNP ratio will lead to approximately a 3 percent increase in investment per capita. Notice that the debt-income ratio affects investment with a lag. The effect of the combination of this lagged response of investment to changes in the debt level and the lagged response of output to changes in investment (arising inter alia from gestation lags) means that the impact of a reduction in the debt stock on output will not be instantaneous. Rather, it will manifest after some time lag.

The results also provide confirmation of a crowding-out effect in sub-Saharan African countries. The coefficient of the crowding-out effect variable, proxied by the ratio of total debt-service payments to exports of goods and services, is negative and significantly different from zero at the 1 percent level. The elasticity of investment with respect to the crowding-out variable is –0.335. Thus, a 10 percent reduction in the debt-service ratio would increase investment by approximately 3 percent.

To summarize, we find, using econometric techniques, that sub-Saharan Africa’s heavy external debt stock and debt-service payments act to reduce investment through both the debt overhang effect and the crowding-out effect. There is evidence of a distributed lagged response of investment to changes in the debt-income ratio (and of a distributed lagged response of output to changes in investment). Thus, reductions in debt stock will affect investment and output after some lag in time (rather than instantaneously).

A Simultaneous-Equations Model of External Debt and Economic Growth in Sub-Saharan Africa

To make explicit allowance for interaction between external debt and economic growth, a simultaneous-equations model is now specified. In particular, the output equation and investment demand function are considered as a system of simultaneous equations. The simultaneous-equations system is given by these modified equations:

where u1t and u2t are stochastic error terms. In this system of equations, there are two endogenous variables, ln GDP and PCI. There are seven predetermined variables: lnL, lnPCI, rt-1,MPK, GDPGR, (D/Y)t-1, and DS/X. Using the order condition of identification, it can be ascertained that both equations are overidentified. The two-stage least-squares regression method can therefore be used to estimate the equations of the model. Given the identification status of the equations, econometric theory assures us that the resulting estimates of the structural coefficients will be consistent and asymptotically efficient.

Using the two-stage least-squares technique, the following estimates were obtained, with t-values reported in parentheses below the coefficients:

R2 = 0.943 R2¯=0.938 F(2, 21) = 173

SEE = 0.039

Mean of dependent variable = 12.26

D-W statistic = 1.02

R2 = 0.9 R2¯=0.86 F(6, 17) = 25

SEE = 8.96

Mean of dependent variable = 78.33

D-W statistic = 1.85.

To test whether it makes much difference empirically which definition of debt is used—total external debt stock (EDT) or debt outstanding and disbursed (DOD)—the system of equations was reestimated using DOD/GNP instead of EDT/GNP as the measure of debt overhang. The estimated results using DOD instead of EDT (with t-values in parentheses under the coefficients) were

R2 = 0.96 R2¯=0.956 F(2, 21) = 250

SEE = 0.03

Mean of dependent variable = 12.26

D-W statistic = 0.78.

R2 = 0.909 R2¯=0.876 F(6,17) = 28.2

SEE = 8.465

Mean of dependent variable = 78.33

D-W statistic = 1.84.

A comparison of Equations (12) and (14) shows that the overall fits are about the same.13 Also, the same variables—GDPGR, the debt-income ratio, and the debt-service ratio—are significant in both. Thus, there is little difference as to whether we use EDT or DOD as our measure of debt. However, we choose to use EDT since it is a broader and more comprehensive measure of external debt.

Capital Accumulation Identity

This is one of the two identities used to close the model. This study uses the basic empirical relationship between the capital stock and investment: abstracting from depreciation, change in the capital stock is equal to investment. After including depreciation, which has the effect of reducing the capital stock, the capital accumulation identity is specified as

where δ is the average rate of depreciation of capital in sub-Saharan Africa,14KSTOCK is the capital stock, and INV is investment.

Debt Accumulation Identity

The model is closed with a second identity, a debt accumulation identity. Given knowledge of the debt situation in sub-Saharan Africa, it was decided to use the following identity:

where DTOTAL is total debt, DSPAY is total debt-service payments, and AVINT is average interest paid on debt.

Simulation Results

In the course of this econometric study, both a historical simulation and a policy simulation were undertaken. These are discussed in turn.

Historical Simulation

A historical simulation, also called an ex post simulation, is performed over the estimation period of a macroeconometric model. The main reasons for historical simulation are model validation and evaluation. A comparison of the actual or historical series for the endogenous variables with the simulated series for the same variables is often used to test the validity of a macroeconometric model. Such a comparison allows an analyst or policymaker to determine how well a macroeconometric model tracks the economy and, specifically, how well each simulated data series tracks or approximates the corresponding actual series. Measures used to test how closely a simulated series of an endogenous variable tracks its corresponding historical series include

  • the root-mean-square (rms) simulation error;

  • the root-mean-square percent error; and

  • Theil’s inequality coefficient, U.15

In this study, the root-mean-square simulation error and Theil’s inequality coefficient (Theil, 1978, p. 368) are used in addition to the correlation coefficient between the actual and simulated data series. To achieve a more efficient simulation and obtain richer results, the number of endogenous variables was increased to seven by specifying identities for the logarithm of per capita investment, the debt-GNP ratio, and the debt-service ratio.

Dynamic simulation was then undertaken. Simulations were done for the seven endogenous variables: the logarithm of GDP, per capita investment, total external debt, the debt-GNP ratio, the debt-service ratio, the logarithm of per capita investment, and the capital stock. Graphical representations of the historical simulations for output and investment are shown in Figures 9 and 10, respectively. Table 17 contains summary statistics for the historical simulations—the correlation coefficient, root-mean-square error, and Theil’s inequality coefficient—for five key endogenous variables. Notice that the values for Theil’s inequality coefficient for output and investment are close to zero, signifying a close fit between the simulated and actual series of the two variables. Also, the correlation coefficients between their simulated and actual series are 97 percent and 85 percent, respectively.

Figure 9.Sub-Saharan Africa: Actual and Dynamically Simulated Values of GDP, 1971–94

(In logarithmic values)

Source: Author’s calculations.

Figure 10.Sub-Saharan Africa: Actual and Dynamically Simulated Values of Per Capita Investment, 1971–94

(In 1987 U.S. dollars)

Source: Author’s calculations.

Table 17.Historical Simulation: Summary Statistics
VariableCorrelation CoefficientRoot-Mean-Square ErrorTheil’s Inequality Coefficient
19611966
GDP0.970.040.00160.003
Investment0.8513.090.0810.160
Debt0.9430.580.1320.254
Debt-GNP ratio0.9112.930.1380.261
Debt-service ratio0.783.730.1320.264
Source: Author’s calculations.
Source: Author’s calculations.

Policy Simulation

Policy simulation is the term applied to experimentation with a macro-econometric model using alternative policy scenarios. In this study the main objective of the policy simulation exercise was to examine the effect of debt stock reduction on the endogenous variables, particularly investment and output. Results of this type of simulation for Nigeria were recently reported by Chhibber and Pahwa (1994) and by Iyoha (1997). As an illustrative example, this study considers policy scenarios of a reduction of 40 percent, 50 percent, and 75 percent in the external debt stock effective from 1986. Dynamic simulations for the 1987–94 period were then undertaken to investigate the effects of the various packages of debt stock reduction on key endogenous variables. In the simulations, it was assumed that a given debt stock reduction led to a corresponding reduction in the debt-service ratio after a one-year lag.

The simulation results came out as expected. A reduction in the debt stock led to a rise in per capita investment, an increase in GDP, a fall in total external debt, a fall in the debt-GNP ratio, and a fall in the debt-service ratio. The results of the policy simulations are summarized in Tables 18, 19 and 20, and graphs of the simulations of output and investment with a 75 percent debt reduction are presented in Figures 11 and 12, respectively.

Table 18.Policy Simulations: Per Capita Investment Under Alternative Debt Reduction Scenarios, 1987–94(In U.S. dollars)
Debt Reduction
None (Baseline)40 percent50 percent75 percent
198762.2866.4673.2989.28
198859.4263.8771.1087.88
198952.0757.0965.0483.06
199040.5446.9256.2576.13
199147.6853.4662.3982.18
199252.5858.1566.9486.83
199356.8462.3271.0991.18
199457.1462.7871.8092.46
Source: Author’s calculations.
Source: Author’s calculations.
Table 19.Policy Simulations: Logarithm of Output Under Alternative Debt Reduction Scenarios, 1987–94(In billions of U.S. dollars)
Debt Reduction
None (Baseline)40 percent50 percent75 percent
198712.3512.3612.3712.40
198812.3712.3812.4012.42
198912.3912.4012.4112.44
199012.3812.4012.4212.46
199112.4212.4412.4612.49
199212.4612.4712.4912.52
199312.4912.5112.5212.55
199412.5212.5312.5512.58
Source: Author’s calculations.
Source: Author’s calculations.
Table 20.Policy Simulations: Effect of a 75 Percent Debt Stock Reduction on External Debt Stock, Debt-GNP Ratio, and Debt-Service Ratio, 1987–94
Total Debt (In billions of U.S. dollars)Debt-GNP ratio (In percent)Debt-Service Ratio (In percent)
Without reductionWith reductionWithout reductionWith reductionWithout reductionWith reduction
198799.4436.5139.3214.4415.145.75
1988115.5941.9447.0817.0815.756.00
1989133.3347.9453.8219.3516.396.22
1990153.8354.8257.2620.4117.106.49
1991176.3062.4063.8922.6217.806.76
1992201.3170.8572.8825.6518.397.00
1993228.9880.2384.6929.6818.997.21
1994259.8190.7096.2733.6119.627.45
Source: Author’s calculations.
Source: Author’s calculations.

Figure 11.Sub-Saharan Africa: Simulation of GDP With and Without a 75 Percent Debt Stock Reduction, 1971–94

(In logarithmic values)

Source: Author’s calculations.

Figure 12.Sub-Saharan Africa: Simulation of Per Capita Investment With and Without a 75 Percent Debt Stock Reduction, 1971–94

(In 1987 U.S. dollars)

Source: Author’s calculations.

An analysis of the results reported in Table 18 shows that, between 1987 and 1994, simulated average per capita investment without a debt stock reduction was $53.57. We may then compare this outcome with an average per capita investment of $58.88 if the debt stock is reduced by 40 percent; $67.23 if the debt stock is reduced by 50 percent; and $86.12 if the debt stock is reduced by 75 percent. Thus, a 75 percent16 reduction in the debt stock in 1986 would have increased per capita investment during the 1987-94 period by 60 percent, on average.

An examination of Table 19 indicates that the simulated average level of lnGDP without debt stock reduction was 12.42. If the external debt stock had been reduced by 40 percent in 1986, the average level of lnGDP during 1987–94 would have risen to 12.44. If the external debt stock had been reduced by 50 percent, the average level of lnGDP would have increased to 12.45, and if the external debt stock had been reduced by 75 percent, the average level of lnGDP during the subsequent eight-year period would have risen to 12.48. Thus, a 75 percent reduction in the external debt stock would have raised the logarithm of GDP from 12.42 to 12.48, or by 6 percent. Note that between 1987 and 1994 the average recorded growth rate of real GDP in sub-Saharan Africa was 2.1 percent. Adding 6.0 percent to this figure, we get 8.1 percent. Therefore, the implication of our results is that if there had been a 75 percent debt stock reduction in 1986, the average growth rate of real GDP during 1987–94 would have been a healthy 8.1 percent instead of the anemic 2.1 percent actually achieved.17 Results obtained in this study are similar to those obtained by Chhibber and Pahwa (1994) for Nigeria. They simulated a 50 percent debt reduction for Nigeria, also starting in 1986; they found that during the 1987–91 period GDP growth would have averaged more than 2 percent higher. Thus, with a 50 percent debt stock reduction, Nigeria’s average GDP growth during 1987–91 would have been 7.28 percent instead of the recorded 5.18 percent (Chhibber and Pahwa, 1994, p. 132).

It is necessary to point out that, mainly because of the operation of the debt overhang effect and the crowding-out effect, there would have been a significant recovery of investment and an increase in growth of real output as a result of the hypothesized reductions in debt stock. There would also have been corresponding improvements in debt indicators. An analysis of Table 20 confirms that a 75 percent debt stock reduction would, by 1994, have reduced the debt-income ratio by 65 percent. Similarly, the debt-service ratio would have declined by approximately 62 percent as a result of a 75 percent debt stock reduction.

Summary and Conclusions

This study presented a small macroeconometric model that facilitated simulations of the impact of external debt on economic growth in sub-Saharan Africa. The complete simultaneous-equation model consisted of two stochastic equations (for output and investment demand) and five identities. The five identities in the dynamic simulation were for debt accumulation, capital stock, the debt-GNP ratio, the debt-service ratio, and the logarithm of per capita investment. The simultaneous-equation model was estimated by two-stage least squares for the 1971–94 period. It was found that there is a significant debt overhang effect as well as a crowding-out effect. In other words, the large stock of external debt and heavy debt-service payments have had a depressing effect on investment in the region.

Historical simulation was undertaken in order to assess the validity of the model. On the whole, the model performed well, with the simulated values of the two key endogenous variables (investment and output) closely tracking their corresponding actual data series. In this respect, Theil’s inequality coefficient (1961 version) of less than 0.002 was obtained for both investment and output.

Policy simulation was also carried out. In particular, the effect of alternative debt stock reduction packages effective in 1986 was simulated. It was found that hypothesized debt reductions of 40 percent, 50 percent, and 75 percent, assumed effective in 1986, would have significantly increased investment and GDP during the subsequent period. Similarly, the various debt stock reduction scenarios would have reduced, to corresponding degrees, total external debt stock, the debt-GNP ratio, and the debt-service ratio. The policy simulations showed that a 75 percent debt stock reduction would have raised per capita gross domestic investment by about 60 percent and increased GDP growth by 6 percentage points, on average, during the 1987–94 period. These results largely mimic those obtained for Nigeria by Chhibber and Pahwa (1994) and Iyoha (1997).

Results obtained in this study confirm that an excessively high stock of external debt depresses investment and lowers the rate of economic growth. Thus, heavily indebted countries in sub-Saharan Africa need to articulate creative strategies for bringing about debt reduction so that the high debt stock and associated crushing debt-service burden do not have such a negative impact on economic growth. Traditional debt relief mechanisms currently being used by these countries include debt restructuring, debt rescheduling, reduced debt servicing, debt buybacks, interest rate options, and various debt conversion schemes such as debt-equity swaps. Overall, the effectiveness of these techniques in reducing the debt stock has been rather limited (Ogbe, 1992). New steps that could be taken to effectively reduce the region’s external debt stock include, inter alia,

  • adoption and implementation of macroeconomic policies that would encourage repatriation of flight capital;

  • adoption of a medium-term economic program approved by the IMF and the World Bank in order to qualify for debt reduction under the enhanced Toronto terms, the Naples terms, and the IDA reduction facility; and

  • pressing for debt forgiveness or debt cancellation through diplomatic action.

Given the potential beneficial effects of debt stock reduction on investment and GDP in sub-Saharan Africa, it is recommended that the international community make a greater effort to provide debt reduction, preferably through debt forgiveness, as a matter of priority. It seems clear that, provided appropriate domestic macroeconomic policies are adopted and implemented along with debt reduction packages, debt reduction would provide a much-needed stimulus to investment recovery and growth in sub-Saharan African countries in the years ahead.

Although plausible, the results obtained in this chapter by the use of two stochastic equations and five identities should be considered preliminary. It is expected that more robust results would be obtained by carrying out the simulations with a larger macroeconometric model of sub-Saharan Africa.

References

The total external debt stock consists of public and publicly guaranteed long-term debt, private nonguaranteed long-term debt, the use of IMF credit, and estimated short-term debt. This is to be distinguished from debt outstanding and disbursed, which is defined as the total debt outstanding and disbursed of long-term official concessional, official nonconcessional, and private loans. Long-term external debt is defined as debt that has an original or extended maturity of over one year that is owed to nonresidents and repayable in foreign currency, goods, and services.

Total debt service measures debt-service payments on long-term debt (public, publicly guaranteed, and private nonguaranteed), use of IMF credit, and interest on short-term debt.

According to the World Bank (1996b), sub-Saharan Africa’s net present value of external debt amounted to 50 percent of GNP in 1994. This compares unfavorably with the ratios for the other low- and middle-income regions of the world, namely, Latin America and the Caribbean (39 percent), Middle East and North Africa (32 percent), East Asia and the Pacific (28 percent), South Asia (26 percent), and Europe and Central Asia (25 percent).

In other words, external borrowing by sub-Saharan African countries in the 1980s was not “purposive” but “reactive.” The foreign borrowing was not undertaken to boost investment and the rate of development. Rather, borrowing was undertaken to finance unexpected and unmanageable current account deficits and to finance burdensome debt-service payments (see Iyoha, 1991b).

There is some debate as to what variable should be used for normalization, that is, as the denominator of total external debt and debt-service payments. Taking a short-term view, exports of goods and services would appear to be the more appropriate scalar, as they are the main source of the foreign exchange resources that can be used to service debt. However, taking a longer-term view, GDP seems to be the appropriate scalar, since the foreign exchange constraint may be lessened by growth in the nontraded sector, provided prices are flexible (Nowzad and others, 1981). Iyoha (1991b) argues for using different scalars for total external debt and debt service: GDP should be used to normalize total external debt, and exports of goods and services as the denominator for debt-service payments. Thus the external debt-GDP ratio may be interpreted as a measure of debt capacity or debt burden and the debt-export ratio as a measure of debt-service capacity or debt-service burden.

The ratio of international reserves to total external debt, which reached 27.3 percent in 1980, collapsed during the period—falling to a low of 6.3 percent in 1988—but recovered slightly to 8.2 percent in 1994. The reserves-import ratio deteriorated during the period, falling from 3 months in 1980 to a low of 1.5 months in 1988 before recovering to 2.1 months in 1994.

This is why Goran Ohlin (1966) maintains that a poor country has no debt-servicing capacity at all. According to Ohlin, for low-income developing countries, debt service competes with essential imports for foreign exchange earnings, which are insufficient to begin with, and with the investment needs of the country for savings, which are often deficient and inadequate. Besides, high debt-service ratios make a country extremely vulnerable to pressures on its balance of payments, further hampering the development effort.

The Paris Club of creditors represents only government-guaranteed creditors. Its membership includes Canada, Germany, France, the United Kingdom, and the United States. These governments guarantee the export activities of their nationals through their official export credit agencies. The first Paris Club meeting was held in 1956. Meetings are held to discuss repayment problems and negotiate debt relief agreements for debtor nations. The London Club of creditors deals with uninsured and unguaranteed debts extended by commercial banks to nationals of debtor countries. Thus, the members are commercial banks based in the industrial countries. The first London Club meeting was held in 1976. Meetings are called to discuss repayment problems and conclude restructuring agreements with debtor nations.

Following Borensztein (1990), we approximate MPK by the real average product of capital multiplied by the relative share of capital in output. The capital stock and MPK series used are available on request.

Iyoha (1997) also finds econometric evidence of a private debt overhang for Nigeria: there is both a debt overhang and a crowding-out effect with respect to private investment. Thus, the evidence is that heavy debt-service payments do crowd out (that is, reduce) not only public investment but also private investment.

The following alternative estimate was obtained using D/XGS instead of D/Y as a measure of the debt overhang:

R2 = 0.878 R2¯=0.84 F(5,18) = 25.9

SEE = 9.5

Mean of dependent variable = 78.33

D-W statistic = 1.35.

A comparison of the two equations shows that the fits are similar, although the equation with D/Y is slightly better since both the debt overhang variable and the crowding-out variable are significantly different from zero. Therefore, the equation in the text, using D/Y as the debt overhang variable, is preferred.

Since this is a linear equation, the elasticity-at-the-mean of a given independent variable is obtained as the coefficient of that variable multiplied by its mean and divided by the mean of the dependent variable.

The estimated coefficients for the output equation in the two sets are identical because the system is recursive.

In the empirical work, a depreciation rate of 7 percent was found to give satisfactory results.

The three measures are defined thus:

A 75 percent reduction in the debt stock is close to the effective debt reduction for which low-income highly indebted countries are eligible under the Naples terms of December 1994 (Central Bank of Nigeria, 1996, p. 10).

Note that the small macroeconometric model used in this study permits debt (or debt stock reduction) to affect growth only via investment. A larger model would, in addition, allow for a direct transmission channel from debt to growth. These direct effects are linked to the impact of debt on growth-supporting expenditures such as human capital and on expenditures that enhance investment productivity. In such a case, debt stock reduction would increase growth via a rise in investment and directly by inducing increased growth promotion expenditures. Conceivably, then, such a specification would increase the impact of debt stock reduction on economic growth.

    Other Resources Citing This Publication