Discussion Paper No.

2001/116 The Impact of External Debt on Economic Growth in Kenya

An Empirical Assessment

Maureen Were *
October 2001 Abstract
A group of low-income countries classified as HIPCs have continued to experience difficulties in managing and servicing their huge stocks of external debt. Most of these countries including Kenya are in Sub-Saharan Africa. The relatively high level of Kenyas external indebtedness and rising debt burden has serious implications on the countrys development and debt sustainability initiatives. While the economic performance continue to deteriorate, there has been significant net outflow of resources to meet the debt obligations in the 1990s. This paper examines the structure of Kenyas external debt and its implications on economic growth. The findings of the study indicate that Kenyas external debt is mainly official, of which a bigger proportion is from multilateral sources. External debt accumulation has been rising over the years with debt burden indicators increasing steadily in the early 1990s. Using time series data for the period 197095, the empirical results indicated that external debt accumulation has a negative impact on economic growth and private investment. This confirms the existence of a debt overhang problem in Kenya. However, the results also indicated that current debt inflows stimulate private investment. Debt servicing does not appear to affect growth adversely but has some crowding-out effects on private investment. Several policy implications emerge from the study. The simultaneous attainment of sustainable levels of economic growth and external debt appear difficult at the moment and could remain elusive if aggressive /..

Keywords: Kenya, debt, growth JEL classification: O55, F34, O11

Copyright Author(s) 2001 * Kenya Institute for Public Policy Research and Analysis (KIPPRA), Nairobi This is a revised version of the paper originally prepared for the UNU/WIDER development conference on Debt Relief, Helsinki, 17-18 August 2001. UNU/WIDER gratefully acknowledges the financial contribution from the governments of Denmark, Finland and Norway to the 2000-2001 Research Programme.

measures are not undertaken. In view of the current economic recession and the negative net outflows, the results obtained from this study support the need for Kenya to be considered for comprehensive debt relief measures. There are prospects that availability of these resource flows can stimulate private investments if used productively. A key challenge to the government remains that of ensuring efficiency in delivery of services and increased productivity of public investments. In addition, creating credibility including commitment and political will to reform implementation is required to spur investor confidence for both local and foreign investments.

Abbreviations BOP ECT EDT/GNP EDT/XGS GDP GNP HIPC IMF INT/GNP INT/XGS LDCs SSA TDS/XGS TOT balance of payments error correction term total external debt as a ratio of GNP total external debt as a ratio of exports of goods and services gross domestic product gross national product heavily indebted poor countries International Monetary Fund interest payments a ratio of GNP interest payments on debt as a ratio of export of goods and services less developed countries Sub-Saharan Africa total debt service payments as a ratio of exports of goods and services terms of trade

UNU World Institute for Development Economics Research (UNU/WIDER) was established by the United Nations University as its first research and training centre and started work in Helsinki, Finland in 1985. The purpose of the Institute is to undertake applied research and policy analysis on structural changes affecting the developing and transitional economies, to provide a forum for the advocacy of policies leading to robust, equitable and environmentally sustainable growth, and to promote capacity strengthening and training in the field of economic and social policy making. Its work is carried out by staff researchers and visiting scholars in Helsinki and through networks of collaborating scholars and institutions around the world.

UNU World Institute for Development Economics Research (UNU/WIDER) Katajanokanlaituri 6 B, 00160 Helsinki, Finland Camera-ready typescript prepared by Liisa Roponen at UNU/WIDER Printed at UNU/WIDER, Helsinki The views expressed in this publication are those of the author(s). Publication does not imply endorsement by the Institute or the United Nations University, nor by the programme/project sponsors, of any of the views expressed. ISSN 1609-5774 ISBN 92-9190-054-0 (printed publication) ISBN 92-9190-055-9 (internet publication)

Introduction

During the three decades beginning in the 1950s, deficits in the current account were considered normal. Countries were encouraged to borrow abroad and create an environment conducive to foreign investment to boost their economic growth. In the process, little attention was paid to the liabilities side of the current account deficit which increased the external indebtedness of these countries, until when Mexico, despite being an oil exporter, declared in August 1982, that it could not service its debts. Ever since, the issue of external debt and its servicing has assumed critical importance and introduced the debt crisis debate. External indebtedness is not harmful per se. Nor does heavy external debt automatically imply that growth must necessarily be low. What is detrimental for many African countries is their inability to meet current debt obligationscompounded by the lack of information on the nature, structure and magnitude of the external debt. A country may be able to export enough to generate the foreign exchange needed to buy the increasing imports associated with rapid growth and still service a high level of debt. Or it may be able to generate the necessary foreign exchange by borrowing more. But the concept of solvency implies that this is a process that cannot go on forever (Williamson 1996). Most of the countries classified as heavily-indebted poor countries (HIPCs) not only face solvency problems, but also face liquidity problems. It has become widely accepted that the heavily-indebted countries, particularly in Sub-Saharan Africa (SSA), require debt relief initiatives beyond mere rescheduling to have a turn-around in their economic performance and fight against poverty. In the late 1990s, this understanding appears to have stirred the international community to consider deeper, broader and faster external debt reliefthe HIPC debt initiative. Eligibility has been based on a good track record of reforms, pursuance of sound policies and ability to translate the resources into better prospects for the poor (IMF 2001a and IMF 2001b). By the end of June 2001, 23 countries, 19 of them in Africa, had benefited from debt service relief amounting to some US$ 34 billion (IMF 2001b). That notwithstanding, doubts are still being cast about the ability of the initiative to solve Africas immense debt problems. Despite the relatively high level of Kenyas external indebtedness, the country has not been included in the list of beneficiaries. Although it is stated that Kenya is expected to reach sustainable levels of debt without special help from the initiative (IMF 2001a), this is unlikely to happen, given the countrys current economic situation. While the country is grappling with high poverty levels (with 56 per cent of the population living below the poverty line), economic performance continues to deteriorate. For the first time since independence, the country recorded a negative growth of 0.3 per cent in gross domestic product (GDP) in 2000. In fact, the exclusion of Kenya from the HIPC debt initiative is likely to have been partly based on its poor record of reforms and economic performance rather than its ability to attain sustainable levels of external debt. Kenyas external debt indicatorsdebt-to-GDP ratio and debt-toexports ratiohave risen from an average of 38.5 per cent and 121.1 per cent for the 1970-80 period to 89.2 per cent and 268.2 per cent for 1991-99 period, respectively. Meanwhile, there have been significant net outflows since 1991 to service the debt obligations. This implies that

Kenya has been paying out more funds than it receives, thereby reducing domestic resources available for development. At the same time, there are very limited options for government to co-finance development activities through domestic debt. Although domestic debt constitutes less than a third of the total formal debt, it is almost ten times as expensive as external debt (GoK 1997). In view of the current economic situation and progress in relief initiatives, this paper examines the structure of Kenyas external indebtedness and its implications on economic growth. The paper is organized as follows. Section 2 provides the background information on Kenyas economic structure and performance since independence. Section 2.1 analyses the debt profile and the magnitude of external debt in Kenya. Section 2.2 examines the structure and composition of external debt, while Section 2.3 describes the determinants of the debt. Section 3 provides a synthesis of the theoretical and empirical literature on external debt and growth. Model specification and time series properties of the data used are presented in section 4. This is followed by the empirical results in section 5. Section 6 provides the conclusions and policy implications of the findings.

Background information

The period from independence (1963) to 1973 is a period when Kenyas economic growth was most rapid. GDP grew at an average of 6.5 per cent and per capita income remained positive despite high population growth rates. An examination of the commonly used debt burden indicators in Table 1 shows that the debt servicing ratio (debt service payments as a ratio of total export of goods and services) in the early 1970s was too low to cause concern when compared with that of African countries as a whole which was over 10 per cent in 1972. However the oil crisis of 1973/74 created severe balance of payments (BOP) problems and changed the picture. To meet the BOP crisis, the government resorted to heavy external borrowing. The external debt stock grew by 45.3 per cent in 1973 from the previous year (see Table 2). The growth rate decelerated to less than 4 per cent, being only 2.9 per cent in 1975 (see Table 1). The coffee boom of 1977 led to an abrupt increase in export earnings and therefore a temporary drop in the debt-servicing ratio in 1978. The coffee boom was, however, followed by the second oil crisis and a sharp deterioration in world commodity markets. As Kenyas export earnings almost stagnated, the debt-servicing ratio began to explode. This was accompanied by rising debt to GNP ratio as seen from Table 1. Drought conditions in the 1980s led to food imports, made possible by the availability of external loan finance. Increased interest rates on international loans raised the debt service charges substantially. This led to a decrease in net transfers on debt, being negative in 1981, 1984, and 1986 and have remained negative since 1991 (see Table 2). This transfer of capital to foreign creditors poses serious implications on the economy.

Table 1 Kenyas debt burden indicators and real GDP growth rate (per cent) Year 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 EDT/XGS 63 94 100 120 119 128 131 104 141 167 165 180 207 238 211 260 242 333 307 305 316 338 321 305 269 249 228 220 240 244 EDT/GNP 31 29 28 35 40 41 45 39 43 45 48 49 55 63 59 71 66 76 71 74 87 98 91 156 107 85 77 64 62 63 TDS/XGS 5 10 8 9 10 15 15 21 14 18 21 27 31 34 35 39 36 40 39 37 35 33 31 27 33 30 28 22 21 27 INT/XGS 2 3 4 4 4 5 4 4 5 7 11 13 14 14 14 15 14 17 17 15 15 15 12 11 12 10 9 7 7 7 INT/GNP 1 1 1 1 1 2 2 1 2 2 3 4 4 4 4 4 4 4 4 4 4 4 3 6 5 3 3 2 2 2 GDP growth rate 6.2 4.9 6.4 4.0 3.1 2.9 4.4 8.1 7.7 4.9 3.9 6.0 3.4 3.0 0.4 5.1 5.5 4.9 5.2 5.1 4.2 2.1 0.5 0.3 3.0 4.9 4.6 2.4 1.8 1.4

Source: World Bank (Global Development Finance [CD] 2001) and series of economic surveys.

Table 1 also shows that three of the four key indicatorsdebt-to-GNP ratio (50 per cent), debtto-exports ratio (275 per cent), debt service ratio (30 per cent) and interest-to-export ratio (20 per cent)have been above the critical levels (the numbers in brackets)1 since early 1980s. The debt service and the debt-to-GNP ratios have been above the critical levels since 1982, except in the late 1990s where the debt service ratio showed a slight decline. The debt-to1 See World Bank (World Debt Tables, 1989-90, vol.1).

exports ratio has remained above the critical level from 1987 to 1994. These indicators show that the external debt problem began to rise faster in the early 1980s. The GDP growth rate remained below 5 per cent in the early 1980s. Similarly, the significant rise in the debt burden indicators (debt-to-exports and debt-to-GNP ratios) in the first half of the 1990s also coincides with a deterioration in GDP growth rates during that period. The highest debt-to-exports and debt-to-GNP ratios of 338 per cent and 156 per cent were attained in 1991 and 1993, respectively. Concurrently, the GDP growth rate declined from 4.2 per cent in 1990 to 2.1 per cent in 1991, before declining further to 0.5 per cent and 0.3 per cent in 1992 and 1993, respectively. The growth performance has remained depressed to the new millennium. The relationship between GDP growth rate and external debt as a percentage of GDP over time is illustrated in Figure 1. The figure shows that while GDP growth rate (GRATE) depicts a consistently declining (linear) trend, external debt (EDGDP) trend has been rising over time. Moreover, the latter has remained above the trend of GDP growth rate since the mid-1980s. Based on this background, the severity of Kenyas external debt crisis cannot be underestimated. Although the debt burden indicators show a declining trend in the late 1990s, the huge transfers are alarming. The resources that could have been allocated to consumption and investment are instead being channelled abroad. This may act as a strong disincentive not only to invest but also to partake in any adjustment programmes aimed at increasing growth.

Figure 1 Trends of GDP growth rate and external debt as a percentage of GDP
9 8 7
GDP growth

140.00 120.00 100.00 80.00 60.00 40.00 20.00 0.00

19 72 19 74 19 76 19 78 19 80 19 82 19 84 19 86 19 88 19 90 19 92 19 94 19 96 19 98
debt as % of GDP

6 5 4 3 2 1 0
19 70

GRATE

EDGDP

Linear (EDGDP)

Linear (GRATE)

2.1 Size and magnitude of external debt in Kenya Table 2 shows the size of Kenyas stock of external debt and debt service payments for the period 1970-99. The total nominal debt stock rose from US$ 477.5 million in 1970 to US$ 7,412.4 million in 1995 while total debt service payments rose from US$ 50 million to US$ 901.4 million in the same period. However, the stock of debt and debt service payments have now declined to US$ 6,561.5 million and US$ 716.0 million, respectively in 1999. As seen from Table 2, the growth in debt stock shows a slight decline in 1990s.
Table 2 Kenyas external debt stock, debt service and net transfers on debt (million US $) Net transfers on debt Principal arrears on long-term debt Growth in debt Concessional debt/total debt (in %) 0.3 0 0 0 0 0 0.6 1.1 1.7 2.3 3.3 3.8 4 4 4 4.1 6.2 12.6 25.6 49.4 71.8 155 263.2 409.8 9.3 6.1 29.3 76.1 174.3 237.6 4.3 16.7 45.3 36.5 11.9 15.7 11.1 31.0 25.2 24.5 -4.7 4.3 7.7 -3.2 19.1 10.1 25.6 0.4 1.4 19.8 5.6 -7.4 3.1 1.3 2.9 -6.5 -4.7 5.2 -5.5 35.6 36.1 34.5 31.4 27.7 25.5 25.0 25.1 19.9 20.1 20.2 23.6 27.4 26.4 30.6 30.3 33.6 32.4 35.0 33.7 33.7 36.4 40.4 43.4 47.1 52.3 56.8 57.0 59.4 63.1

Year

External debt Debt service

(in million US$) 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 477.5 497.9 581.2 844.7 1,152.7 1,290.2 1,493.3 1,658.9 2,173.7 2,721.0 3,386.8 3,228.2 3,367.8 3,628.3 3,511.5 4,181.3 4,603.6 5,783.7 5,809.7 5,890.1 7,058.1 7,452.9 6,898.1 7,111.3 7,202.3 7,412.4 6,931.0 6,602.8 6,943.3 6,561.5 50.0 52.4 48.3 65.2 97.6 151.0 169.3 326.0 215.7 299.3 433.5 485.0 496.9 515.0 578.7 621.2 677.3 691.4 737.6 708.8 790.9 719.4 669.9 631.5 880.8 901.4 844.4 669.1 611.7 716.0 97.1 16.5 78.5 194.1 271.5 138.5 198.6 85.1 469.3 403.9 489.2 -214.4 11.7 194.3 -92.1 41.4 -252.1 176.8 95 287.4 281.2 -40.4 -146.5 -111.4 -651.4 -197.5 -462 -199.1 -373.7 -500.9

Source: World Bank (Global Development Finance [CD] 2001).

Table 2 shows that a significant rise in Kenyas external indebtedness corresponds to the periods 1973-74, 1978-80, 1985-87 and 1990 in which increased borrowing was made. The first and second periods coincided with the first and second oil crises, respectively. These periods also registered a significant growth in debt service payments. The third period coincided with the funding of structural adjustment programmes by International Monetary Fund (IMF) and World Bank. The decline in growth of external debt in 1988 and 1989 is partly due to debt write-offs and a decline in bilateral and private debt. In 1989, Kenya was forgiven its external debt amounting to US$ 463 million. The decline in the 1990s can be attributed partly to the negative netrepayments and aid embargos resulting in no new external debt contracts. For instance, the twoyear aid freeze in official capital inflow in 1991 and 1992 resulted in an increase in Kenyas external payment arrears. In addition, there was a heavy reliance on domestic borrowing in relation to external borrowing in the 1990s. A relatively tight fiscal stance was also witnessed during the period. Despite the magnitude of external debt in the 1980s, Kenya was able to service its debts without rescheduling. This is also reflected by the fact that there was zero or negligible accumulation of arrears in 1970s and a better part of 1980s (see Table 2). However, by early 1990s, the debt burden became so acute that Kenya had to reschedule its debt in 1994 for the first time. With the curtailment of donor funding, the government began to accumulate arrears on official debt. Table 2 shows that there was a significant accumulation of arrears in the early and late 1990s. Although there was a dramatic build-up in nominal aid flows during the 1980s, external financial support has been slackening in the 1990s. Consequently, the level of external indebtedness has been falling. Although Kenya may not be as heavily indebted as other HIPCs, its present poor economic performance and inability to meet its debt obligations have serious implications on development and debt sustainable objectives. 2.2 Structure, type and composition of external debt Like most low-income SSA countries, a greater proportion of Kenyas external debt consists of official debts. A decomposition of official debt shows that in 1970s, official debt was mainly from bilateral sources. From early 1980s onwards, however, multilateral debt constitutes a major proportion of total debt stock. The share of multilateral debt increased moderately in 1980s mainly as a result of large disbursements of adjustment lending from the World Bank (OBrien and Ryan 1999). Since the early 1990s, the proportion of concessional debt has been rising. The proportion of concessional debt rose from 20 per cent in 1979 to 34 per cent in 1989 and to 63 per cent in 1999, respectively (see Table 2). This has given Kenya the advantage of contracting loans on soft terms. This also explains why the interest to exports and interest to GNP ratios have remained relatively low. Private debt has remained relatively low over the years. Such loans are normally obtained on hard terms and conditions implying high debt service payments.

Long-term debt constitutes a major proportion of total debt outstanding. A greater proportion of long-term debt outstanding is contracted by the public sector. Private non-guaranteed debt has not only been low, but has also been rising relatively slowly over the years. Short-term debt has also remained relatively low over the years. 2.3 Determinants of external debt The debt crisis has evolved from a complex combination of factors, some of which are external while others are the direct result of economic policies pursued. However, Ajayi (1991) has argued that the division of the factors into external and domestic is not correct because external factors impinge crucially on what happens domestically and vice versa. That notwithstanding, determinants of Kenyas external debts can be attributed to both internal and external factors. The major external factors include the following. The 1973/74 oil price increases led to a deterioration in terms of trade leading to BOP deficits. The oil shock also contributed to a tremendous increase in the availability of international credit at very low interest rates and encouraged oil-importing developing countries (including Kenya) to borrow (Sachs and Larrain 1995). Many creditors overstated the potential capabilities of the debtor countries to meaningfully absorb and pay for debts (Ajayi 1991). Similarly, the second oil crisis, which also coincided with the 1979/80 drought, led to increased borrowing. In the early 1980s the world interest rates increased sharply as a consequence of anti-inflationary programmes in the industrialized countries. At the same time, the terms of trade deteriorated for the debtor world as raw material prices fell. Kenyas growth of export earnings declined tremendously from 26 per cent in 1980 to about 13 per cent in 1981. The external balance continuously deteriorated in the early 1980s and was further worsened by the drought of 1983/84 and the early 1990s. Increased protectionism policies by developed countries have tended to discriminate against less developed countries (LDCs) exports (including Kenya), thus lowering their earnings. Besides external factors, Kenyas external indebtedness can be partly attributed to internal factors. These mainly refer to the overly expansionary fiscal policies and highly distorted trade policies, especially policies that created a heavy bias against exports. Public sector deficits have been a major problem since late 1970s. Following the 1977 coffee boom, the initial response was to expand public expenditure and, since revenue from taxation did not rise as fast, the government resorted to foreign borrowing. When commodity prices later fell, expenditures were not reduced accordingly and previous borrowing was supplemented with new borrowing to maintain expenditure levels. By early 1980s, the public sector was overextended. The overextension of the public sector showed up in economic inefficiency.

Other factors were an overvalued exchange rate,2 negative real interest rates as well as an import-substituting industrial strategy, which was characterized by overprotection. The greater pervasiveness of the import licensing system and regulations on business activities created enormous opportunities for rent-seeking and for executive discretion.

Debt and economic performance

3.1 Theoretical perspectives The question that needs to be answered is whether the large debt burden in HIPCs is one of the factors contributing to the weak economic performance and the uneven pace of economic reform in these countries, particularly in SSA. There are two competing hypotheses: the debt overhang hypothesis and the liquidity constraint hypothesis. The debt overhang theory is based on the premise that if debt will exceed the countrys repayment ability with some probability in the future, expected debt service is likely to be an increasing function of the countrys output level. Thus some of the returns from investing in the domestic economy are effectively taxed away by existing foreign creditors and investment by domestic and new foreign investors is discouraged (Claessens et al. 1996: 17). Under such circumstances, the debtor country shares only partially in any increase in output and exports because a fraction of that increase will be used to service the external debt. The theory implies that debt reduction will lead to increased investment and repayment capacity and, as a result, the portion of the debt outstanding becomes more likely to be repaid. When this effect is strong, the debtor is said to be on the wrong side side of the debt Laffer curve. In this case, the debt Laffer curve refers to the relationship between the amount of debt repayment and the size of debt.3 However, the idea of debt Laffer curve also implies that there is a limit at which debt accumulation stimulates growth (Elbadawi et al. 1996). In reference to an aid Laffer curve, Lensink and White (1999) argue that there is a threshold at which more aid is detrimental to growth. The liquidity constraint is captured as a crowding out effect, by which the requirement to service debt reduces funds available for investment and growth. A reduction in the current debt service should, therefore, lead to an increase in current investment for any given level of future indebtedness (Cohen 1993). Other channels through which the need to service a large amount of external obligations can affect economic performance include lack of access to international
2 An overvalued currency reduces the price of imports and thus worsens the BOP, leading to higher capital inflows. It can also raise expectations for devaluation, leading to capital flight. Currency depreciation can raise the stock of external debt. 3 For detailed information about debt overhang and debt laffer curve, see Claessens (1990), Cohen (1989) and (1993).

financial markets and the effects of the stock of debt on the general level of uncertainty in the economy (Claessens et al. 1996). The scope of debt overhang is much wider in that the effects of debt do not only affect investment in physical capital but any activity that involves incurring costs up-front for the sake of increased output in the future. Such activities include investment in human capital (in terms of education and health) and in technology acquisition whose effects on growth may be even stronger over time. 3.2 An overview of empirical studies There have been some limited attempts to empirically assess the debt overhang and crowding-out effects. Most of the empirical studies include a fairly standard set of domestic, debt, policy and other exogenous explanatory variables. The majority find one or more debt variables to be significantly and negatively correlated with investment or growth (depending on the focus of the study). For instance, Borensztein (1990) found that debt overhang had an adverse effect on private investment in Philippines. Iyoha (1996) found similar results for SSA countries. He concluded that heavy debt burden acts to reduce investment through both the debt overhang and the crowding out effect. However, Cohens (1993) results on the correlation between developing countries (LDCs) debt and investment in the 1980s showed that the level of stock of debt does not appear to have much power to explain the slowdown of investment in developing countries during the 1980s. It is the actual flows of net transfers that matter. He found that the actual service of debt crowded out investment. Elbadawi et al. (1996) also confirmed a debt overhang effect on economic growth using crosssection regression for 99 developing countries spanning SSA, Latin America, Asia and Middle East. They identified three direct channels in which indebtedness in SSA works against growth: current debt inflows as a ratio of GDP (which should stimulate growth), past debt accumulation (capturing debt overhang) and debt service ratio. The fourth indirect channel works through the impacts of the above channels on public sector expenditures. They found that debt accumulation deters growth while debt stock spurs growth. Using data for Cameroon, Mbanga and Sikod (2001) found that there exist a debt overhang and crowding-out effects on private and public investments, respectively. Other studies that have found a negative effect of external debt on growth include Degefe (1992). Some studies simply use simulation analysis to show the impact of the debt burden indicators on economic growth under different scenarios (e.g. Ajayi 1991; Osei 1995 and Mbire and Atingi 1997). Generally, most studies tend to confirm debt overhang/crowding-out effects. Nonetheless, the empirical literature is mainly focused on evaluating the impact of external debt on investments rather than on economic growth per se. Moreover, these studies are mainly based on data across countries in disregard to each countrys uniqueness. While the findings are quite revealing, there is need for case-by-case studies in view of each countrys unique characteristics. This is particularly important given the stringent conditionalities for debt relief initiatives. In addition, most empirical studies suffer from methodological limitations in the sense that there is a

tendency to ignore the non-stationarity of time series data. This is despite the fact that working with non-stationary variables lead to spurious regression results (Charemza and Deadman 1992). That notwithstanding, very few empirical studies on Kenyas external debt exist and, even then, they do not focus on the analysis of external debt and economic growth (for example: Ochieng 1991; Manundu 1984; Gulamhussein 1987; and Ngeno 1991). This paper is motivated on the basis of these factors. The paper follows the approach used by similar studies (specifically Elbadawi et al. 1996), but adopts an error correction formulation in assessing the debt overhang and crowding-out effects in Kenya.

Model specification

It has been argued that expansion of investment facilitates economic growth, depending on the quality of investment. Consequently, it is important to evaluate the impact of external indebtedness on investment. However, as argued by Fosu (1996, 1999) debt can also affect economic growth directly through its effects on the productivity of investment. The existence of a debt overhang would influence the nature of investment mix, with a bias towards short-term investment projects with faster returns that are less productive. In other words, external debt may still affect output growth even if investment levels remain unaffected (ibid.). Therefore, estimating only the investment equation would underestimate the effect of external indebtedness on economic growth. From the literature, the channels through which the indebtedness affects growth have been identified as: stock of external debt as a ratio of GDP which should stimulate growth; past debt accumulation (lagged debt-GDP ratio) which impacts negatively on growth; the debt-service ratio, which captures the crowding out effect. The final indirect channel incorporated in this paper is net foreign financing as a proportion of the total deficit. As the stock of debt and cost of external debt servicing rise, there is little left to finance public development projects and social services. This leads to severely compressed budgets and/or fiscal deficits. These fiscal deficits aggravate further external borrowing as a source of financing the deficits. Besides these variables, the model also incorporates other policy, fundamental and shock variables. The model adopted is based on Elbadawis et al. (1996) model. Like in similar studies, the debt burden indicators enter the production function directly. 4.1 Growth equation
GRATE = (EDGDP, EDGDPt-1, DSR, FFDC, PINV, TOT, RER, SER, INFL, GPUIVt-1) (1)

Where GRATE EDGDP

= =

real GDP growth rate; stock of external debt to GDP ratio (-); stock of external debt to GDP ratio lagged by one period (reflect debt accumulation) (-);

EDGDPt-1 =

10

DSR FFDC PINV TOT SER INFL RER GPUIV

= = = = = = = =

the debt service as a ratio of export earnings (reflect the crowding-out effect) (-); net foreign financing as a proportion of total deficit (-); private investment as a ratio of GDP (captures the accelerator principle) (+); terms of trade (captures external shocks) (- or +); primary school enrolment rate as a proxy for human capital development (+); rate of inflation (reflects macro- economic stability) (- or +); movements in real exchange rate (reflects incredibility of policies) (-); Public investment as a ratio of GDP (+).

Private and government investments are included separately in the model. The former captures the accelerator principle. In other studies, total investment (with the two combined) is used. The shock variable is captured by terms of trade. The macroeconomic policy variables are captured by inflation rate and by real exchange rate. These variables show the extent of vulnerability of the economy to external factors and consequently to reliance on foreign resource financing. They also show the extent of credibility of policies and their effect on economic growth. In contrast to Elbadawi et al. (1996), the current study assumes that the role of human capital development is more important in explaining growth than just population growth. This is captured by primary school enrolment rate. 4.2 Investment equation Interest rate as a variable is included in the investment equation to capture its effect on private investment. The investment equation is specified as:
PINV = f(EDGDPt-1, EDGDP, GRATE, DSR, FFDC, TOT, SER, GPUIV, INT,INFL, RER) (2)

where INT = Interest rate (treasury bill rate). The other variables are as defined before. Before estimating the above equations, a test for simultaneity between investment and growth equations was carried out to ascertain whether simultaneous equation-bias exists. The test showed that only weak simultaneity exists. This was handled by normalizing the variables through the error correction process. Therefore the two equations are estimated independently, each incorporating an error correction term.

11

Data, time series properties and empirical results

The time series data used in the estimation covers 1970-95. For consistency and ease of comparison, all the data (except GDP growth rates, investment and school enrolment rate) were drawn from international publicationsthe Global Development Finance and International Finance Statistics.4 Data from government publications were drawn from Statistical Abstracts and Economic Surveys. The results of data analysis and estimation were obtained using the Generalized Instrumental Variables Estimators econometrics computer package (PCGIVE, version 8.0). 5.1 Time series properties Non-stationarity of time series data has often been regarded as a problem in empirical analysis. Working with non-stationary variables leads to spurious regression results from which further inference is meaningless. The conventional Dickey-Fuller (DF) and Augmented Dickey-Fuller (ADF) tests were used to test for stationarity of the series. The results of the test for the variables in levels are presented in Table 3. The tests show that the variables SER, GRATE, INFL, FFDC and PINV are stationary (integrated of order zero) at 5 per cent level of significance. A graphical analysis of GPUIV showed stationarity. The rest of the variables were found to be stationary after differencing them once. These variables are therefore integrated of order one (I I(1)). The next step was to establish whether the non-stationary variables are cointegrated. Differencing of variables to achieve stationarity leads to loss of long-run properties. The concept of cointegration implies that if there is a long-run relationship between two or more non-stationary variables, deviations from this long-run path are stationary. To establish this, the Engel-Granger two-step procedure was used. This was done by generating residuals from the long-run equation of the non-stationary variables, which were then tested for stationarity using the DF and ADF tests. The residuals were found to be stationary at 5 per cent level of significance for both tests. Consequently, an error correction formulation was adopted. The next step was to establish whether the non-stationary variables are cointegrated. Differencing of variables to achieve stationarity leads to loss of long-run properties. The concept of cointegration implies that if there is a long-run relationship between two or more non-stationary variables, deviations from this long-run path are stationary. To establish this, the Engel-Granger two step procedure was used. This was done by generating residuals from the long-run equation of the non-stationary variables, which were then tested for stationarity using the DF and ADF tests. The residuals were found to be stationary at 5 per cent level of significance for both tests. Consequently, an error correction formulation was adopted.

4 There is substantial variation between data (on external debt and debt-related figures) from international and locally available publications.

12

Table 3 Unit root tests (variables in levels) Variable GRATE EDGDP TOT DSR INT GPUIV PINV RER SER FFDC INFL Critical values at 5 per cent Note: DF -2.816 -1.419 -2.471 -1.809 -2.461 -2.270 -3.485 0.6729 -4.332 -2.06 -2.604 -2.985 *The number in parenthesis refers to number of lags ADF (2)* -3.631 -1.133 -2.319 -0.3838 -1.795 -2.030 -2.030 0.4319 -4.364 -4.249 -3.433 -3.622 Order of integration I (0) I (1) I (1) I (1) I (1) I (1) I (0) I (1) I (0) I (0) I (0)

Both the investment and growth equations were re-specified to include the error correction term as follows:
GRATE = a0 +a1DEDGDP + a2DDSR + a3FFDC + a4 PINV +a5DTOT + a6SER +a7INFL + a8GPUIV + a9DRER + a10ECTt-1 + e1t

(3)

Where ECT = EDT - co - c1DSR + c2TOT - c3RER.

PINV = bo + b1DEDGDP + b2DDSR + b3FFDC + b4DTOT + b5DINT + b6SER + b7DRER + b8INFL + b9GRATE + b10GPUIV + b11ECTt-1 + e2t

(4)

ECT is the error correction term. This term captures the long-run relationship. It reflects attempts to correct deviations from the long-run equilibrium path and its coefficient can be interpreted as the speed of adjustment or the amount of disequilibrium transmitted each period to economic growth. The cs can be interpreted as parameters of equilibrium relationship about which economic theory is informative (Ndungu 1993). 5.2 Empirical results Equations 3 and 4 with their associated lags were estimated using OLS. Before running the regressions, a simple correlation analysis between the dependent and the explanatory variables was carried out. The results are reported in Table 4. 5.2.1 Correlation results Table 4 shows that there is relatively high negative correlation between economic growth (GRATE) and external debt (EDGDP) and between private investment (PINV) and external

13

debt. There is negative correlation between private investment, interest rate (INT) and real exchange rate (RER). The relationship between private investment and economic growth is relatively high and positive as would be expected. Terms of trade (TOT) have positive relationship with both growth and investment.
Table 4 Correlation matrix of variables in levels GRATE GRATE PINV 1.000 PINV 0.613 1.000 EDGDP -0.560 -0.556 DSR -0.293 -0.271 TOT 0.518 0.421 INFL -0.592 -0.380 INT -0.585 -0.514 GPUIV 0.328 0.318 RER -0.49 -0.60

5.2.2 Estimation results for the growth equation Using the general to specific estimation procedure, the preferred model for the growth equation is reported in Table 5. An impulse dummy for 1984 was added after critical analysis of residuals, which showed a shock to the system during that year. The diagnostic test outcomes are satisfactorythat is, AR for autocorrelated residuals, the ARCH for heteroscedastic errors, normality test for the distribution of the residuals and the RESET test for the regression specification. Most of the variables considered in the determination of economic growth have their hypothesized signs. The coefficient of current debt flows (DEDGDP) was expected to be positive but it is negative. A rise in current debt flows as a ratio of GDP leads to a decline in economic growth. This implies that even current debt flows deter economic growth in the short run. On the other hand, the coefficient of past debt accumulation (debt lagged twice) is negative as expected. These results confirm the existence of debt overhang problem as earlier postulated. The results tally with the findings of similar studies (e.g. Elbadawi et al. 1996). The positive effect of debt service ratio on economic growth was unexpected. However, it has been argued that actual debt service payments are inadequate indicators of the debt burden (Fosu 1999). Nevertheless, the debt service ratio for Kenya has not been overly high compared with other low-income countries. The impact of the other variables is as follows. Net foreign financing of deficit as a ratio of fiscal deficit (lagged once) has a negative impact on economic growth as postulated. Private investment as a ratio of GDP as well as investment in human capital development, proxied by primary school enrolment rate, positively influence growth. However, the coefficient of the latter is negligible when compared with the impact of private investment on growth. Similarly, current public investment promotes economic growth but unexpectedly, past public investment (GPUIVt-1) was found to have a negative impact on economic growth. Current inflation rate deters economic growth while past inflation rate (INFLt-1) stimulates economic growth. The impact of real exchange rate on growth was found to be highly statistically insignificant and the

14

variable was thus dropped from the model. A dummy to capture shocks in 1984 was found to have had a negative impact on growth. This is due to the severe drought that occurred during that year, leading to massive food imports.
Table 5 Regression results for the growth equation Variable Constant INFL INFLt-1 GPUIV GPUIVt-1 PINV PINVt-1 FFDCt-1 SER DEDGDP DEDGDPt-2 DDSR DDSRt-1 DTOT DTOTt-1 ECTt-1 D1984 Note:
2

Coefficient -3.996 -0.053 0.101 0.727 -0.514 0.303 0.285 -0.045 0.039 -0.055 -0.064 0.090 0.085 0.035 0.049 -0.308 -2.966

Std. error 0.9018 0.013 0.016 0.100 0.107 0.036 0.032 0.007 0.008 0.014 -0.011 0.026 0.028 0.004 0.004 0.050 0.407

t-value -4.433 -4.093 6.285 7.283 -4.809 8.417 8.866 -6.465 4.881 -3.805 -5.790 3.399 3.056 8.563 11.757 -6.158 -7.284

D before the variable symbol implies first difference of the variable.

R = 0.996 F[16,6] = 88.748[0.000] = 0.2514 n =23 DW = 2.45 AR 1 -2F(2,4)

2

= 2.5724[0.1913]

ARCH 1 F(1,4) = 0.17225[0.6985] Normality chi (2) = 2.3681[0.3060] RESET F(1,5) = 0.29745[0.6089]

The lagged error correction term (ECTt-1) included in the model to capture the long-run dynamics between the cointegrating series is correctly signed (negative) and statistically significant. The coefficient indicates a speed of adjustment of 31 per cent from actual growth in the previous year to equilibrium rate of economic growth. The relatively low speed of adjustment implies that all errors/deviations are not corrected within one year and most of the time the economy is operating out of equilibrium. A further discussion of what ECT entails is worthwhile as it reveals long-run relationships of the non-stationary variables. ECT is specified as: ECT = EDGDP - 42.960 - 0.386DSR + 0.095TOT - 1.73RER.

15

It can be seen that although debt service ratio had unexpected positive sign, it enters the error correction term with the expected sign. Debt service rises with growth in external debt. Unfavourable terms of trade prompt increased external borrowing. Estimation results for the investment equation
Table 6 Regression results for the investment equation Variable Constant INFLt-1 DEDGDP DEDGDPt-1 SER DINT FFDCt-1 DDSR DDSR t-1 DRER ECTt-1 GPUIV
2

Coefficient 11.139 -0.278 0.361 -0.203 0.060 -0.492 0.057 -0.245 0.366 0.521 0.936 0.335

Std. error 3.696 0.074 0.093 0.066 0.044 0.144 0.032 0.150 0.158 0.372 0.220 0.426

t-value 3.014 -3.737 3.871 -3.068 1.351 -3.421 1.811 -1.632 2.320 1.401 4.264 0.786

R = 0.849 F(11,12) = 6.1471[0.0020] = 1.8098 n=24 AR 1 -2F(2,10) = 1.5035[0.2686] ARCH 1 F(1,10) = 0.49461[0.4979] Normality Chi (2) = 3.4464[0.1785] RESET F(1,11) = 0.93786[0.3536]
2

The results in Table 6 indicate that current debt flows (DEDGDP) stimulate private investment while past debt flows (DEDGPDt-1) deter investment. These results are as expected. Foreign borrowing can help relieve resource shortages and if used productively will stimulate investments and therefore promote growth in the short run. A rise in current debt service ratio negatively affects private investment. This confirms the crowding-out effect of debt service on private investment. Contrarily, variation in past debt service ratio (DDSR t-1) has a positive effect on private investment. The impact of other variables on private investment is as follows. Past inflation (INFLt-1) discourages current private investment. This could imply that economic agents expect the previous years high level of inflation rate to persist in the current period, thus discouraging current private investment. Similarly, high interest rates (DINTr) discourage private investment. Both public investment and human capital development proxied by primary school enrolment rate have positive but highly statistically insignificant coefficients. Similarly, the impact of a change in the real exchange rate (depreciation) on private investment is not significant. Unlike

16

in the growth equation, net foreign financing of the deficit as a ratio of total fiscal deficit has a positive and statistically significant effect on private investment. Unlike in the growth equation, the error correction term lagged once (ECTt-1) has a positive sign. This coincides with the hypothesis that current stock of debt is likely to stimulate private investment. The model reports a speed of adjustment of around 94 per cent, which is relatively high. This implies that the deviations/errors from the long-run equilibrium path are almost corrected in one period. Investors react fairly fast to changes in external resource inflows.

Conclusion and policy implications

The central focus of this study was to establish the impact of Kenyas external indebtedness on economic growth. The paper also examined the structure, magnitude, composition and determinants of Kenyas external debt. A greater proportion of Kenyas external debt consists of official debts mainly from multilateral sourcesmainly in the form of concessional loans. This has enabled the country to borrow on fairly soft terms. The causes of external debt in Kenya can be traced to both internal and external factors. Internal factors are mainly overly expansionary fiscal policies and highly distorted trade policies, especially policies that created a heavy bias against exports. The external factors include deterioration of terms of trade leading to BOP deficits, high world interest rates and increased protectionism by developed countries, which tended to discriminate against LDCs exports. In addition to these factors, drought conditions have also contributed to the external debt burden. Using an error correction formulation, the estimation results showed a possibility of a debt overhang problem in both the growth and investment equations. Not only does past debt accumulation deter growth, but so do current debt flows in the short run. In the private investment equation, it was however found that current debt flows stimulate investment while debt accumulation deters investment. In general, the results seem to support the argument that external debt still affect growth directly even if investment is not adversely affected. The investment model also showed that there was some crowding out of current investment as a result of servicing relatively large amounts of external debt. In contrast, debt servicing does not appear to have a direct negative impact on economic growth. Arguably, the debt service ratios for Kenya have been considered to be relatively lower compared to other low-income HIPCs. As indicated earlier, past attempts to resolve SSAs external debt crisis have not borne much fruit. The only hope now lies in the current HIPC debt Initiative, which is expected to relieve the HIPCs of their debt burdens, with positive implications on poverty. In as much as external debt burden is a reality in Kenya, it is also true that the country cannot achieve its goal of becoming an industrialized nation by the year 2020 without external financial assistance. In view of the current economic recession and the negative net outflows, the results obtained from

17

this study support the need for Kenya to be considered for comprehensive debt relief measures. The simultaneous attainment of sustainable economic growth and external debts appear difficult at the moment and could remain elusive if aggressive measures are not undertaken. The government could play an important role in stimulating the economy if the resources obtained from the debt relief initiatives are targeted at productive public investments with the resultant crowding-in effects on private investment, and social spending for the poor. Being the key source of indebtedness, the challenge to the government remains that of ensuring efficiency in delivery of services and increased productivity of public investments. Efforts are already being directed towards this direction through privatization of parastatals and downsizing of the civil service, but more has to be done to revamp the economy to a higher, sustainable growth path. In the long run, foreign savings should supplement but not replace domestic savings. Proper macroeconomic management of the economy as a whole is important since it also determines the volume and servicing of external debt, as well as the credit rating. Availability of external finance should be consistent with a policy framework that is credibly maintained (fiscal stance, exchange rate policy, interest rate policy, pricing policy, etc.). It is important to create credibility including political will in order to spur investor confidence for both local and foreign investments. Commitment to re-building credibility is a key challenge for Kenya. Development activities could also be financed through increased export earnings spearheaded by an export-led growth strategy. As part of broader strategy to assist the HIPCs out the debt crisis, the international community should provide a conducive environment for exports from the low-income countries including Kenya. Kenya still has a chance of overcoming its external debt problems by cultivating the right policies, but will need considerable support through debt relief/reduction initiatives.

18

Bibliography Ajayi, S. I. (1991). Macroeconomic Approach to External Debt: The Case of Nigeria. AERC Research Paper No. 8. Nairobi: African Economic Research Consortium. Aliber, R. Z. (1980). A Conceptual Approach to the Analysis of External Debt of the Developing Countries. World Bank Staff Working Paper No. 421. Washington, DC: World Bank. Borensztein, E. (1990). Debt Overhang, Debt Reduction and Investment: The Case of the Philippines. International Monetary Fund Working Paper No. WP/90/77, September. Washington, DC: IMF. Charemza, W. W., and F. B. Deadman (1992). New Directions in Econometric Practise: General to Specific Modeling, Cointegration and Vector Autoregression. Cheltenham: Edward Elgar. Claessens, S. (1990). The Debt Laffer Curve: Some Empirical Estimates. World Development, 18 (12). Claessens, S., E. Detragiache, R. Kanbur, and P. Wickham (1996). Analytical Aspects of the Debt Problems of Heavily Indebted Poor Countries. Paper presented to IMF/World Bank seminar in December. Washington, DC: World Bank. Cohen, D. (1989). How to Cope with Debt Overhang: Cut Flows rather than Stocks, in I. Diwan and I. Husain (eds), Dealing with the Debt Crisis. Washington, DC: World Bank. Cohen, D. (1993). Low Investment and Large LDC Debt in the 1980s. The American Economic Review, June. Cline, W. R. (1996). Debt Relief for the Highly Indebted Poor Countries: Lessons from the 1980s Debt Crisis. Paper prepared at the IMF/World Bank conference on External Financing for Low-Income Countries, December. Washington, DC: IMF/World Bank. Degefe, B. (1992). Growth and Foreign Debt: The Ethiopian Experience: 1964-86. AERC Research Paper No. 13. Nairobi: African Economic Research Consortium. Diwan, I., and I. Husain (eds) (1989). Dealing with the Debt Crisis. Washington, DC: World Bank. Davis, M. J. (ed.) (1992). Macroeconomic Adjustment: Policy Instruments and Issues. Washington, DC: IMF. Elbadawi, A. I., J. B. Ndulu, and N. Ndung'u (1996). Debt Overhang and Economic Growth in Sub-Saharan Africa. Paper presented at the IMF/World Bank Conference on External Financing for Low-income Countries, December. Washington, DC: IMF/World Bank. Eaton, J. (1993). Sovereign Debt: A Primer. The World Bank Economic Review, 7 (2).

19

Fosu, A. K. (1996). The Impact of External Debt on Economic Growth in Sub-Saharan Africa. Journal of Economic Development, 12 (1). Fosu, A. K. (1999). The External Debt Burden and Economic Growth in the 1980s: Evidence from Sub-Saharan Africa. Canadian Journal of Development Studies, XX (2). GoK (1997). Public Expenditure Review. Nairobi: Public Expenditure Review Secretariat, Ministry of Finance. Mimeo. Greene, J. E., and S. M. Khan (1990). The African Debt Crisis. AERC Special Paper 3. Nairobi: African Economic Research Consortium. Gulamhussein, M. J. (1987). External Shocks, Adjustment and Debt Problem in Kenya, 197486. Nairobi: University of Nairobi. Masters dissertation. Government of Kenya (various years). Economic Surveys, various issues. Nairobi: Government Printer. Government of Kenya (various years). Statistical Abstracts, various issues. Nairobi: Government Printer. Hendry, F. D., and A. J. Doornik (1991). PcGive 8.0: An Interactive Econometric Modelling System. London: International Thomson Publishing. Iyoha, M. A. (1996). External Debt and Economic Growth in Sub-Saharan African Countries: An Econometric Study. Paper presented at AERC workshop, Nairobi. IMF (International Monetary Fund) (various years). International Financial Statistics Yearbook 1995, 1996. Washington, DC: IMF. IMF (International Monetary Fund) (2001a). Debt Relief under the Heavily Indebted Poor Countries (HIPC) Initiative. A factsheet, April. Washington, DC: IMF. IMF (International Monetary Fund) (2001b). 100 Percent Debt Cancellation? A Response from the IMF and the World Bank. A factsheet, July. Washington, DC: IMF. Krumm, K. L. (1985). The External Debt of Sub-Saharan Africa. World Bank Staff Working Paper No. 741, June. Washington, DC: World Bank. Klein, T. M. (1994). External Debt Management. World Bank Technical Paper No. 245. Washington, DC: World Bank. Lensink, R., and H. White (1999). Is there an Aid Laffer Curve? Credit Research Paper No. 99/6. Nottingham: Centre for Research in Economic Development and International Trade, University of Nottingham. Lipumba, N. H. I., and P. E. M. Noni (1990). Foreign Debt Management and Economic Growth in Tanzania. AERC Discussion Paper. Nairobi: African Economic Research Consortium. Manundu, M. (1984). Debt Management Strategies for Kenya. Summary of findings of a workshop organized by Friedreich Ebert Foundation in November.

20

Mbanga, G. N., and F. Sikod (2001). The Impact of Debt And Debt-Service Payments on Investment in Cameroon. Final report presented at the AERC Biannual Research Workshop, May 26 31. Nairobi: AERC. Mbire, B., and M. Atingi (1997). Growth and Foreign Debt: The Ugandan Experience. AERC Research Paper No. 66. Nairobi: African Economic Research Consortium. Mehran, H. (ed.) (1985). External Debt Management. Washington, DC: IMF. Ngeno, N. K. (1991). The External Debt Problem of Kenya. Paper presented at the AERC workshop in May. Nairobi: AERC. Ndungu, N. (1993). Dynamics of the Inflationary Process in Kenya. Gothenburg: University of Gothenburg. Ph.D. thesis. Ndungu, N. (1997). Price and Exchange Rate Dynamics in Kenya: An Empirical Investigation (1970-93). AERC Research Paper 58. Nairobi: AERC. Nyoni, S. T. (1997). Foreign Aid and Economic Performance in Tanzania. AECR Research Paper No. 61. Nairobi: African Economic Research Consortium. OBrien, F. S., and T. C. I. Ryan (1999). Aid and Reform in Africa: Kenya Case Study. Aid and Reform in Africa Papers. Washington, DC: The World Bank Group. Ochieng, J. S. (1991). Determinants of the External Debt Burden: An Analysis of Kenya, 1970-89. Nairobi: University of Nairobi. Masters degree dissertation. Osei, B. (1995). Ghana: The Burden of Debt Service Payment under Structural Adjustment. AERC Research Paper No. 33. Nairobi: African Economic Research Consortium. Okurut, N. F. (1997). The Impact of Foreign Aid Flows on Private Investment in Uganda. Paper presented at the AERC Research Workshop in May. Nairobi: AERC. Sachs, J. D., and F. B. Larrain (1995). Macroeconomics in the Global Economy. Englewood Cliffs, NJ: Prentice Hall. Svendsen, K. E. (1987). The Failure of the International Debt Strategy. Copenhagen: Centre for Development Research. Williamson, J. (1996). Constraints on Macroeconomic Management Imposed by Heavy External Debt. Paper presented at the IMF/World Bank seminar, December. Washington, DC: IMF/World Bank. Woodland, D. (1992). Debt, Adjustment and Poverty in Developing Countries. London: Printer Publisher. World Bank (various years). World Debt Tables, various issues. Washington, DC: World Bank. World Bank (1997). Global Development Finance, Country Tables. Washington, DC: World Bank.

21