A substantial part of this report focuses on the relationship between public debt and economic crises. This is not surprising, given the fact that Latin America’s history has been punctuated by devastating debt and financial crises. However, it should not be forgotten that one of the reasons that countries issue debt is to finance investment in both human and physical capital, which enhances long-run growth. So, in theory, the net effect of higher levels of debt on economic development depends on whether the positive effect of the investment that it finances is stronger than the effects of potential crises. In fact, countries may face a trade-off in which higher levels of external debt could potentially increase growth in the long run but also increase short-run volatility.
A better understanding of the relationship between debt and economic outcomes is also key to forming an opinion on the desirability of debt relief for poor countries. In fact, there are two standard arguments for debt relief. The first has to do with the composition of public expenditure, especially the need to free up resources to increase social spending on health and education and to address infrastructure bottlenecks. The second has to do with the growth-reducing effects of high debt. The first argument is straightforward. For a given path of government revenues, debt relief permits an expansion in public spending by easing a government’s budget constraint. In fact, in the enhanced Heavily Indebted Poor Countries (HIPC) initiative, countries receiving debt relief are required to use the resources saved as a result of debt relief to finance increased public spending in areas like health and education. The second argument is based on the idea that large debts have negative effects on private investment and hence depress growth.
With this question in mind, this chapter examines the interaction between debt and economic development, discusses how debt affects the composition and level of public expenditure, and reviews the economic literature on the desirability of debt relief. As the following discussion will show, the literature does not provide clear support for either of these two views. In fact, while there is no clear evidence that debt relief promotes growth and social outcomes across the board, there is some evidence that debt relief has had a positive effect in a restricted group of countries.
THE COMPLEX INTERACTION BETWEEN DEBT
AND ECONOMIC GROWTH
The reason that foreign borrowing by developing countries should have a positive effect on growth is straightforward. Almost by definition, developing countries tend to be capital scarce and hence have numerous unsatisfied investment needs with a potential return that is higher than the international interest rate.[1] In a perfect world, the funds borrowed by developing countries would be used to finance these investments with high returns, and this would not only make the country better off, but also generate the resources necessary to repay the debt. As a consequence, foreigners would profit handsomely from lending to developing countries and would thus have an incentive to provide these countries with all the capital needed to equalize returns to investment across countries.
Unfortunately, the real world is not perfect. Politicians may forgo high-return projects and overborrow to finance “white elephant” projects to satisfy their own egos or a particular constituency.[2] They may also try to maximize their probability of re-election by financing consumption booms (Chapter 9 includes a discussion of various political economy theories of debt and deficit). In the worst of cases, corrupt politicians may steal from the public purse.[3] This leads to a situation in which foreign lenders, who cannot perfectly monitor the uses of the financial resources they are supplying, may be subject to panic episodes and suddenly decide to stop lending. Thus, even countries that are indeed using foreign borrowing to finance high-return projects may be subject to destructive sudden stop episodes (Calvo, 2005c).
Furthermore, debt could have a negative impact on growth even if politicians do not misuse or steal money. Debt overhang theories show that high levels of debt may lead to lower levels of investment and, possibly, worse policies (Box 10.1). In addition, high levels of debt may reduce growth because they make countries vulnerable to herding phenomena and self-fulfilling crises regardless of domestic policies (Calvo, 2005c). In fact, one issue that has not been addressed by the literature on the link between debt and growth is the volatility channel. Interesting questions include the following: (1) Does debt increase output volatility? (2) Does debt favor long-run growth if a crisis is avoided? (3) Does debt favor long-run growth even after the higher propensity for crisis associated with debt is factored in? (4) If so, does the extra growth pay for the loss in welfare due to added volatility?
So much for the theory; what do the data say? Dijkstra and Hermes (2001) survey the early empirical literature on the relationship between debt and growth and find that there is no conclusive evidence in support of the debt overhang hypothesis. One problem with this early empirical literature is that it assumed a linear relationship between debt and growth and hence did not allow for the possibility that moderate levels of foreign borrowing can be associated with higher levels of growth (because they permit the financing of high-return investment projects) and that high levels of debt can be detrimental to growth through debt overhang effects. Recent work (summarized in Table 10.1) explicitly recognizes the possibility of a nonlinear relationship between debt and growth. Pattillo, Poirson, and Ricci (2002) and Clements, Bhattacharya, and Quoc Nguyen (2003) find that low levels of debt have a positive effect on growth, but that growth reaches a maximum at intermediate levels of debt (point A in Figure 10.1 is the debt overhang threshold) and becomes negative when debt reaches another threshold (point B in Figure 10.1). The key difference between these two studies is in the estimated threshold. While the former authors find that the growth-maximizing debt level is 10–20 percent of GDP, the latter authors (who use the same methodology but focus on a sample of low-income countries) find that growth reaches a maximum when debt is about 50 percent of GDP. This difference in turning points is extremely important for the debt relief debate. In fact, the results of Clements, Bhattacharya, and Quoc Nguyen indicate that the amount of debt reduction considered by the HIPC initiative will bring the beneficiary countries close to the debt levels that maximize growth, but the results of Pattillo, Poirson, and Ricci suggest that the debt relief offered by the HIPC initiative will not be enough to maximize growth in the beneficiary countries.
More recent studies suggest that the nonlinear estimations adopted by Pattillo, Poirson, and Ricci may be too simple to capture the complex relationship between external debt and growth. Cordella, Ricci, and Ruiz-Arranz (2005) show that there is both a “debt irrelevance threshold” and a debt overhang threshold. Focusing on the full sample of countries for which they have data, these authors show that the debt level that maximizes growth is about 20 percent of GDP (point A in Figure 10.2); at this point debt overhang kicks in and higher debt becomes associated with lower growth until a second threshold (the debt irrelevance threshold, point B in Figure 10.2) is reached; beyond this second threshold, there is no significant correlation between debt and growth. The debt overhang and debt irrelevance thresholds tend to be higher for countries with good policies and lower for countries with bad policies. In fact, it is not even clear whether debt is relevant to growth at all for countries with bad policies. Again, these results have important implications for the debt relief debate because they suggest that, for countries (like most HIPCs) that are situated in the debt irrelevance region, a moderate amount of debt relief (for instance, enough for them to move from point C to point C’ in Figure 10.2) would be completely useless. The implications of this study are thus that only a large amount of debt relief—enough to move a country away from the debt irrelevance area—can have a positive effect on the country’s growth rate.
Along similar lines, Imbs and Rancière (2005) find that there is no statistically significant relationship between debt and growth regardless of whether debt is low or high, indicating that there might be two debt irrelevance areas. These authors also provide direct tests for the channels through which debt may reduce growth and find that debt overhang episodes are indeed associated with lower levels of investment and worse policies.[4]
All the studies surveyed above focus on the relationship between external debt and GDP growth rather than on the relationship between total public debt and GDP growth. These are different concepts, because total external debt includes both public and private external debt but does not include domestic public debt. At the same time, total public debt does not include private external debt. Tanzi and Chalk (2000) are among the few authors who explicitly focus on total public debt.[5] They argue that there are several channels through which public debt may affect growth. The first is the standard crowding-out channel, through which debt leads to higher interest rates, which are in turn detrimental to private investment. Tanzi and Chalk test the crowding-out hypothesis using a sample of European countries and find that, contrary to what is predicted by Ricardian equivalence (see Chapter 1 for a discussion of this concept), higher levels of public debt do seem to be associated with higher interest rates and lower private investment. Similarly, back-of-the-envelope estimates for the U.S. economy show that, through crowding out, each dollar of debt reduces net output by approximately six cents every year (Elmendorf and Mankiw, 1999). This suggests that in the case of the U.S. economy, only projects that have a social return higher than 6 percent are worth being financed through the issuance of public debt.[6]
The second channel has to do with the fact that governments that face high levels of debt are often forced to reduce public expenditure drastically and often end up doing so by cutting the most productive but least politically costly component of expenditure, that is, public investment and operation and maintenance expenditure. Data for European countries show that higher levels of debt are associated with lower levels of public investment (Tanzi and Chalk, 2000), a result that does not seem to hold for Latin America (Lora, 2006). Financial repression is another channel though which public debt may affect growth. There are several examples in which countries with high levels of debt adopted tax structures that favored public debt and forced institutional investors to hold a large amount of public debt (Tanzi and Chalk, 2000).[7] Finally, countries with a large amount of public debt denominated in domestic currency are likely to be tempted to resort to higher inflation to reduce the burden of debt service. Whether or not such an inflationary policy is implemented, the fact that high levels of public debt generate inflationary expectations may lead to a climate of uncertainty that makes agents reluctant to engage in long-term contracts and translates into lower growth.
DEBT AND PUBLIC SECTOR INVESTMENT
IN PHYSICAL AND SOCIAL CAPITAL
The previous discussion highlighted the idea that debt may affect growth through both productivity and factor accumulation and suggested that higher levels of public debt can have a negative effect on the most productive components of public expenditure.[8] Unless it is totally wasteful, public investment in infrastructure is likely to be associated with the accumulation of physical capital, and social expenditure is likely to be associated with the accumulation of human capital.[9] Hence, it is relevant to consider whether there is a relationship between public debt and these components of public expenditure.
Public Debt and Investment in Infrastructure
While in fast-growing East Asian developing countries, investment in infrastructure ranges between 4 and 6 percent of GDP (Fay and Morrison, 2005), during the late 1990s, Latin America’s investment in infrastructure was hovering around 2.25 percent of GDP, about two-thirds of which was private and one-third public (Figure 10.3). It is interesting to ask whether these relatively low levels of public investment in infrastructure (PII) were driven by the high debt levels and recurrent debt crises that affected the region in the previous decade.
Although there are several studies that focus on the relationship between the composition of public expenditure and fiscal retrenchments,[10] work on the relationship between public capital expenditure and debt levels is more limited and tends to yield inconclusive results. While Mahdavi (2004) finds that external debt has an adverse effect on capital expenditure in a sample of 47 developing countries (the effect is not significant for the Latin American subsample), Clements, Bhattacharya, and Quoc Nguyen (2003) find that the stock of external debt has no significant effect on public investment. Instead, they find that higher levels of debt service (as opposed to the stock of external debt) crowd out public investment. All of these studies focus on capital expenditure, but none of them examines the behavior of PII. These are different concepts, because capital expenditures reported by standard cross-country data sets such as the International Monetary Fund’s Government Finance Statistics are an incomplete measure of actual PII, which in many countries is mostly carried out by state-owned enterprises or local governments, whose operations are not well captured by this source.
Lora (2006) looks at the experience of the seven largest Latin American economies and is the only study that explicitly focuses on PII. The most striking finding is that PII seems to be positively correlated with the stock of public debt. In particular, a 10 percentage point increase in the debt-to-GDP ratio is associated with an increase of PII of approximately 0.13 percent of GDP. While this effect may seem small in absolute terms, it is very large in relative terms, as it corresponds to approximately 10 percent of average PII in Latin America. Furthermore, the long-run effect is almost twice as large. There is also a strong negative association between fiscal balance and PII, implying that PII tends to decline during periods of fiscal adjustment, a result which is consistent with the findings of previous studies. PII also responds directly to changes in total primary expenditures, confirming that PII is susceptible to expenditure cuts (though not to changes in fiscal revenues).
A topic of intense debate is whether lending by international financial institutions has an impact on the level and the composition of public expenditure. In theory, if governments have access to international capital markets, multilateral lending does not necessarily increase public expenditure or alter its broad composition, though such lending may affect the quality and economic and social impact of public expenditure. However, when access to private external finance is limited, two opposing effects may operate. On the one hand, multilateral loans may finance projects that could not take place otherwise. On the other hand, the international financial institutions may make their support conditional on tighter fiscal policy and hence reduced overall expenditure. Lora (2006) finds that official lending as a whole has a minor and negative, though not statistically significant, effect on PII. However, when official lending is broken down by type of lender, it becomes clear that lending by the International Monetary Fund is associated with lower PII. There are two possible reasons for this finding. The first (supported by the critics of the IMF) is that this institution’s conditionality results in lower investment in infrastructure. The second (and more likely) reason is that the IMF tends to step up lending at times of crisis, which are also times when fiscal adjustments are required.
Lora (2006) further tests how public debt affects the composition of public expenditure and finds that an increase in the stock of public debt equivalent to 1 percent of GDP leads to a short-run increase of 0.15–0.18 percentage points in the share of PII in primary expenditures (or an increase of 0.22–0.30 in the long run).[11] These results point towards a symbiotic relationship between PII and public debt that may operate in the following way. Like other primary expenditures, PII increases when fiscal resources grow. However, when public debt is on the rise, PII is at an advantage vis-à-vis other primary expenditures, possibly because infrastructure is considered a more productive type of expenditure than, say, social expenditure, especially in the short to medium run, and possibly because there are legal and institutional constraints that tie debt to physical investment projects. In periods of fiscal consolidation, PII is adversely affected through a decline in expenditures and a reduced use of (or access to) credit.
On a note of caution, it is important to mention that while the above results suggest that higher levels of public debt are good for public investment, they also convey the more standard message that fiscal retrenchment is associated with lower PII. Hence, the fact that debt accumulation may eventually lead to fiscal retrenchment is likely to mitigate, or even reverse, the result described above. In fact, calculating the total effect of debt accumulation on PII would require a statistical model that jointly estimated how debt accumulation affects both PII and the probability of a future fiscal consolidation—a difficult task indeed.
Public Debt and Social Expenditure
Human capital accumulation is another channel through which public debt can affect economic growth, and if well used, social expenditure can promote human capital accumulation. The conflict between honoring public debt commitments and alleviating the lot of the poor is a recurrent topic among social policy activists and politicians in the developing world. For instance, at the World Social Forum held in Porto Alegre, Brazil, in 2002, participants claimed that external debt payments absorb a substantial amount of resources and that poor developing countries should stop repaying their debt. Funds previously earmarked for debt repayment should be redirected, the participants asserted, to finance socially just and ecologically sustainable development (Toussaint and Zacharie, 2002). Debt relief, either granted by lenders or obtained unilaterally through outright default, is often seen as an expeditious way to increase social public expenditure and improve the welfare of the poor. As argued by the World Bank and the IMF in support of the HIPC initiative, “debt relief can also be used to free up resources for higher social spending aimed at poverty reduction to the extent that cash debt-service payments are reduced” (IMF and World Bank, 1999). Economist Jeffrey Sachs (quoted in Fritschel, 2004) has gone even further: “No civilized country should try to collect the debts of people that are dying of hunger and disease and poverty.” These arguments resonate strongly in Latin America, where interest payments on debt absorb on average 2.8 percent of GDP, which would be enough to increase total social expenditures by 25 percent (Lora and Olivera, 2006).
Considering the attention that this issue attracts in public debate, it is striking how little empirical research has been devoted to assessing whether countries burdened with heavier debt commitments do indeed spend less in social sectors. While there are a few studies that look at the impact of fiscal adjustment measures on social expenditure,[12] these studies do not shed any light on the impact that debt and debt service payments may have on levels of social expenditure or their share in total expenditures. Lora and Olivera (2006) is the only study that aims at assessing the effects of total public debt (external and domestic) on social expenditure worldwide and in Latin America. Lora and Olivera find that higher debt ratios do reduce social expenditures, as popular opinion holds, but that the effect is rather small. Quantitatively, they find that an increase of 10 percentage points in the debt-to-GDP ratio is associated with a decline of approximately 0.15 percentage points in the ratio of social expenditure to GDP.[13] As the average social expenditure in Lora and Olivera’s sample is 6 percent of GDP, these estimates suggest that a 10 percentage point increase in the public debt ratio will reduce social expenditures by approximately 2.5 percent. Lora and Olivera also find that fiscal adjustments have a negative impact on social expenditure and that a one-dollar reduction in the overall or the primary fiscal deficit is associated with an average decline in social expenditures of around 3 cents in the current year (or nearly 5.5 cents in the long run). It is important to note that this is an average effect which varies widely depending on how fiscal adjustment is achieved. If primary expenditures are cut by one dollar, the decline in social expenditures may be as high as 13 cents, while if the same adjustment is achieved by raising more revenues, social expenditures may increase by 4 cents.
Lora and Olivera (2006) also show that higher levels of debt have a negative effect on the share of social expenditures in total primary expenditure. A 10 percentage point increase in the debt-to-GDP ratio leads to a decrease in the share of social expenditure in total government expenditure of approximately 0.5 percentage points (with long-run effects about twice as large). Social and other public expenditures behave in different ways. When total primary expenditures decline by 1 percent of GDP, the share of social expenditures in primary expenditures increases by nearly 0.4 percentage points (a finding consistent with previous literature showing that social expenditures are resilient in the face of fiscal adjustments). Ironically, this amplifies the negative effect of debt on social expenditure. In fact, an increase of one dollar in the stock of debt is associated with an increase of 4.9 cents in the primary balance and 1.3 cents in interest payments on debt in the following year. Hence, the net effect on the overall fiscal balance is an increase of 2 cents in the short run (or 3 cents in the long run). The typical response that produces the improvement in the primary balance is a mix of higher revenues (2.6 cents in the following year or 3 cents in the long run) and lower expenditures (2.5 cents in the following year or 4.4 cents in the long run).
Thus, following an increase in the stock of debt, governments typically react by reducing total expenditures and increasing total revenues by an amount beyond the increase in interest payments resulting from the increase in the debt stock, thus in general tightening the overall fiscal balance. In the process, social expenditures are hit twice, as they are sensitive not only to changes in total expenditures (and somewhat less to changes in revenues), but also to the direct impact of the stock of debt. A surprising finding is that increases in debt service payments (which may be the result of higher debt ratios) have only a minor effect on social expenditures. This suggests that debt displaces social expenditure not so much because it raises a country’s debt burden, but because it reduces the country’s space (or appetite) for further indebtedness.
Another interesting question is whether borrowing from international financial institutions makes a difference in regard to social expenditures. Loans from official sources in general, and from multilateral organizations in particular, do not seem to ameliorate the adverse consequences of debt for social expenditures. However, while increases in total official debt have no additional effect on social expenditures, different types of official lending have different effects on social expenditure. In particular, borrowing through bilateral lending and from the IMF attenuates the negative effect of debt on social expenditure, and borrowing from the other multilaterals amplifies this negative effect (Lora and Olivera 2006).[14] When social expenditure is measured as a share of total expenditure (rather than as a share of GDP), the effect of all types of official lending on social expenditures becomes insignificant, suggesting that the differential influence that each type of official lending has on social expenditures is due basically to how it influences total expenditures and not social expenditures directly.
As Latin America is often associated with macroeconomic instability and debt crises and has levels of social expenditures which are well below both the world average and the average for other developing regions, it is interesting to examine whether the previous results also hold for a subsample of Latin American countries.[15] Increases in debt stocks and in interest debt payments have much larger effects on social expenditures in Latin America than in the rest of the developing world. When debt stocks in Latin America increase by one dollar, social expenditures decline by 3 cents more than in other regions (where the decline is about 1 cent). For each additional dollar of debt payments in Latin America, social expenditures decline by around 23 cents (while in the rest of the world they increase by about 8 cents) (Lora and Olivera, 2006).[16]
Summing up, Lora and Olivera’s findings give credence to many of the widely held views about the deleterious effects of high levels of indebtedness. Higher debt ratios do reduce social expenditures, and not just because of the extra cost in interest payments (an effect that is especially important in Latin America), but because these higher debt ratios are associated with cuts in total expenditures that affect the social sectors.
These findings suggest that orthodoxy in debt management may be the best way to protect social expenditures. In fact, an improvement equivalent to 1 percent of GDP in the primary balance should initially cause a decline in social expenditures of 0.034 percent of GDP, but this initial reduction should be partly offset by an increase in social expenditures of 0.014 percent of GDP the following year because the stock of debt has fallen. In the third year, the initial reduction should be fully offset, and beginning with the following year, social expenditures should rise above the initial level (Figure 10.4). Furthermore, it should be possible to have social expenditures rising from the outset if the fiscal adjustment is based on an increase in revenues rather than on a reduction of expenditures, which is the reason why social expenditures fall in periods of fiscal retrenchment.
WHAT CAN BE LEARNED FROM THE DEBT RELIEF EXPERIENCE?
Supporters of debt relief argue that canceling the debt of the poorest countries has several potential advantages. First of all, it is an exercise in transparency, as several institutions are de facto granting debt relief by continuously evergreening past loans and refusing to recognize that some of their loans cannot be collected. Second, debt relief may create space for much-needed expenditure on poverty-reducing activities. Third, debt relief can kick-start growth in countries that suffer from debt overhang. Those who oppose debt relief argue, instead, that recipient governments use the resources freed up as a result of debt relief for wasteful activities and that debt relief makes it possible to perpetuate bad policies. Furthermore, providing debt relief to highly indebted countries may just end up promoting bad behavior and distributing resources to countries with a history of poor macroeconomic policies.
So far, this chapter has concentrated on the relationship between debt and economic outcomes, but it has not focused specifically on attempts to test directly the economic impact of debt relief (Boxes 10.2 and 10.3 provide a description of the main debt relief initiatives). A set of interesting questions on the impact of debt relief includes the following:
1. Has debt relief been successful in permanently reducing debt levels?
2. Has debt relief been successful in increasing GDP growth and social expenditure?
3. Is debt relief preferable to aid?
4. Does debt relief bring additional resources?
Has Debt Relief Been Successful in Permanently Reducing Debt Levels?
Past experience suggests that the answer to the question “Has debt relief been successful in permanently reducing debt levels?” is nuanced. Easterly (2002) studies the behavior of a set of countries that received substantial debt relief and finds that the net present value of debt service for these countries increased throughout the period during which these debt relief initiatives were implemented. How can increasing debt be reconciled with a sizable debt relief effort? A possible explanation is bad luck. However, Easterly finds no evidence that HIPCs suffered worse shocks than other developing countries. Easterly’s favored explanation is that high levels of debt are due to the presence of policymakers with a high discount rate (economic jargon to describe individuals who prefer to consume as much as they can today without worrying too much about what will happen tomorrow) who always borrow as much as they can. If higher levels of debt are due to the government’s high discount rate and if debt relief does not affect the government’s discount rate, then debt relief will be ineffective in the long run and lead to a merely temporary consumption boom.[17] Easterly argues that the data support the above interpretation. In particular, he studies the outcome of debt relief in 41 HIPCs and finds that, while debt relief over the 1989–1997 period amounted to US$33 billion, new borrowing by the countries receiving this debt relief was US$41 billion, consistent with the idea that debt relief leads to new borrowing.
However, a formal test leads to less striking conclusions. A regression of new borrowing on debt relief finds that for each 1 percent of GDP in debt relief, there is new net borrowing of 0.34 percent of GDP, suggesting that one dollar of debt relief leads to about 65 cents of effective debt reduction. Furthermore, the effect is not constant across periods. By splitting the sample into three periods, Easterly (2002) shows that debt relief was accompanied by increasing debt over the 1979–1987 period, constant debt over the 1988–1994 period, and decreasing debt over the 1995–1997 period. The latter result provides some evidence that the latest debt relief initiative may have been more successful than the previous ones, probably because HIPC debt relief is conditional on the implementation of macroeconomic policies that should lead to debt sustainability. It should be noted, however, that according to IMF reports, up to half of the countries that received debt relief under the enhanced HIPC initiative will soon return to an unsustainable debt situation (Birdsall and Deese, 2004).[18]
While Easterly (2002) favors the high discount rate interpretation, an alternative, and to some extents observationally equivalent, interpretation of these facts is that borrowing by poor countries remains high not because of the behavior of spendthrift politicians who waste resources, but because these countries have many unsatisfied basic needs. According to this interpretation, as soon as a debt relief program relaxes their budget constraint, politicians spend as much as they can trying to satisfy these unmet basic needs. Hence, money is not wasted but used trying to escape from a poverty trap; the problem is that there is too little money coming in, and hence poor countries cannot escape from their poverty traps. A policy consistent with this interpretation is to cancel debt and also provide more aid. In fact, Sachs’s (2005) proposal for ending poverty and the United Nations (2005) report on how to achieve the Millennium Development Goals make this point explicitly. Poor countries need to increase their expenditure on poverty reduction activities, and this can be achieved only through cancellation of their debt and increases in aid flows to them. A debt cancellation without an increase in aid flows would just result in an immediate buildup of debt—an interpretation consistent with Easterly’s empirical finding, but not necessarily an indication that debt relief is useless.[19]
Has Debt Relief Been Successful in Increasing GDP Growth and Social Expenditure?
In regard to whether debt relief has been successful in increasing GDP growth and social expenditure, the answer again is not so clear. Arslanalp and Henry (2005, 2006b) provide convincing evidence that debt relief is beneficial for countries that suffer from debt overhang. In particular, they show that the debt reduction brought about by the Brady Plan provided substantial benefits for both lenders and borrowers. However, they argue that, while debt relief may be beneficial for middle-income countries, HIPCs do not suffer from debt overhang, because in these countries the main obstacle to investment is not excessive debt but the lack of basic market institutions. Bird and Milne (2003) also argue that debt overhang should not matter for low-income countries, because this group of countries receives a positive net resource transfer. Furthermore, while high levels of debt can affect growth by increasing the probability of a financial crisis in the presence of volatile capital flows, this is an unlikely scenario for low-income countries, which have most of their debt with official creditors (Rajan, 2005a).
Depetris Chauvin and Kraay (2005) and Hepp (2005) provide empirical evidence on the effects of debt relief on growth and the composition of public expenditure. Depetris Chauvin and Kraay use a sample of 62 low-income countries during the 1989–2001 period. They find that (1) there is a positive but not statistically significant correlation between debt relief and GDP growth; (2) there is a positive and statistically significant correlation between debt relief and government spending in health and education, but this positive correlation is mostly driven by two outliers (Mozambique and Yemen); (3) there is a positive but not very robust correlation between debt relief and changes in policies; and (4) there is no significant correlation between debt relief and investment. Depetris Chauvin and Kraay interpret these findings as indicating that past debt relief efforts, which amounted to US$100 billion, were ultimately wasted, as they did not yield any concrete result. A more positive view would suggest that debt relief did not hurt any country and was beneficial in some countries.
Hepp (2005) shows that debt relief has a differential effect in HIPCs and non-HIPCs and provides support for Arslanalp and Henry’s (2005, 2006a) claim that HIPCs do not suffer from debt overhang. In particular, Hepp finds that, within the sample of HIPCs, neither debt service nor debt stock relief have any significant effect on growth. When he focuses on the sample of non-HIPCs, however, he finds that debt relief that leads to a one percentage point drop in debt service generates a 0.2 percent increase in GDP growth.[20]
One possible assertion is that the current debt relief wave (under the HIPC initiative and the Multilateral Debt Relief Initiative) is better designed and will have a bigger impact on growth than previous initiatives. World Bank (2006) evaluates the HIPC initiative and also finds modest progress in policy performance, growth, and poverty reduction but argues that lack of data makes it extremely hard to provide an evaluation of this initiative.
Is Debt Relief Preferable to Aid?
The question of whether debt relief is preferable to aid has not been systematically addressed by the empirical literature. Arslanalp and Henry (2006a) argue that the answer to this question depends on the type of country. In particular, they claim that debt relief is preferable in middle-income countries that suffer from debt overhang. However, they argue that in low-income countries that are part of the HIPC initiative, low growth is driven not by debt overhang, but by the lack of appropriate market institutions and economic infrastructure.[21] To support this claim, they show that HIPCs always have positive net transfers (a fact inconsistent with debt overhang) and that only a negligible part of capital flows to HIPCs goes to the private sector, indicating that international investors never considered these countries as having vibrant private sectors with viable investment projects. While some authors claim that (at least in the short run) countries care only about net resource flows,[22] Arslanalp and Henry (2006a) stress that debt relief and aid are different concepts and argue that aid is more efficient than debt relief in building market institutions. As aid might be crowded out by debt relief, they conclude that HIPCs should be the target of aid and not debt relief. Along similar lines, Rajan (2005a) points out that if the main obstacle to growth in a particular country is an impossible business climate, reducing the level of debt without providing additional resources or improving policies is unlikely to have any positive effect on growth in that country.
Birdsall and Deese (2004) agree that debt relief and aid are different concepts, but they present a completely different view. According to these authors, debt relief is one of the most effective forms of aid for at least five reasons. The first is the standard debt overhang reason. The second is that debt relief cannot be tied (“tied” refers to a situation in which aid donors force recipient countries to purchase goods or services from the donor country), and they point out that tying aid reduces the value of that aid by as much as 30 percent. The third reason is that debt relief stops defensive lending—lending that is not dictated by a country’s needs or the quality of its policies but by the level of its debt stock.[23] The fourth reason is that debt relief reduces the transaction costs of conventional aid programs because it liberates recipient countries’ government officials from satisfying the different needs and approaches to development of the various donor agencies (which implies that the marginal value added in the requests of these agencies is lower than the efforts that the recipient countries need to make in order to satisfy these requests).[24] The fifth reason is that debt relief provides flexible budget support and increases government accountability because it allows governments of recipient countries to set their own priorities instead of focusing on the pet projects of the various donors.
It should be clear from the above discussion that different views on debt relief versus aid depend partly on differing opinions regarding the value added that can be provided by donor agencies. Those who think that such agencies can increase the development impact of public expenditure by directing external resources towards the development of better institutions and infrastructure will tend to favor aid. Those who think that these agencies will only generate a useless bureaucratic apparatus and result in waste of resources will tend to favor debt relief.
Does Debt Relief Bring Additional Resources?
Systematic research on whether debt relief brings new money is limited. Birdsall, Claessens, and Diwan (2002) look at debt relief to African countries over the 1990s and conclude that because of poor data quality, it is hard to find solid evidence in either direction. However, their results provide some evidence suggesting that the debt reduction of the 1990s crowded out other forms of aid and hence did not provide any additional resources. Ndikumana (2002) finds that beneficiaries of debt relief received more aid than similar countries that did not benefit from debt relief. While this result points towards additionality of debt relief, Ndikumana also finds an overall decline in aid disbursements since the early 1990s, generating a situation in which beneficiaries of debt relief receive more net transfers than nonbeneficiaries, but not necessarily more than what they were receiving before becoming beneficiaries. Powell (2003) finds that debt relief neither crowds out nor generates additional resources. This, when considered along with the fact that economic aid to poorer nations has been decreasing overall, is consistent with the previous results that suggest no increase in net resource transfers to HIPCs. This point is also made by Arslanalp and Henry (2006a), who show that over the 2000–2003 period, net resource transfers to HIPCs were lower than those prevailing over 1980–1995 in terms of both recipient and donor countries’ GDP (Figure 10.5).
A World Bank (2006) evaluation of the HIPC initiative which uses more recent data provides a more positive view of the potential additionality of the initiative and shows that since 2000 there has been a substantial increase in net resource transfers to HIPCs. In particular, this evaluation finds that net annual transfers to the 28 decision point HIPCs increased from US$7.3 billion in 2000 to US$15.8 billion in 2004 and that more than half of this increase was due to debt relief. The same study also conducts a counterfactual exercise in an attempt to explore what would have happened without debt relief and finds that the HIPC initiative did bring additional resources in at least 17 of the 28 HIPCs considered.
THE RESEARCH AND POLICY AGENDA IS STILL OPEN
Well-used public debt can be a powerful tool for economic development. However, political distortions may lead to overborrowing, and volatile capital markets may lead to economic instability even with moderate debt levels. The main message of this chapter is that it is not easy to identify the relationship between debt and economic development, but there is some evidence that moderate levels of debt can promote growth, while higher levels of debt are likely to have a negative effect on growth. The problems with operationalizing these results is that, depending on countries’ economic and debt structures, “moderate” may mean very different things. In fact, one of the main themes of this report is that debt levels are only one—and probably not the most important—factor that determines debt vulnerabilities.
The chapter also explores the relationship between debt levels and the composition of public expenditure and does provide evidence suggesting that, while Latin American countries do borrow to finance investment in infrastructure,[25] higher levels of debt have a negative effect on social expenditure.
Finally, the chapter points out that there is still much that is not known on how debt relief affects economic growth and poverty reduction. The two key unanswered questions are whether debt relief is associated with more and better poverty reduction policies and higher growth and whether debt relief initiatives bring additional resources or only crowd out aid and concessional lending.
While lack of good data has been one of the main obstacles to detailed evaluations of past debt relief initiatives, one positive aspect of the HIPC initiative is that the coordinating role of the multilateral financial institutions and the need for these institutions to conduct internal evaluations of the initiative has greatly improved the information available to the research community. Furthermore, the high visibility of the initiative is providing incentives to conduct in-depth evaluations of debt relief not only among the multilateral financial institutions, but also in the academic and nonprofit communities. The recent World Bank (2006) update evaluation of the HIPC initiative is a welcome step in this direction and, although some of the results of this evaluation are inconclusive, it is important to note that this was also the case for the early evaluations of the Brady exchange (Husain and Diwan, 1989; Fernández-Arias, 1993), which is now considered a very successful debt relief program.
