Reviewing Two Decades of Debt Relief

WorldDevelopment Vol. 30, No. 10, pp. 1677–1696, 2002 Ó 2002 Elsevier Science Ltd. All rights reserved Printed in Great Britain www.elsevier.com/locate/worlddev 0305-750X/02/$ - see front matter PII: S0305-750X(02)00073-6 How Did Heavily Indebted Poor Countries Become Heavily Indebted? Reviewing Two Decades of Debt Relief

WILLIAMEASTERLY * Center for Global Development, Institute for International Economics, Washington, DC, USA Summary. — The paradox of debt is that heavily indebted poor countries (HIPCs) became heavily indebted after two decades of debt relief efforts. Average policies in HIPCs 1980–97 were worse than other less-developed countries (LDCs), controlling for income. Terms of trade and wars do not show a different trend in HIPCs than in non-HIPC LDCs. Financing HIPCs shifted away from private and bilateral nonconcessional sources toward International Development Assistance and other multilateral concessional financing––but this implicit form of debt relief also failed to reduce net present value debt. The record is not encouraging for the success of current debt relief efforts. Ó 2002 Elsevier Science Ltd. All rights reserved.

Key words — debt, Africa, adjustment, International Organizations, foreign aid, World Bank/IMF policies

1. INTRODUCTION debt servicing difficulties have been a feature of the world economy throughout history. 3 The central paradox of the heavily indebted The problems of the HIPCs are very much in poor countries (HIPCs) is that they became the news today (Third World debt was even heavily indebted after two decades of partial mentioned in the hit movie Notting Hill, star- debt relief and concessional lending. How did ring Hugh Grant and Julia Roberts.) A coali- this happen? This may suggest that the factors tion of nongovernmental organizations called that lead to high debt are long-lasting and not Jubilee 2000 asked for a write-off of all debt of easily changed by debt relief. Consider the poor countries on the occasion of the turning of following example. the millenium (Jubilee 2000). Support for Ju- The HIPC of Haiti is not growing. The ratio bilee 2000 has been expressed by such diverse of foreign debt service to exports has reached figures as Bono from the rock group U2, 40%, well above the 20–25% thought to be the Pope, Jeffrey Sachs, Muhammad Ali, ‘‘sustainable.’’ 1 The debt was accumulated not to finance productive investments, but to finance the government’s patronage employment * I am grateful to two anonymous referees, Craig and large military and police forces. Corrup- Burnside, David Dollar, Bernhard Gunter, Mary Hall- tion has been endemic, so there is the suspicion ward-Draiemeier, Aart Kraay, Robert Powell, Sergio that some of the proceeds of foreign loans Schmukler, and Axel Van Trotsenburg for helpful com- found their way into the pockets of the rulers. ments, to seminar participants at the IMF Institute, This is a description of Haiti’s experience in the Johns Hopkins School of Advanced International Stud- 90s. The 90s to which these facts refer are not ies, Oberlin University, the London School of Eco- the 1990s, but the 1890s. 2 nomics, and the World Bank for their comments, to The problem of heavily indebted countries is Punam Chuhan for providing access to debt service not a new one. From the two Greek city-states projections used in the calculation of the present value who defaulted on loans from the Delos Temple of debt series, and to Shelley Fu and Hairong Yu for in the fourth century BC to Mexico’s default on processing the debt service data. Any errors and omis- its first foreign loan after independence in 1827 sions are my responsibility alone. Final revision accep- to Haiti’s 1997 ratio of debt to exports of 484%, ted: 13 May 2002. 1677 1678 WORLD DEVELOPMENT

Mikhail Gorbachev, and the Dalai Lama. 4 The only problem with these arguments for Jubilee 2000 said that with debt forgiveness, the salutary effects of debt relief is the lack of ‘‘the year 2000 could signal the beginning of recognition that debt relief is not a new phe- dramatic improvements in healthcare, educa- nomenon. In the past, debt relief brought little tion, employment and development for coun- of the benefits promised for a new wave of debt tries crippled by debt.’’ 5 relief. In fact, debt relief did not even bring a Demonstrators from Washington to Prague reduction in debt, as poor country governments to Gothenburg to Genoa have thrown stones borrowed anew until they had again become for debt relief. The successor to the Jubilee 2000 heavily indebted. movement is a coalition called Jubilee Plus, Although there were intimations as long ago which calls for an unconditional cancellation of as 1967 that ‘‘debt-service payments have risen debt of the poor countries. 6 Kofi Annan in to the point at which a number of countries face April 2001 noted critical situations,’’ the current wave of debt relief for poor countries really got underway in 9 The Jubilee 2000 movement to cancel the debts of the 1979. The 1979 World Debt Tables of the poorest countries was an inspiration to us all. But its World Bank noted ‘‘lagging debt payment’’ on work did not finish with the Jubilee Year. We should official loans to poor countries, although ‘‘debt all be grateful that it is carrying on in the short term as or debt service forgiveness has eased the prob- ‘‘Drop the Debt,’’ and broadening its agenda. ... In lems for some.’’ The 1977–79 UNCTAD meet- the Millennium Declaration, world leaders called for ings led to official creditors writing off $6 all the bilateral debts of the least developed countries to be cancelled, in return for their making demons- billion in debt to 45 poor countries. The mea- trable commitments to poverty reduction. And they sures by official creditors included ‘‘the elimi- promised to deal ‘‘comprehensively and effectively’’ nation of interest payments, the rescheduling of with the debt problems of low- and middle-income debt service, local cost assistance, untied com- countries. pensatory aid, and new grants to reimburse old debts.’’ 10 The World Bank and the International The 1981 Africa report by the World Bank Monetary Fund (IMF) now have a pro- (usually known as the Berg Report) noted that gram called the HIPC initiative to provide debt Liberia, Sierra Leone, Sudan, Zaire, and writedowns––including for the first time, write- Zambia (all of which would become HIPCs) downs of IMF and World Bank claims in had already experienced ‘‘severe debt-servicing present value terms––for poor countries with difficulties’’ in the 1970s and ‘‘are likely to good policies. The G-7 summit in Cologne in continue to do so in the 1980s.’’ The Berg Re- June 1999 and the World Bank/IMF annual port hinted of debt relief, namely ‘‘longer-term meetings in October 1999 agreed on an expan- solutions for debt crises should be sought’’ and sion of this program, increasing the number ‘‘the present practice of donors separating aid of eligible countries, speeding up the process of and debt decisions may be counterproduc- receiving relief and increasing the amount of tive.’’ 11 debt relief provided for each country. The ex- The 1984 World Bank Africa report was pansion increased the total cost––in net present more forthright: ‘‘where monitorable programs value terms––of the HIPC initiative from exist, multiyear debt relief and longer grace US$12.5 billion to US$27 billion. 7 The IMF, periods should be part of the package of fin- World Bank, and other multilateral and bilat- ancial support to the program.’’ 12 The word- eral creditors had committed HIPC debt re- ing got even stronger in the World Bank’s, 1986 lief to 26 countries by April 1, 2001, for total Africa report: low income Africa’s financing commitments of $40 billion. (The World Bank needs will ‘‘have to be filled by additional bi- defines 41 countries as heavily indebted poor lateral aid and debt relief.’’ 13 The Bank’s 1991 countries––HIPCs. The HIPC problem has an Africa report continued escalating the rhetoric: Africa slant, as 33 of the 41 HIPCs are in Africa; ‘‘Africa cannot escape its present economic four are in Latin America. A number of HIPCs crisis without reducing its debt burden siz- are in the midst of violent conflict and so cannot ably.’’ 14 be considered for debt relief yet.) Jeffrey Sachs Meanwhile, the June 1987 G-7 summit in suggests that that the World Bank, IMF, com- Venice called for interest rate relief on debt of mercial banks and rich country governments low-income countries. The World Bank noted could absorb a writeoff of the $106 billion the ‘‘the past year has brought increasing recog- poorest countries currently owe to them. 8 nition of the urgency of the debt problems of HIPCs AND DEBT RELIEF 1679 the low-income countries of sub-Saharan Besides explicit debt relief, there also has Africa.’’ 15 One year later, the June 1988 G-7 been an implicit form of debt relief going on summit in Toronto agreed on a menu of op- throughout the period, which is the substitution tions, including partial forgiveness, longer ma- of concessional debt for nonconcessional debt. turities, and lower interest rates (these became It’s remarkable that the net present value of known as the ‘‘Toronto terms’’). 16 Meanwhile, future debt service for HIPCs rose through- in order to help African countries service their out the period despite the large net transfers of official debt, the World Bank in December 1987 resources from concessional lenders like the initiated a special program of assistance (SPA) International Development Association of the to low-income Africa. The IMF complemented World Bank and the concessional arms of bi- the SPA with the enhanced structural adjust- lateral and other multilateral agencies. ment facility. Both programs sought to pro- The necessity to provide continuing waves of vide ‘‘substantially increased, quick-disbursing, debt relief one after another, from UNCTAD highly concessional assistance to adjusting to Venice to Toronto to Houston to Trinidad countries.’’ 17 The 1990 Houston G-7 summit to London to Naples to HIPC to expanded considered ‘‘more concessional reschedulings HIPC, all the while substituting concessional for the poorest debtor countries.’’ The UK and for nonconcessional debt, may suggest some- the Netherlands proposed ‘‘Trinidad terms’’ thing is wrong with the implementation of debt that would increase the grant element of debt relief. There is the paradox that a large group reduction to 67%, from 20% under the ‘‘To- of countries came to be defined as heavily in- ronto terms.’’ 18 The 1991 London G-7 sum- debted at the end of two decades of debt relief mit agreed ‘‘on the need for additional debt and increased concessional financing. relief measures... going well beyond the re- This paper reviews possible explanations. lief already granted under Toronto terms.’’ 19 The revealed preference of debtors for high Through November 1993, the Paris Club (the debt may simply lead to new borrowing to re- club of official lenders) applied enhanced To- place old cancelled debts. Even if borrowing is ronto terms that were even more conces- constrained, poor countries that have a high sional. 20 In December 1994, the Paris Club discount rate against the future may run down announced ‘‘Naples terms’’ under which eligi- country assets. This is the external adjustment ble countries would receive yet additional debt equivalent to the fiscal adjustment ‘‘illusion’’ relief. 21 discussed by Easterly (1999a). Then, in September 1996, the IMF and The granting of progressively more favorable World Bank announced the HIPCs debt ini- terms for debt relief may also have perverse tiative, which was to allow the poor countries incentive effects, as countries borrow in antici- to ‘‘exit, once and for all, from the rescheduling pation of debt forgiveness and delay policy re- process’’ and to resume ‘‘normal relations with forms waiting for the best deal. Burnside and the international financial community, charac- Dollar (2000) and World Bank (1998b) sug- terized by spontaneous financial flows and the gest that aid does not raise growth in countries full honoring of commitments.’’ The multi- with poor economic policies. The World Bank’s lateral lenders for the first time would ‘‘take landmark Africa report (World Bank, 1994b) action to reduce the burden of their claims on suggested that many African countries failed to a given country,’’ albeit conditional on good depart from poor economic policies during the policies in the recipient countries. The Paris process of receiving adjustment loans from the Club at the same time agreed to go beyond World Bank and IMF. Naples terms and provide an 80% debt reduc- Since private lending withdraws because of tion in net present value terms. 22 the poor creditworthiness of HIPCs, the pro- Finally, as we saw above, the IMF and cess of debt relief has also led to a substitution World Bank expanded the ‘‘once and for all’’ of official lending for private lending and for- program in, 1999. Nor is the story over, as in- eign direct investment (FDI), which raises the dependent analysts like Birdsall, Williamson, concern that official lending may have not fol- and Deese (2002) point out that there remain lowed the same standards of creditworthiness HIPCs outside the program such as Indonesia, as private lending. There has been a redistri- Nigeria, and Pakistan, while the IMF and bution of roles even among official lenders, World Bank assumed optimistic projections for with some agencies making net transfers (debt export growth to make even existing HIPCs’ flows net of interest) to HIPCs and others re- post-relief situation manageable. 23 ceiving net transfers from HIPCs. 1680 WORLD DEVELOPMENT

2. THEORETICAL CONSIDERATIONS Therefore, if the discount rate is unchanged ON DEBT RELIEF before and after debt relief, the government will respond to debt relief by new borrowing A country that has gotten an ‘‘excessive’’ until the old ratio of net worth to consumption external debt may be one with a high discount is restored. In the same vein, if the terms of rate against the future––reflecting factors such lending are made more favorable by substitut- as a profligate government, political instability, ing concessional for nonconcessional debt then or interest group polarization. 24 After receiv- countries will reborrow to maintain the net ing debt relief, the high-discount rate country present value of debt service. Alternatively, the would like to accumulate the same amount of country could run down assets to restore external debt again. There will be an amount of the old ratio of net worth to consumption. 25 new borrowing corresponding to the amount (The country does benefit from higher con- of debt relief, until the old ratio of net worth sumption than would have been possible in the to GDP is restored. Alternatively, debt relief absence of debt relief.) conditionality could try to control new bor- On the other hand, what would happen if the rowing by constraining a country’s noninterest discount rate of the government changes? If current account deficit. Even this constraint a reformist government succeeds a spendthrift could be ineffective, however, because a coun- one, then debt relief would successfully provide try can reduce its assets to restore its desired a painless transition to a higher ratio of net low level of net worth in the long run. Finally, a worth to consumption (higher assets and lower government can impose its own high discount debt to consumption ratios). rate on the rest of the economy through policies Above, I described one possible reaction that tax private sector capital accumulation. If to debt relief is for the country to reborrow the government’s discount rate is unchanged enough to restore the old ratio of net worth to before and after debt relief, then these bad GDP. But, the external creditors (many of them policies will persist with debt relief. official lenders) may impose a limit on bor- Poor countries may have a higher discount rowing. A common formulation is to provide rate because individuals with shorter expected enough loans as to maintain a certain target lifetimes have higher discount rates (Blanchard debt ratio (usually a ratio to GDP or to ex- & Fischer, 1989, Chap. 3.3), and lifetimes are ports). I will suppose here that a country’s ex- shorter in poor countries. Alternatively, the ternal creditors supply an amount of credit government in poor countries may have a such that its debt to GDP ratio is equal to some higher discount rate because its expected tenure stable constant. 26 in office is shorter, because poor countries have Suppose that debt relief lowers the permitted more political instability (Easterly, 1999b). The debt ratio and imposes the lower level of bor- government may then impose its higher dis- rowing associated with maintaining the new count rate on the whole economy, as I argue debt ratio. This kind of debt relief could simply below. cause a one-for-one reduction in national assets The ‘‘high discount rate’’ can also be seen as with the amount of debt reduction as percent- shorthand for political economy factors that age of GDP. Since liabilities have been reduced, cause the government to overspend, prey on assets will in the long run decrease as well. private enterprise, and overextract rents from Being prevented from running up as much the economy to distribute as patronage. There debt as previously to finance consumption, is a large literature on the ‘‘neopatrimonial’’ the country will compensate by running down and ‘‘predatory’’ state (see Nafziger, 1993 and assets instead. If the current debt level was Van de Walle, 2001 for African examples). The ‘‘unsustainable’’ in that it represented too ruling elite in impoverished societies keeps itself heavy a burden relative to assets, then the new in power by buying off potential rivals and re- debt level will be equally ‘‘unsustainable’’ be- warding supporters, not to mention repressing cause society’s assets will decrease with the opposition by force. All of this requires the debt. 27 state to mobilize resources, which it does by So far I have not focused on the government, borrowing against the future as well as explic- leaving it unclear whether a high discount rate itly or implicitly taxing current production at could also characterize the private sector. We the cost of future growth. Given the elite does would generally expect that the government not feel secure, the future does not have a will be more impatient than the private sector, strong voice in elite circles. because of uncertainty of tenure and lower HIPCs AND DEBT RELIEF 1681 concern for future generations of government. market interest rate, then policy-makers will Governments in poor countries are subject to wait to ‘‘sell’’ policy reforms. greater instability (e.g., more coups) than in Going further, we can think of a Hotelling- rich countries, thus have shorter expected ten- type model for the depletion of the ‘‘stock’’ of ures in office, and thus have a higher discount needed policy reforms. If there is a supply of rate than in rich countries. Governments in poor needed reforms in HIPCs and a demand for countries could however, impose their high reform by donors, then the equilibrium ‘‘price’’ discount rate on the whole economy through of a marginal reform will rise at the rate of high tax rates and other policies adverse to interest. If HIPCs reform ‘‘too fast,’’ this would growth. drive down the price below the interest rate The government has a tradeoff between tax- trajectory––which means that HIPCs prefer to ing the private sector to finance government wait in such a case, driving the price back up to consumption today versus government con- the equilibrium interest rate trajectory. This sumption tomorrow financed by the future tax suggests policy-makers will adopt a gradualist base (which is decreasing in the tax rate today). rather than big-bang strategy of economic re- The private sector accumulates net worth and form in response to gradual debt relief, only grows faster the more that the rate of return to gradually depleting their stock of ‘‘necessary capital exceeds the discount rate, except that reforms.’’ This result is undesirable because it the government imposes a tax on the rate of means that countries will be stuck longer with return to capital. poor policies. The optimal tax rate for the government is There is also a perverse incentive created by increasing in the government’s discount rate. the response of debt relief to changes rather Intuitively, the government is trading off con- than the level of policies. Obviously, countries sumption today (increasing in the tax rate) with worse initial policies have more scope for versus consumption tomorrow (increasing in improvement. If debt relief responds exclusively private wealth tomorrow and thus decreasing in to changes, it may result in aid resources going the tax rate). A high discount rate government to countries with a worse level of policies on will choose to tax the private sector heavily. average. Countries could even engage in zig- The government will succeed in imposing its zag behavior, getting debt relief as they im- intertemporal preferences on the whole econ- prove policies and then backsliding to the old omy through its policies. The policies may in- level of policies. This is the kind of result that clude predatory behavior that implicitly rather Burnside and Dollar (2000) depicted as un- than explicitly taxes capital accumulation, such productive aid. as high corruption, real overvaluation, a high Finally, I have been dealing with the demand black market premium, high inflation, or fin- for external loans, but not with their supply. ancial repression. Countries that have negative growth, falling The empirical prediction is that a high dis- assets, bad policies, and increasing debt are count rate government will have bad policies poor credit risks. The prospect of debt for- that explicitly or implicitly tax the private sec- giveness also would tend to chill private lend- tor. If the government’s high discount rate is ing. We could expect that private creditors unchanged over time, then we would expect will stop lending at some point. If multilateral these bad policies to remain unchanged before and other official lenders perceive their role as and after debt relief. ‘‘filling the financing gap,’’ then their role will There are other ways in which debt relief increase over time in countries with falling creates perverse incentives for new borrowing. assets and increasing debt. The way that debt relief has been granted, of- The official lenders may want to keep lending fering progressively more favorable terms over even when the loans do not promote develop- time for two decades, also has perverse incen- ment because multilateral and donor agencies tive effects. Most obviously, it creates moral are often rewarded for volumes of assistance hazard incentives to borrow in the expectation rather than results. The official lenders may feel that part of this debt will be forgiven. the need to keep lending so the country does More subtly, incremental debt relief creates not default on earlier obligations to private or incentives to delay policy reforms, waiting for official creditors. The International Financial a progressively higher ‘‘price’’ at which to ‘‘sell’’ Institutions will thus fail to enforce conditions policy reforms. If the rate at which the amount even as they keep giving new loans. (The World of relief is increasing exceeds the international Bank (1998b) mentioned that it had given loans 1682 WORLD DEVELOPMENT to finance the same agricultural policy reforms group to see if the prediction of unchanged in Kenya five separate times.) The official behavior before and after debt relief hold rel- lenders should then bear some of the blame for ative to other developing countries. financing bad governments who pursue policies detrimental to their own citizens. (a) Debt accumulation andasset decumulation I will not try to distinguish these stories from each other in explaining becoming heavily The theoretical stories predicted that a high- indebted after debt relief. One alternate hy- discount rate country would be characterized pothesis to these that I will test would be that not only by high debt accumulation but also by HIPCs became heavily indebted through bad low asset accumulation, or even asset decumu- shocks such as adverse terms of trade growth lation. This contrasts with the traditional view and war. I test this hypothesis in the results that debt accumulation finances asset accumu- below. The other testable predictions from lation. The natural place to look for evidence on these stories are that high-debt countries will asset accumulation is investment. This is a poor show other signs of heavily discounting the indicator, however, as Devarajan, Easterly, future (such as asset decumulation), that new and Pack (2001) have found that traditionally borrowing will be associated with debt relief, measured investment is not productive in Africa and that policies will be worse in high debt where most of the HIPCs are concentrated. countries. The irresponsible official lender story A better albeit indirect way of getting at predicts that public debt will substitute for productive asset accumulation is to look at the private debt. These are sharp predictions con- behavior of per capita output. If we take per trasting with conventional wisdom that debt capita output as proportional to a broad con- relief finances or encourages asset accumulation cept of productive capital per capita, includ- and that actual debt falls over time with im- ing physical and human capital, technological proved terms on the debt. capital, knowledge, etc., then the evolution of per capita output would tell us something about the tangible and intangible forms of asset 3. THE EMPIRICAL EXPERIENCE accumulation. WITH DEBT RELIEF The natural measure of HIPCs’ external liabilities is their debt to GDP ratio. But since We can examine successively the response much of the HIPCs’ debt is concessional, the of new debt and assets to debt relief. I examine face value of the debt is a poor measure of the the 41 HIPCs as so classified by the IMF debt burden. I use the present value of debt and World Bank. 28 The countries are Angola, service as a ratio to GDP as the debt indicator. Benin, Bolivia, Burkina Faso, Burundi, Cam- Surprisingly, despite the attention given to the eroon, Central African Republic, Chad, Congo poor countries’ debt problem, I was unable to (Dem. Rep.), Congo (Rep.), Coote^ d’Ivoire, Equ- find time series of the present value of debt atorial Guinea, Ethiopia, Ghana, Guinea, service for HIPCs. (The World Bank’s Global Guinea-Bissau, Guyana, Honduras, Kenya, Lao Development Finance reports an estimate of the PDR, Liberia, Madagascar, Malawi, Mali, present value of debt service for the latest year, Mauritania, Mozambique, Myanmar, Nicara- while earlier reports reported three year mov- gua, Niger, Rwanda, Sao Tome and Principe, ing averages going back to 1991. These mov- Senegal, Sierra Leone, Somalia, Sudan, Tan- ing averages do not give internally consistent zania, Togo, Uganda, Vietnam, Yemen, and numbers for individual years, so I do not use Zambia. them.) Using data on scheduled debt service The reader may worry that we have a sample from the Debt Reporting System of the World selection bias, because these countries were Bank, a time series 1979–97 for each of the classified as HIPCs at the end of the period. HIPCs’ present value of debt obligations was Hence, it would not be so surprising if we find calculated for this paper. 29 that things did not go well for these countries in Figure 1 shows the evolution of the HIPCs’ the period prior to their classification. This per capita output in 1997 prices and their sample selection is justified, however, because it median debt to GDP ratio in present value is this group that the debt relief efforts targeted. terms. 30 If we take the trend fall in output over We can think of the following results as docu- 1979–97 as representing a drop in potential menting the extent of adverse selection in debt output, and potential output as proportional to relief efforts. We will retrace the path of this a broad notion of productive assets, then there HIPCs AND DEBT RELIEF 1683

Figure 1. External debt/GDP (present value terms) and per capita income in HIPCs. was asset decumulation at the same time as atization foreign exchange revenues for 1988– there was high debt accumulation. The HIPCs’ 97. Over this period, total sales of state enter- debt problem arose not just because of new bor- prises in the HIPCs amounted to US$4 billion. rowing, but because of disinvestment in pro- This is an underestimate, because not all priv- ductive potential. This is consistent with a story atization revenues are recorded in the official in which the HIPCs can be characterized as statistics. Even using this flawed data, there is persistently high discount rate countries. a positive and significant correlation of 0.35 There is some possibility of a break point across the 41 HIPCs between the amount of toward the end of the period in which the debt debt forgiveness and the amount of privatiza- ratio went down and output went up. This tion foreign exchange revenues. Privatization corresponds to the period after the new HIPC may have been done for efficiency reasons or debt relief initiative was launched, which could even as a condition for debt relief, but it also indicate more success for this latest debt relief may suggest a high discount rate economy attempt. But, the period after the break is too running down its assets. short to evaluate whether it is a permanent change. (b) Debt relief andnew borrowing I next turn to oil production, for which we have 1987–96 data. There are 10 HIPCs that The data on debt relief from the World are oil producers. Oil production is a form of Bank’s World Debt Tables only go back to asset decumulation, since it takes an asset in the 1989. The relationship between debt relief and form of oil in the ground and turns it into cash new borrowing over this period is interesting: that can be an alternative form of financing total debt forgiveness for 41 heavily indebted consumption if conventional debt is constrained. poor countries over 1989–97 totaled US$33 Did HIPCs have higher oil production growth billion, while their new borrowing was US$41 over this period of debt relief than did the non- billion. This seems to point in the direction of HIPC oil producers? The answer is yes. The the prediction above that debt relief will be average log growth in oil production is 6.6 met with an equivalent amount of new bor- percentage points higher in the HIPCs than in rowing. 31 the non-HIPCs, which is a statistically signifi- Was new borrowing the highest in the cant difference. The average log growth in oil countries that got the most debt relief ? Run- production in HIPCs was 5.3%; in non-HIPCs, ning a regression for the 40 HIPCs that have it was )1.3%. complete data, there is a statistically signifi- Another form of asset decumulation taking cant association between average debt relief as place at this time was sales of state enterprises a percent of GDP and new net borrowing as to foreign purchasers. We have data on priv- percentage of GDP. The offset in this case is 1684 WORLD DEVELOPMENT less than one for one: one percentage point of significantly higher in 1997 than it was in 1979. GDP higher debt forgiveness translated into Again this result is not surprising given that we 0.34% of GDP new net borrowing. have selected the sample based on their debt at Another bit of evidence that debt relief did the end of the period. Still, it suggests that for not lower debt significantly is to look at exter- a large group of 41 countries, new borrowing nal debt to export ratios over 1979–97. I again (more than) kept pace with the amount of debt use the present value of debt service as a mea- relief, as would have been predicted by the sure of external debt, but now as a ratio to model for countries with unchanged discount exports. I again use 1979 as a base year because rates. 32 it was the year the UNCTAD summit inaugu- rated the current wave of debt relief. I have (c) Regression analysis of HIPCs’ data for 28–37 HIPCs over 1979–97. Despite macroeconomic imbalances and the ongoing debt relief, the median present country policies value debt to export ratio rose strongly during 1979–97 (Figure 2). We can see three dis- In this section, I develop summary statistics tinct periods: (i) 1979–87 when debt ratios rose of HIPCs’ policy stance. I regress an average strongly; (ii) 1988–94 when debt ratios re- over the debt relief period 1980–97 of each mained constant; and (iii) 1995–97 in which policy indicator or macroeconomic imbalance debt ratios fell. The behavior in periods (i) and on the log of initial income, and a dummy for (ii) is consistent with failed debt relief, while the HIPCs for the whole sample of less-developed drop in the last period may indicate that the countries (LDCs). 1996 HIPC debt relief program has been more Table 1 shows the results. We see that the successful than earlier efforts. average levels over 1980–97 of current account Despite the drop in the last period, however, deficits, budget deficits (with or without grants), the median debt to export ratio is statistically M2/GDP, and real overvaluation, were worse

Figure 2. 95% conﬁdence interval for median present value of debt of HIPCs as a ratio to exports. HIPCs AND DEBT RELIEF 1685

Table 1. Regression results for policies in LDCs 1980–97, controlling for income (sample of all LDCs) Dependent variable, Current account balance/GDP Budget deficit excl. grants/GDP average 1980–97 Coefficient t-Statistic Coefficient t-Statistic Log income, 1979 0.08 0.11 1.47 2.08 Dummy for HIPCs )5.58 )4.36 )4.26 )3.67 R2 0.25 0.32 # Observations 77 81 Budget deficit incl. grants/GDP M2/GDP Log income, 1979 )0.34 )0.46 1.50 0.48 Dummy for HIPCs )4.97 )3.94 )15.65 )2.96 R2 0.19 0.15 # Observations 84 83 Log (1 þ inflation rate) Index of overvaluation (based on Dollar, 1992) Log income, 1979 0.13 2.60 9.07 1.13 Dummy for HIPCs 0.15 1.79 64.19 4.92 R2 0.08 0.30 # Observations 82 68 Real interest rate Log (1 þ black market premium) Log income, 1979 )0.01 )0.47 0.04 0.60 Dummy for HIPCs )0.05 )1.79 0.09 0.78 R2 0.05 0.01 # Observations 74 77 CPIA (1–5 scale) Log income, 1979 0.07 0.72 Dummy for HIPCs )0.33 )2.15 R2 0.11 # Observations 77 for HIPCs. The differences in HIPCs’ real in- deficit. Table 2 shows some intriguing pat- terest rate, black market premium, and infla- terns. First, HIPCs received less FDI than tion rates from the rest of the LDC sample are other LDCs, controlling for income. This may not statistically significant (although inflation be an indirect indicator of the bad policies and real interest rates are marginally significant found on the other indicators: investors do not at the 10% level). want to invest in an economy with high budget The HIPCs also were worse on the broad deficits, high overvaluation, and high corrup- measure of policy given by the World Bank’s tion. Investors may also have worried what Country Policy and Institutional Assessment debt relief may have meant for other external (CPIA). This measure of policies not only liabilities like the stock of direct foreign in- includes a rating of policy stance, but also of vestment. It also is a confirmation of the pre- institutional quality––like the prevalence of diction that private capital flows will dry up in corruption. The HIPCs’ average CPIA 1980–97 high discount rate economies with falling assets was worse than the CPIA for other LDCs. and increasing debt. The result on the current account deficit is Second, despite their poor policies, HIPCs not surprising: obviously HIPCs got to be received more in World Bank and IMF fin- HIPCs by borrowing a lot! The results on poli- ancing than other LDCs. The result on World cies are not as obvious, as the debt accumula- Bank financing is controlling for initial income tion could have come from bad external shocks (negatively related to World Bank financing). (on which more in a moment) rather than bad The effect (0.96% of GDP) is small relative to policies like real overvaluation, low financial the size of the current account deficit, but large depth, and poor CPIA. relative to the mean amount of World Bank Even more interesting is to examine the financing (1.1% of GDP). The share of World composition of financing the current account Bank financing in gross disbursements also was 1686 WORLD DEVELOPMENT

Table 2. Financing composition of debt accumulation, 1979–97 Dependent variable, FDI/GDP Coeﬃcient t-Statistic average 1980–97 Coeﬃcient t-Statistic Log income, 1979 0.11 0.66 Dummy for HIPCs )0.84 )2.92 R2 0.17 # Observations 77 World Bank Financing/GDP IMF Financing/GDP Log income, 1979 )0.40 )3.76 0.05 0.41 Dummy for HIPCs 0.96 5.35 0.73 3.40 R2 0.53 0.15 # Observations 83 83 World Bank share of disbursements/GDP IMF share of disbursements/GDP Log income, 1979 )8.10 )5.72 0.69 0.79 Dummy for HIPCs 7.17 3.14 4.37 3.12 R2 0.54 0.13 # Observations 76 76

Table 3. Terms of trade shocks and war, 1979–97 Dependent variable, Least-squares log growth in terms of trade Percent of period at war average 1979–97 Coefficient t-Statistic Coefficient t-Statistic Log income, 1979 0.00 )0.97 )0.04 )0.75 Dummy for HIPCs 0.00 )0.05 )0.09 )1.10 R2 0.02 0.02 # Observations 77 76 significantly higher (by 7.2 percentage points) terms of trade shocks. Table 3 shows, how- in HIPC than in non-HIPCs. This confirms ever that the least-squares log growth in terms the prediction that multilateral lenders ‘‘filling of trade over 1979–97 was not significantly the financing gap’’ will have a significant role worse for HIPCs. The LDC sample as a whole in financing high-discount rate economies. shows significantly worsening terms of trade The results are similar for the IMF. I re- over 1979–97, but the HIPCs do not stand out gressed IMF financing on a constant, initial as any different than their less heavily indebted per capita income and the HIPCs dummy. The neighbors. HIPC dummy is indeed significant. Like the Another possible shock that might have World Bank HIPC dummy, the effect is small caused HIPCs to have high debt ratios is war, relative to current account deficits (0.73% of since it both destroys productive assets and GDP), but large relative to the non-HIPCs causes additional government spending that average IMF financing (0.5% of GDP). The has to be financed. But, as shown in Table 3, HIPC effect for the IMF’s share of disburse- HIPCs were not more likely to be at war than ments is of the same sign and significant––the the rest of the LDC sample. 33 IMF had 4.4 percentage points more of gross In sum, we have a pattern of poor policy disbursements to HIPCs than to non-HIPCs, indicators that most needed to be improved controlling for income. The HIPCs got to be to avoid a debt crisis. Not surprisingly, HIPCs’ HIPCs in part by borrowing from the World policies were worse precisely in those areas–– Bank and IMF. I will go into more detail on high current account deficits and budget defi- who gave loans to the HIPCs (and when) in a cits––that led to high debt accumulation. Less later section. obvious were bad policies on financial repres- One explanation of the HIPCs’ becoming sion and exchange rate overvaluation. This is heavily indebted is that they suffered adverse consistent with these countries having a high HIPCs AND DEBT RELIEF 1687 discount rate that was unchanged before and GDP over the period of incremental debt relief after debt relief. This is also consistent with 1979–97. policy-makers waiting for the best deal during The budget deficit to GDP ratio also fails the incremental process of debt relief. It is also to improve over the debt relief period 1979–97 consistent with the moral hazard problem that (Figure 3), for a sample of 23–35 countries, if after the initial debt relief in 1979, HIPCs may anything deteriorating to the very high level of have rationally anticipated that much of their around 10% of GDP. These figures treat grants new borrowing would be later forgiven. as a source of financing. This would be justified if we think of grants as temporary, with the (d) Current account deficits and budget donors planning that the country exit from deficits over time needing foreign aid after a certain interval. But, grants in practice may be permanent and they In addition to averages over 1980–97, it is do not imply future debt servicing requirements, important also to look for trends. Did HIPCs’ so it’s of interest to see the budget deficit in- policies get better over the two decades of debt cluding grants. The grant-inclusive budget def- relief ? On the current account deficit, perhaps icit still fails to improve for HIPCs (Figure 3). the most important measure of policy stance The results on the current account deficit and for heavily indebted countries, the news is not budget deficit do not show a clear improvement good. (This measure of the current account in behavior during the process of incremental deficit treats grants as revenue rather than fi- debt relief. This is consistent with the HIPCs nancing.) The median current account deficit being persistently high-discount rate econo- has stayed high and constant at around 7.5% of mies.

Figure 3. Current acount andﬁscal balances over time in HIPCs. 1688 WORLD DEVELOPMENT

(e) Debt relief andother country policies The evidence is very mixed, as shown in over time Figure 4. The real interest rate for HIPCs is an indicator of either the private return to capital How have other HIPC policies behaved if interest rates are uncontrolled or financial during the period of incremental debt relief repression if there is a nominal interest rate 1979–97? As noted in the theoretical section, ceiling. HIPCs had flat real interest rates over poor policies is one mechanism by which the time. Contrary to the stereotype of HIPCs as government imposes its own high discount rate financially repressed, the median real interest on the rest of the economy. There is also the rate was positive for most of the period (al- worry that countries would respond to incre- though not significantly different than zero). mental debt relief by postponing policy re- A different variable related to financial reforms, waiting for a higher ‘‘price’’ at which to pression, the ratio of M2 to GDP (financial ‘‘sell’’ policy reforms. Alternatively, countries depth) in HIPCs, shows a different picture. We could slowly reform, selling off pieces of reform have already seen that HIPCs had worse fin- as the price rises. The intent of the debt relief ancial depth than other LDCs. Financial depth, efforts, in contrast, was that policies would which King and Levine (1993a,b) identified as improve immediately as a condition for getting a critical determinant of growth, does not im- new debt relief. Which happened? prove in the HIPCs over time.

Figure 4. HIPCs country policy indicators over time (95% conﬁdence interval for median current account balance/GDP in HIPCs). HIPCs AND DEBT RELIEF 1689

The inflation rate oscillated in the HIPCs International Development Association (IDA) without any clear trend over 1979–97. The in- financing alone more than tripled its share in flation rate was not in the range that (Bruno disbursements. The share of private credit be- & Easterly, 1998) identified as associated with gan the period 3.6 times higher than the IDA negative growth performance (40% and above), share; by the end of the period, the share of although it spent a few years in the 20–40 dan- IDA was 8.6 times higher than that of pri- ger zone where there is a high risk of slipping vate financing. The private credit flows do not into the above 40% zone (Bruno, 1995). take into account private capital flight, and so HIPCs spent a good part of the debt relief probably understate the degree to which private period with the black market premium above capital flows reversed themselves. A recent the 20% threshold defined by Sachs and Warner study found that Africans held 39% of private (1995) as one of the criteria for being a ‘‘closed’’ capital outside of the home country during economy. After a wild period in the mid-1980s, the period in which Africa’s high debt was ac- however, there is a tendency for both the me- cumulated (Collier, Hoeffler, & Patillo, 1999). dian and variance of the black market premium Similarly, Ajayi (1997) finds that the stock of to fall over time in the HIPCs. 34 accumulated capital flight over 1980–91 was on There is good news and bad news on another average 40% of the external debt outstanding exchange rate measure, the measure of devia- in the HIPCs, with such extremes as Rwanda tion of local prices from purchasing power (94.3%), and Kenya (74.4%). parity at the official exchange rate. I construct The share of IMF financing, which began an purchasing power parity index of Dollar at the same level as IDA financing, remained (1992) to benchmark the real exchange rate roughly unchanged. The other important change as an average of 1976–85 for each country, is away from bilateral financing in favor of then convert it to a time series using the usual IDA and other multilateral concessional fin- definition of the real exchange rate ðPDomestic= ance. ðEPUSÞÞ. The good news is that the real ex- Another important thing to examine is net change rate depreciates over 1979–97 in the transfers (net flows minus interest payments). HIPCs. This is one of the major achievements On debt that carries a market interest rate, of this 20-year process of adjustment and debt positive net transfers imply that the debt is relief. growing faster than the interest rate. This im- The bad news is that the initial position was plies the debt is unsustainable (if the recipient extreme overvaluation and the improvement continued to borrow to pay the interest and was only gradual, so that the average exchange then some, this would imply the present value rate in the HIPCs for the period is severely of debt is unbounded). Net transfers from con- overvalued (as we saw in the regression analy- cessional sources, on the other hand, carry a sis). Another piece of bad news is that other large grant element and so do not have the LDCs also had a tendency toward real depre- same implications for debt sustainability; if any- ciation, so that at the end of the period the thing higher concessional net transfers should HIPCs were still 24% overvalued relative to increase the likelihood of sustainability. other LDCs. Figure 6 shows that all the nonconcessional The HIPCs fared worse on our broadest net transfers were positive, and so contributed measure of policy, the World Bank’s subjective to the rapid growth of debt during 1979–87 rating called the CPIA. 35 The HIPCs display (recall Figure 2). But, there were also large net no clear trend over time. This is consistent with transfers from concessional sources (IDA, the story that intertemporal preferences were other multilaterals, and the bilaterals)––total unchanged before and after debt relief, and the net transfers to the HIPCs of US$33 billion–– government used poor policies to impose its which makes it all the more striking that these high discount rate on the whole economy. countries became increasingly highly indebted in net present value terms over this period. (f) Supply of financing Figure 7 shows that there was a huge shift in net transfers from 1979–87 to 1988–97, a period Figure 5 shows the composition of gross in which debt ratios stabilized. Large posi- disbursements to HIPCs over 1979–97. The tive net transfers from IDA and bilateral con- prediction that private credit would disappear cessional sources offset negative net transfers and multilateral financing assume an increased for IBRD, IMF, bilateral nonconcessional, and share are more than confirmed. World Bank private sources. 36 This was another form of 1690 WORLD DEVELOPMENT

Figure 5. Composition of gross disbursements to HIPCs.

Figure 6. Net transfers to HIPCs by creditor, 1979–87 (billion US$). HIPCs AND DEBT RELIEF 1691

Figure 7. Net transfers to HIPCs by creditor, 1988–97 (billion US$).

‘‘debt relief,’’ since it exchanged concessional stabilized inflation by 1987, but growth was debt with a large grant element for noncon- poor, real interest rates went from excessively cessional debt. The net present value of debt negative to excessively positive, and overvalu- however, remained roughly unchanged over this ation remained. period, at least until the last few years, sug- A cynical interpretation would be that as gesting that these economies persisted in ‘‘high countries could not or would not pay their discount rate behavior.’’ nonconcessional debt, official lenders replaced This increase in multilateral lending (a good their nonconcessional debt with concessional part of it structural adjustment lending) took debt that had a large grant element. This place despite the poor policies noted earlier, should have significantly eased the debt servic- which casts doubt on the wisdom of official ing burden of the HIPCs. Even so, the HIPCs lending that took place. For example, Zambia still had enough of a debt problem at the end received 18 adjustment loans over 1980–99 of the period that lenders initiated more debt from the IMF and World Bank but had sharply relief. negative growth, large current account and A major motivation of the HIPC Initiative budget deficits, high inflation, a high black mar- has been to use the resources freed up by debt ket premium, massive real overvaluation, and a relief to help the poor. It is quite a challenge negative real interest rate for most of that pe- however for the HIPC governments to imple- riod. As of the year 2000, when it received a ment effectively conditions on increasing pov- commitment of debt relief under the HIPC erty-reducing spending when they have such initiative, Zambia still had high inflation and a mixed record on conditions on improving high budget deficits. macroeconomic policies––macropolicies are usu- Coote^ d’Ivoire got 26 adjustment loans over ally considered easier to implement than pov- 1980–99 but had negative growth, high current erty reduction programs. Moreover, the data account deficits, and an overvalued real ex- are not in place for governments to even know change rate. After the initiation of adjustment whether spending is reaching the poor. A sur- lending, Bolivia had a hyperinflation, negative vey in March 2001 found that only two of 25 of real interest rates, and overvaluation. Bolivia HIPCs would be able to carry out satisfactory 1692 WORLD DEVELOPMENT expenditure-tracking systems within one year Coote^ d’Ivoire received 1,276 times more per (IMF & IDA, 2001). A year later, in March capita aid net flow than India in 1997. 38 2002, none of the HIPCs’ expenditure tracking The results on composition of financing are systems was rated as satisfactory and Uganda also rather alarming. The HIPCs’ debt crisis was the only HIPC to have reported actual developed because of the expansion of official poverty-reducing spending in fiscal year 2000/ lending. The official lenders did not seem to 2001 (IMF & IDA, 2002). follow the same prudential rules as private capital, which pulled out of the HIPCs. The Concessionary finance used unproductively leads to IMF and World Bank provided more financing indebtedness which is then used as an argument for to HIPCs over 1979–97 than other countries further concessionary finance (Bauer, 1972, p. 127). of their income level, despite their worse policies. In the second half of the period, positive net transfers from IDA and bilateral conces- 4. CONCLUSIONS sional sources offset negative net transfers from IBRD, IMF, bilateral nonconcessional and pri- The theoretical concepts in this paper predict vate sources. that governments with unchanged discount What are the policy implications? Debt relief rates in the long run will respond to debt relief is futile for governments with unchanged long- by running up new debts or by running down run preferences (i.e., governments that continue assets. There are some signs that the incre- to be dominated by rent-seeking elites). At best, mental process of debt relief over the past two only governments that display a fundamental decades fulfilled these predictions. New bor- shift in their development orientation should rowing was correlated with debt relief so that be eligible for debt relief. To assess whether debt ratios actually got worse. Per capita out- governments have made such a fundamental put had a trend decline, suggesting decumula- shift in preferences, some track record of devel- tion of productive assets, broadly defined. Oil opment-oriented behavior should be required reserves were depleted more rapidly and sales prior to granting debt relief. There were im- of state enterprises to foreign owners were portant steps in this direction in the 1996 HIPC higher in countries that got debt relief. initiative, which unfortunately may have been Policies by which government implicitly or weakened by the 1999 ‘‘enhanced HIPC.’’ explicitly taxes asset accumulation displayed Official lenders should not keep ‘‘filling the a mixed pattern of some gradual policy im- financing gap’’ in violation of prudential stan- provements and some failures to improve. The dards of creditworthiness. most important policy indicators for heavily Perhaps what has been most damaging to indebted countries––the current account deficit incentives for new borrowing and delayed re- and the budget deficit––failed to improve, and forms is the creeping process of debt relief over they remained above other LDCs’ levels con- the past 20 years. Although debt relief is done trolling for their initial values in 1979. in the name of the poor, the poor are worse off There is also some good news. HIPCs’ ex- if debt relief creates incentives to delay reforms change rate overvaluation and black market necessary for growth. premium improved over time. Debt ratios fell A once-and-for-all program is greatly supe- in the past three years, and per capita income rior to a gradual program of increasing relief. rose. This could indicate that the most recent The once-and-for-all program has to attempt to HIPC debt relief initiative has been more suc- establish a credible policy that debt relief will cessful than earlier debt relief efforts, although never again be offered in the future, and that we have only a few years of data on which to it is only giving debt relief to governments with draw conclusions. Debt relief at least makes a shift in development orientation. If this is possible higher consumption in HIPCs, if noth- problematic, then the whole idea of debt relief ing else. is problematic. It results in more resources go- Still, the problem of the adverse selection of ing to countries with bad policies than poor HIPCs remains a serious one. By 1997, with the countries with good policies. It is ironic that the coming of the new multilateral debt relief ini- aid community allegedly arrived at the con- tiative, HIPCs received 63% of the flow of re- sensus ‘‘aid works in a good policy environ- sources devoted to poor countries despite only ment’’ while one of the principal development accounting for 32% of the population of those efforts has been a program that selects countries countries. 37 Including debt reduction as aid, based on past bad policies. HIPCs AND DEBT RELIEF 1693

NOTES

1. World Bank (1998a, p. 56). and Williamson (1986), Mistry (1988), Nafziger (1993), and Parﬁtt and Riley (1989). For more recent compila- 2. Dupuy (1988, p. 116) and Lundahl (1992, p. 39, 41, tions of analysis, see Iqbal and Kanbur (1997) and 244). Brooks et al. (1998).

3. Dommen (1989) and Wynne (1951, pp. 5–7). 16. World Bank (1988b, p. xxxviii).

4. On September 23, 1999, a delegation including U2’s 17. World Bank (1989, p. 31). Bono, pop entertainment figures Quincy Jones and Bob Geldof, and Jeffrey Sachs met with Pope John Paul II 18. World Bank (1990, p. 29). on Third World debt relief. For more on Jubilee 2000, see the web sites www.jubileeusa.org and www. 19. World Bank (1991b, p. 31). jubilee2000uk.org. 20. World Bank (1993, p. 6). 5. http://www.jubilee2000uk.org/main.html. 21. World Bank (1994a, p. 42). 6. http://www.jubilee2000uk.org/ In 2001, there was also a campaign called ‘‘drop the debt,’’ featured at 22. Boote, Kilby, Thugge, and Van Trotsenburg (1997, http://www.dropthedebt.org/home.html. On June 19, 2001, p. 126, 129). the coalition unveiled a controversial ad featuring a healthy Western baby breast-feeding from a malnour- 23. Other analysts like Roodman (2001) also point out ished African mother and asked ‘‘have not we taken that Indonesia, Nigeria, and Pakistan have as good a enough?’’ As of April 2002, the ‘‘drop the debt’’ web site claim to be HIPCs as the official HIPCs according to was no longer operating but the www.jubileeusa.org site most objective criteria. uses the same slogan.

24. See Easterly and Levine (1997) on ethnic polariza- 7. International HeraldTribune: June 14, 1999, p. 1; tion. Financial Times: June 21, 1999, p. 3; see also the World Bank web site on the HIPC initiative www.worldbank. org/hipc. 25. The consumption path will also shift up by the annuity value of the lump-sum transfer implied by debt relief. In a real life example of part of this consumption 8. International HeraldTribune: June 12, 1999, p. 6; see eﬀect, the President of Nicaragua gave workers a half also Center for International Development (1999). day oﬀ to celebrate being part of the HIPC program.

9. The quote is from UNCTAD (1967, p. 3). 26. The idea of maintaining a stable external debt to GDP ratio as one criterion for current account 10. World Bank (1979, pp. 7–8); UNCTAD (1983, sustainability is common in oﬃcial agencies and in the p. 3). academic literature. See for example, Cohen (1996), Dadush, Dhareshwar, and Johannes (1994), Milesi- 11. World Bank (1981, p. 129). Ferretti and Razin (1996), Roubini and Wachtel (1998), Van Wijnbergen, Anand, Chhibber, and Rocha (1992), 12. World Bank (1984, p. 46). and World Bank (1998a).

27. I have treated all assets as domestic capital stock, 13. World Bank (1986, p. 41). and have not introduced the possibility of foreign assets. It is straightforward to extend the definition of 14. World Bank (1991a, p. 176). A to include foreign assets (capital flight). Therefore, the country could reduce its accumulation of flight capital 15. World Bank (1988a, p. xix). The general literature abroad in response to a reduction in available new started noticing low-income African debt at about the borrowing. There is ample scope for flight capital to same time. See Greene (1989), Humphreys and Under- adjust at the margin, and flight capital is a major factor wood (1989), Husain and Underwood (1991), Lancaster in HIPCs (see below). Of course, the flight capital is in 1694 WORLD DEVELOPMENT private hands while the debt is public, so there is the 32. The calculation for this paper that the median debt ‘‘transfer problem’’ of taxing the private sector to pay to export ratio in 1997 is 221% is lower than the World the public debt. Bank’s Global Development Finance (GDF) estimate of 278%. Obviously, the present discounted value is sensi- 28. See the World Bank web site www.worldbank. tive to the assumption on the discount rate. Still, the cor- orgnhipc. relation across HIPCs between the debt to export ratios from GDF and those from this paper in 1997 is 0.78. 29. The discount rate used is the average LIBOR over 1979–97. 33. The war variable was the percent of time at war on national territory during 1979–94. 30. Since debt is not in PPP prices, I also use a non- PPP measure of output––the World Bank’s World 34. Drazen and Easterly (2001) find that inflation and Development Indicators Atlas method per capita income the black market premium display a ‘‘crisis provokes in 1997, and then apply median real per capita growth in reform’’ property, whereas the growth rate, the budget HIPCs to get the series. The HIPCs’ median debt to deficit, and the current account deficit do not. They also GDP ratio is somewhat lower than that in the World find that aid is reduced at high levels of inflation and the Bank’s Global Development Finance (50% here com- black market premium, while it increases with current pared to 70% in GDF), because the discount rate I used account deficits and budget deficits. is higher. Nevertheless, the correlation of debt to GDP ratios between GDF and mine across the HIPCs is 35. The CPIA has four components, which are Macro- 0.90. economic Management and Sustainability of Reforms, Policies for Sustainable and Equitable Growth, Policies for Reducing Inequalities, and Public Sector Manage- 31. Unfortunately, these figures are in nominal rather ment. It is available for 1977–98. These results should be than NPV terms. But, since NPV of debt to exports is taken with a grain of salt, not only because of the fairly stable over this period, this supports the idea that subjective element but also because the methodology for new borrowing replaced forgiven debt. Moreover, the the rating has changed over time. relationship between debt relief and new borrowing year by year is not contemporaneous. New borrowing is concentrated toward the beginning of the period, while 36. IDA is the concessional lending arm of the World debt relief is concentrated toward the end of the period. Bank, while IBRD is the nonconcessional lending part One possibility is that the high level of new borrowing of the World Bank. caused a threshold to be passed that resulted in debt relief; this possibility suggests a potentially serious 37. This calculation sums net flows of long-term debt problem with moral hazard. Another related possibility and debt stock reductions going to HIPCs and to other is that borrowing countries expected progressively more low income economies, where low income is defined as in favorable terms of debt relief and engaged in pre- the World Bank’s World Development Indicators. emptive new borrowing to keep their long-run ratio of net worth to GDP unchanged. In this case, debt relief 38. India’s low per capita aid receipts represent not was an illusion. Finally, it is possible that the debt relief only its suffering from the adverse selection of aid efforts of 1996–97 were more successful than earlier donors, but also from the tendency of large countries to efforts. receive small amounts of aid per capita.

REFERENCES

Ajayi, S. I. (1997). An analysis of external debt and Birdsall, N., Williamson, J., & Deese, B. (2002). Deliver- capital ﬂight in the severely indebted low-income ing on debt relief: from IMF gold to a new aid architec- countries. In Z. Iqbal, & R. Kanbur (Eds.), External ture. Washington, DC: Center for Global Development ﬁnance for low-income countries. Washington, DC: and Institute for International Economics. International Monetary Fund. Blanchard, O., & Fischer, S. (1989). Lectures on macro- Bauer, P. T. (1972). Dissent on development: studies and economics. Cambridge, MA: MIT Press. debates in development economics. Cambridge, MA: Boote, A., Kilby, F., Thugge, K., & Van Trotsenburg, Harvard University Press. A. (1997). Debt relief for low-income countries and HIPCs AND DEBT RELIEF 1695

the HIPC debt initiative. In Z. Iqbal, & R. Kanbur International Monetary Fund and International Devel- (Eds.), External finance for low-income countries. opment Association (2001). Tracking of poverty- Washington, DC: International Monetary Fund. reducing public spending in heavily indebted poor Brooks, R., Cortes, M., Fornasari, F., Ketchekmen, B., countries (HIPCs). Prepared by the IMF’s Fiscal Metzgen, Y., Powell, R., Rizavi, S., Ross, D., & Ross, Affairs Department and the World Bank’s Poverty K. (1998). External debt histories of ten low-income Reduction and Economic Management Network, developing countries lessons from their experience. in collaboration with other departments of the IMF In IMF Working Paper WP/98/72. Washington, DC: and World Bank, March 27. IMF. International Monetary Fund and International Devel- Bruno, M. (1995). Does inflation really lower growth? opment Association (2002). Actions to strengthen Finance anddevelopment,32 , 35–38. the tracking of poverty-reducing public spending in Bruno, M., & Easterly, W. (1998). Inflation crises and heavily indebted poor countries (HIPCs). Prepared long-run growth. Journal of Monetary Economics. by the IMF’s Fiscal Affairs Department and the Burnside, C., & Dollar, D. (2000). Aid, policies, and World Bank’s Poverty Reduction and Economic growth. American Economic Review. Management Network, in collaboration with other Center for International Development (1999). Imple- departments of the IMF and the World Bank, March menting debt relief for HIPCs. Cambridge, MA: 22. CID, Harvard University. King, R. G., & Levine, R. (1993a). Finance and growth: Cohen, D. (1996). The sustainability of African debt. Schumpeter might be right. Quarterly Journal of WorldBank Policy Research Paper 1621 . Economics, 108, 717–737. Collier, P., Hoeffler, A., & Patillo, C. (1999). Flight King, R. G., & Levine, R. (1993b). Finance, entrepre- capital as a portfolio choice. WorldBank Policy neurship, and growth: theory and evidence. Journal Research Paper 2066, February. of Monetary Economics, 32, 513–542. Dadush, U., Dhareshwar, A., & Johannes, R. (1994). Lancaster, C., & Williamson, J. (Eds.) (1986). African Are private capital flows to developing countries debt and financing. Washington, DC: Institute for sustainable? WorldBank Policy Research Working International Economics. Paper 1397. Lundahl, M. (1992). Politics or markets: essays on Devarajan, S., Easterly, W., & Pack, H. (2001). Is haitian underdevelopment. London: Routledge. investment in Africa too high or too low? Macro- Milesi-Ferretti, G. M., & Razin, A. (1996). Sus- and microevidence. WorldBank Research Working tainability of persistent current account deficits. Paper 2519. NBER Working Paper 5467. NBER: Cambridge, Dollar, D. (1992). Outward-oriented developing eco- MA. nomies really do grow more rapidly: evidence Mistry, P. S. (1988). African debt: the case for relief for from 95 LDCs 1976–85. Economic Development and sub-Saharan Africa. Oxford: Oxford International Cultural Change, 40(3), 523–544. Associates. Dommen, E. (1989). Lightening the debt burden: some Nafziger, E. W. (1993). The debt crisis in Africa. sidelights from history. UNCTAD Review, 1(1), 75– Baltimore: The Johns Hopkins University Press. 82. Parfitt, T. W., & Riley, S. P. (1989). The African debt Drazen, A., & Easterly, W. (2001). Do crises induce crisis. London: Routledge. reform? Simple empirical tests of conventional Roodman, D. M. (2001). Still waiting for the jubilee: wisdom. Economics andPolitics, 13 (2), 129–158. pragmatic solutions for the thirdWorlddebtcrisis. Dupuy, A. (1988). Haiti in the worldeconomy: class race Worldwatch Paper 155, Washington, DC: World and underdevelopment since 1700. Boulder: Westview Water. Press. Roubini, N., & Wachtel, P. (1998). Current account Easterly, W. (1999a). When is fiscal adjustment an sustainability in transition economies. NBER Work- illusion? Economic Policy (April), 57–86. ing Paper 6468. Cambridge MA: NBER. Easterly, W. (1999b). Life during growth. Journal of Sachs, J., & Warner, A. (1995). Economic reforms and Economic Growth, 4(3), 239–275. the process of global integration. Brookings Papers Easterly, W., & Levine, R. (1997). Africa’s growth on Economic Activity, 1–118. tragedy: policies and ethnic divisions. Quarterly United Nations Conference on Trade and Development Journal of Economics, November. (1967). The terms, quality, andeffectiveness of finan- Greene, J. (1989). The external debt problem of sub- cial flows andproblems of debtservicing. Report TD/ Saharan Africa. IMF Staff Papers, 36, 836–874. B/C.3/35. Geneva: UNCTAD, February 2. Humphreys, C., & Underwood, J. (1989). The external United Nations Conference on Trade and Develop- debt difficulties of low-income Africa. In I. Husain, ment (1983). Review of arrangements concerning debt & I. Diwan (Eds.), Dealing with the debt crisis. problems of developing countries pursuant to Board Washington, DC: The World Bank. resolution 222 (XXI), Paragraph 15. Report TD/B/ Husain, I., & Underwood, J. (Eds.) (1991). African 945. Geneva: UNCTAD, March 23. external finance in the 1990s. Washington, DC: The Van de Walle, N. (2001). African economies andthe World Bank. politics of permanent crisis, 1979–99. Cambridge: Iqbal, Z., & Kanbur, R. (Eds.) (1997). External finance Cambridge University Press. for low-income countries. Washington, DC: Interna- Van Wijnbergen, S., Anand, R., Chhibber, A., & Rocha, tional Monetary Fund. R. (1992). External debt, fiscal policy, and sustainable 1696 WORLD DEVELOPMENT

growth in Turkey. Baltimore: Johns Hopkins Uni- World Bank (1991a). Sub-Saharan Africa: from crisis to versity. sustainable growth. Washington, DC: World Bank. World Bank (1979). Worlddebttables 1979, Vol. I. World Bank (1991b). Worlddebttables 1991–92, Vol. I. Washington, DC: World Bank. Washington, DC: World Bank. World Bank (1981). Accelerateddevelopmentin sub- World Bank (1993). Worlddebttables 1993–94, Vol. I. Saharan Africa: an agenda for action. Washington, Washington, DC: World Bank. DC: World Bank. World Bank (1994a). Worlddebttables 1994–95, Vol. I. World Bank (1984). Towardsustaineddevelopmentin Washington, DC: World Bank. sub-Saharan Africa: a joint program of action. World Bank (1994b). Adjustment in Africa: reforms, Washington, DC: World Bank. results, andthe roadahead. Oxford: Oxford Univer- World Bank (1986). Financing adjustment with growth in sity Press. sub-Saharan Africa, 1986–90. Washington, DC: World World Bank (1998a). Global development ﬁnance 1998, Bank. Vol. I. Washington, DC: World Bank. World Bank (1988a). Worlddebttables 1987–88, Vol. I. World Bank (1998b). Assessing aid: what works, what Washington, DC: World Bank. doesn’t, and why. New York: Oxford University Press. World Bank (1988b). Worlddebttables 1988–89, Vol. I. World Bank (1999). Global development ﬁnance 1999, Washington, DC: World Bank. Vol. I. Washington, DC: World Bank. World Bank (1989). Worlddebttables 1989–90, Vol. I. Wynne, W. H. (1951). State insolvency andforeign Washington, DC: World Bank. bondholders: selected case histories of governmental World Bank (1990). Worlddebttables 1990–91, Vol. I. foreign bond defaults and debt readjustments. New Washington, DC: World Bank. Haven: Yale University Press.