External Debt Sustainability: An Extended Framework for HIPCs?

Transcription

1 External Debt Sustainability: An Extended Framework for HIPCs? Marco Arnone, Luca Bandiera, and Andrea F. Presbitero Abstract This paper proposes an extended framework on external debt sustainability to address the main shortcomings of the standard analysis, after presenting a review of the literature together with the IMF-World Bank HIPC Initiative. We argue that a comprehensive debt sustainability analysis requires a fully-fledged government budget constraint, which includes not only the external position, but also public domestic debt and the feedback effects of the choices on deficit financing. We also suggest the adoption of country specific debt thresholds, debt service thresholds, and the evaluation of the risk of default, rescheduling and arrears. Keywords: HIPC Initiative, Debt Sustainability, Debt Relief, External Debt. JEL Classification Numbers: F34, H63, O11, O19 Authors Address: marco.arnone@yahoo.com, lbandiera@worldbank.org and a.presbitero@univpm.it Affiliations: Marco Arnone is at the Catholic University of Milan (Italy) and CeMaFiR, Andrea Presbitero (corresponding author, Tel: 071/ ) at the Università Politecnica delle Marche (Ancona - Italy) and CeMaFiR; Luca Bandiera - Catholic University of Piacenza (Italy) and CeMaFir - is a consultant at The World Bank. This paper was started when Marco Arnone was an economist in the IMF s Monetary and Financial Systems Department, and this version was finalized when A. Presbitero was a Summer Intern at The World Bank. The authors wish to thank M. Baussola, S. Bosi, L. Campiglio, T. Cordella, M. Mazzoli, V. Nehru, L. Papi, J.F. Perrault, D. Scalise, M. Sommer, F. Timpano, V. Tulin, and M. Vivarelli for helpful comments. Opinions expressed reflect only those of the authors. Remaining errors are ours alone.

2 1. Introduction and Summary The High Indebted Poor Countries (HIPC) Initiative launched by the IMF and the World Bank deals with long-term debt sustainability and poverty reduction, so as to prevent debtors to re-accumulate high level of public external debt, insure a permanent exit from debt dependence and provide budgetary support to implement the poverty reduction strategies of the HIPCs. This paper reviews different approaches to external debt sustainability, and proposes a more comprehensive approach to assess the sustainability in the HIPCs. According to some authors, debt relief seems to have failed to reach its targets, because the Debt Sustainability Analysis (DSA) used in the Initiative is weakened by major drawbacks. We present the theoretical literature on debt sustainability - with particular attention to the IMF-World Bank approach (Annex A) - and the effects that a large external debt and fiscal deficit have on economic growth in low-income countries (LIC). The accounting approach used in current policy analysis embodied in the criterion of financial sustainability - presents several shortcomings. The most important drawbacks of the standard DSA are the endogeneity of the relevant variables and the exclusion of domestic debt. Some recent models develop a sophisticated analysis, which includes some of the main criticisms received by the DSA used in the HIPC Initiative. These models represent a substantial improvement and could be considered as a starting point for a more comprehensive approach, even if they present some limitations, namely the exclusion of the current account, domestic debt, and/or lack of feedback between fiscal and current account. The design of the HIPC Initiative is based on the adverse effect of large external debts on investment and economic growth. We argue that the theoretical arguments namely the debt overhang and the liquidity constraint should be reconsidered and adapted to the specific experience of the HIPC countries. The empirical evidence on the debtgrowth nexus lacks of robustness. However, even if a standard debt overhang effect does not seem to work in poor countries, external debt could still affect growth because of the uncertainty and macroeconomic instability associated with high public debts ( extended debt overhang). The lack of structural reforms, short-termism and misallocation of resources could further reduce the efficiency of investments and growth (Presbitero, 2006). We underline the importance of institutions and policies, so that DSA and debt relief should not be the same for each country, but tailored on each country s specific macroeconomic and institutional framework. Furthermore, we stress the potential risks associated with the rapid development of domestic debt in LICs: an increasing debt service soaks up resources from the government budget and diverts money from development- and growth-oriented programs, while the rise in internal interest rates impacts negatively on investments. Thus, we call for country-specific debt service thresholds. As a result, a comprehensive framework for debt sustainability requires a fully-fledged government budget constraint, which includes not only the external position, but also domestic debt and the feedback effects of the choices on deficit financing. We propose 2

3 that an extended debt sustainability framework should include the four following aspects: 1. Country specific debt thresholds, according to the findings of the debt overhang theory and to the relevance of institutions and policies. 2. Debt service thresholds, in order not to crowd out public or private investment. 3. The evaluation of the risk of default, rescheduling and arrears, to reduce macroeconomic instability. 4. The inclusion of domestic debt, which have feedback effects on future public deficits, monetary policy and inflation, and the external financing gap. The remainder of the paper is as follows: Section (2) illustrates different concepts of debt sustainability and the different ways in which it is assessed, together with their limitations. Section (3) presents the models of debt sustainability that try to overcome some of the criticisms received by the standard DSA. Section (4) reviews the theoretical effects of a large external debt on economic growth, with specific attention to HIPCs experience and the related empirical evidence. Section (5) is about the rising role of domestic debt as source of deficit financing in poor countries. Section (6) presents our proposal for a comprehensive debt sustainability framework. 2. Different Approaches to External Debt Sustainability External debt sustainability is a widely debated issue in the theoretical and empirical literature, which presents different approaches, depending on the economic targets and on the consideration of lender and borrower behaviour. A possible classification of the different perspectives is as follows: Optimising models: here the marginal benefit equals the marginal cost of borrowing. This approach was the first one developed in the theoretical literautre (for a survey, see Eaton (1993)). Non-optimising models: growth-cum-debt model and debt dynamics approach. In the growth-cum-debt model, external borrowing fills the gap between domestic savings and investments, as in the two-gap model (Chenery and Strout, 1966). The condition to be solvent requires that the rate of growth of the economy must be greater than the rate of interest (the cost of borrowing). One important shortcomings is the lack of attention to the foreign exchange issues. The debt dynamics approach looks at external solvency and export growth; the latter must be greater than interest rates. These models remain unsatisfactory, because the growth paths are assumed to be time invariant, and imports are not considered. Fiscal space models: they stress the adverse effect of debt service on public expenditures: lack of infrastructure and public spending has an adverse effect on private investments, and the result is a slower growth. Disincentive effects: a large stock of debt undermines economic performance through the debt overhang effect, which is related to tax disincentive and macroeconomic instability owing to: (1) exchange rate 3

4 depreciation, (2) increases in fiscal deficit, (3) monetary expansion and inflation and (4) uncertainty deriving from exceptional financing. Large part of the theoretical and the empirical analysis has focused on the capacity of a debtor country to service its debt, ignoring the effect that debt and deficit have on other economic variables and development objectives. A broader classification distinguishes between a simple debt capacity analysis (financial sustainability) and a more complex view that involves the assessment of some basic development targets (economic sustainability). In the following sections, we present the theoretical background of the financial sustainability approach. Its limitations call for a more comprehensive approach in favour of a development perspective, based on the adverse effects of large external debt on a developing economy Financial Sustainability Following the so-called accounting approach (or borrower-based approach), a fiscal deficit is considered sustainable if it generates a constant debt-to-gdp ratio (Cuddington, 1996). The level of primary surplus (or deficit) which stabilizes the debtto-gdp ratio the parameter b in (1) is given by: rt g t (1) SURPt = b 1 + g t where r is the real interest rate and g is the rate of growth of GDP: as long as the economy grows at a rate higher than the interest rate, it is possible to run a sustainable primary deficit. What really matters is the capacity of the government to raise revenues that could balance the expansion of the stock of debt. Since in LICs the grant elements is a substantial fraction of GDP - f in (2) - Cline (2003) argues that the previous condition should be less stringent, and defines the primary balance as: (2) pb = (r-g)b f. He estimates 1 that HIPCs do not have to run any primary surplus, but they can incur a primary deficit equal to 7.2 percent of GDP, without boosting the debt-to-gdp ratio above the current level of 106%. The accounting approach is at the root of the DSA adopted by the HIPC Initiative. It measures the ability to meet current and future external debt service obligations: it is based on the determination of thresholds of solvency, based on the ratios of Net Present Value 2 (NPV) of Public and Public Guaranteed (PPG) external debt to exports and to fiscal revenues. 1 He assumes growth rate of 3%, grants equal to 4.5% of GDP, and real interest rate of 0.5%. 2 The NPV of debt is the sum of all the future debt service obligations interest and principal on existing debt, discounted at market interest rate. Its use is justified because it takes into account the degree 4

5 In the Present Value Constraint (PVC) approach (or lender-based approach) a government is solvent if the flow of expected value of future resources is at least equal to the face value of the stock of debt: (3) B 0 = SURPt + = t 1 ( 1 r) t where B is the initial stock of debt, r is the real interest rate and SURP are the future surpluses, given by the difference between revenues and public expenditures. This criterion is different from the accounting one, which imposes an upper bound to the debt-to-gdp ratio. Under the usual assumption of r greater than GDP growth, the PVC requires only the real growth rate of debt lower than real interest rate. If the rate of growth of debt is between the real interest rate and the rate of growth of GDP, the PVC is satisfied, but the debt-to-gdp ratio can grow over time. The focus is on what level of deficit can be financed. As a consequence, this analysis does not require the full repayment of the entire debt. Eventually, Cohen (2001) suggests a market value approach to include also the risk of non-payment. Thus, the market value of the stock of debt is smaller that its face value and its present value, because it takes into account arrears, rescheduling and constrained refinancing of various sorts (Cohen, 2001: 364). Donors should consider the distinction between nominal and actual debt cancellation. In this approach, part of the debt reduction which is generally considered new aid should be, instead, a reported loss Limitations of the financial sustainability approach The financial sustainability framework has been subject to many criticisms, which encompass a wide range of issues: The framework assumes that HIPCs can borrow only on concessional terms; no attention is paid to domestic debt dynamics and the substitution effect between external and domestic debt 3. Loser (2004) stresses the need for an explicit consideration of an adequate level of indebtedness as a function of the specific institutional and economic characteristics of a country. The HIPC framework does not sufficiently consider the current account balance and foreign exchange constraint. Goldstein (2003) stresses the importance of the exchange rate behaviour for emerging economies, especially when public debt is denominated in foreign currency 4. of concessional lending: when the interest rate on a loan is below the market rate, the NPV of debt is lower than its face value. 3 With respect to the relevance of domestic debt, see Martin (2004), Arnone and Presbitero (2005) and Reinhardt et al. (2004). 4 The new debt sustainability framework for LICs recognises the importance of exogenous shocks on the capacity of repaying external debt. However, shocks are evaluated only in a partial equilibrium, which does not consider feedback effects. 5

6 The use of the ratio of NPV of debt to exports is widely considered inadequate as a sustainability criteria, because exports are highly volatile. Also, considering the ratio of affordable debt service to revenue as the single solvency criterion (Gunter, 2002 and Martin, 2002) is subject to criticisms 5. As a results, sustainability needs to be determined by a borrowing level that is consistent with a sustainable external balance and a sustainable government budget (Hanmer and Shelton, 2001: 5). The implied exogeneity of the main policy variables and great reliance on projections. The repayment capacity is a function of the rate of growth of GDP, of exports, of fiscal revenues and of the interest rates, which are, to some extent, influenced by the level of indebtedness 6. Gunning and Mash (1998) stress the centrality of forecasts and underline also the role of expectations which generates uncertainty and make predictions less reliable. The unfeasibility of the adjustments required in order to reach sustainability (Roubini, 2001). In order to stabilize the debt ratio to a certain threshold, a country could have to run a trade (or primary) surplus that, given the expectation on interest rates and economic growth, is practically unfeasible. Goldstein (2003) agrees with the fact that standard analysis ignores that some targets are literally impossible to deliver, given the volatility and the reliance on market sentiment of interest rates, exchange rates and inflation. As a result, the simple accounting framework does not necessarily ensure sustainability, since, once a certain NPV of debt to export is achieved, it will not be necessarily maintained in the long run. Many limitations of this sort of analysis are recognized also by the World Bank Operations Evaluation Department, which stresses that the debt projections are likely to be biased by the assumptions and the methodology used in forecasting (Eaton, 2002), which does not consider the effects of high external debt and of the required macroeconomic adjustments on growth. More generally, the accounting approach could overcome those problems, but it is weakened by two important shortcomings: (1) the real interest rate and the rate of growth of GDP are generally considered exogenous, while instead they are affected by the government spending, and (2) the assumption about the growth rate of liabilities, which ignores the role of lenders, since HIPCs are expected to keep on borrowing at very high concessional terms Economic sustainability The development perspective criticizes the simple framework above and proposes two different channels to improve it: in the first place, the development perspective investigates the relationships between fiscal deficits, interest rates, economic growth, inflation and balance of payments in order to address the endogeneity of these variables; 5 See Birdsall and Deese (2002). They point out that such a scheme is an incentive for recipient governments to reduce their tax collection effort. 6 This issue of endogeneity is raised, among other, also by Loser (2004) and by Goldstein (2003), who suggests considering the negative feedback that high interest rates and a tight fiscal policy have on growth. 6

7 secondly, it considers a broadening of debt forgiveness targets by taking into account the amount of resources needed to reach specific targets of growth and poverty reduction (like the Millennium Development Goals). A prominent supporters of this view is Sachs (2005), who advocates higher consideration of the financial requirements for basic social and health expenditures. His point of view is widely reflected in a variety of NGOs documents. The WB-IMF analysis, on the contrary, does not explicitly calculate the amount of resources that will be redirected to specific sectorial targets to spur economic growth and to reduce poverty after debt relief 7. Resources freed up by debt relief are assumed to be spent on health, education, nutrition and infrastructures according to government plans. The lack of specific sectorial expenditure targets increases uncertainty about economic growth; therefore, debt sustainability cannot be assured. Some authors (Martin, 2002) move the focus from the debt ratios to the debt service payment, which is a more direct indicator of the resources soaked up by external obligations. Eurodad (2001) calls for a poverty approach to debt sustainability - emphasizing the resources necessary to satisfy essential human needs and to domestic debt (prioritized with respect to the external debt to favour internal macroeconomic stability). Finally, a country should dispose of enough resources to serve its debt obligations and improve economic performance. The IMF and WB have emphasised economic growth as a necessary condition to achieve sustainability (IMF, 2002a: 22-23): While the existing stock of debt (and associated debt service) sets the point of departure for determining long-term debt sustainability, the growth of income, exports, and fiscal revenue which, to a large extent, reflect a country s economic policies are the underlying determinants of the evolution of a country s capacity to service external debt over longer term. 3. Empirical Models of Debt Sustainability There are only few models analysing debt relief and sustainability under a broader macroeconomic framework than the accounting approach. Edwards (2002, 2003) looks at the experience of Nicaragua: in two different analyses he investigates the relationships between debt relief and fiscal sustainability, and debt relief and current account. Edwards (2003) defines macroeconomic sustainability in terms of two key features: (1) fiscal sustainability and (2) current account sustainability. He develops two different dynamic models of sustainability: the first one has a primary balance consistent with a stable debt-to-gdp ratio as a target, while the second a current account consistent with solvency. The main advantages of these studies are the consideration of domestic debt, the distinction between concessional and commercial debt, the inclusion of remittances, aid 7 In the HIPC framework, the use of HIPC savings is monitored only during the interim period, that is, before debt relief becomes irrevocable. It must be stressed, however, that HIPCs have to prepare a Poverty Reduction Strategy Paper in consultation with civil society, and this paper is considered key to stimulate ownership. Trying to impose specific constraints on so many items of the government balance sheet could be seen as an undue interference with (and lessening of) a country s sovereignity. 7

8 and grants, as well as seignorage revenue, which are key elements for developing countries. In both cases, he finds out that sustainability requires strong macroeconomic adjustments (reduced expenditures, higher taxes, devaluation), which are politically difficult to implement and which can divert resources from poverty-reducing and growth-enhancing programs. The main drawback of these works is a lack of integration between the two analysis: in particular, the adjustments demanded for current account sustainability are likely to undermine fiscal stability. The sensitivity of the results with respect to GDP growth underlines the feedback of macroadjustment on growth. Fedelino and Kudina (2003) find that the HIPC Initiative does not assure fiscal sustainability, unless fiscal policies change. They start from the Edwards approach to fiscal sustainability and modify the budget constraint in order to include: (1) the exchange rate, and (2) the distinction between domestic and foreign debt. Their model implies that donor should provide more grants and aid flows to finance the required (and expected) increased poverty-reduction expenditures, considering that tighter fiscal policy in the HIPCs is not viable. The authors conclude that Unless HIPCs improve their primary fiscal positions or grant financing is sustained at current, or possibly higher, levels, debt sustainability in HIPCs may prove elusive in the long term (p. 26). This model has the advantage of including domestic debt into the analysis, but has the same limitations that affect Edwards models. In particular, it takes the crucial economic variables as exogenous and, then, it derives the effect of a change in those variables 8. Burnside and Fanizza (2004) look at the effects that debt relief has on the government budget constraint and on output growth. They analyze the HIPC initiative with a different perspective: the sustainability of debt and external sector are not addressed, but this model has the advantage of looking at the effect of debt reduction on the economy, without assuming a growth rate target. Government spending, investment, tax revenues and output change in response to debt forgiveness, to the level required by conditionality and ex ante spending decisions. The inclusion of a money demand function allows for the simulation of two different scenarios, depending on monetary policy. If the central bank pursues stable inflation, the outcome is a small reduction in the long run level of debt and inflation; while a loose monetary policy results in substantial debt reduction, at the cost of a higher inflation in the short run. These results confirm the unpleasant monetarist arithmetic highlighted by Sargent and Wallace (1981): lower current inflation implies higher inflation in the long run, or, in this case, lower inflation in the future requires a higher rate of current inflation. However, that conclusion could be mitigated by fiscal reforms: if government increases spending in poverty reduction programs and cuts other expenditures, there are sizeable gains in terms of permanent reduction of debt and inflation, both with passive and active monetary policy. The relevance of the growth effect denotes the importance of the effectiveness of poverty reduction programmes in raising growth for fiscal sustainability. 8 However, as the authors themselves admit, interest rates, inflation, exchange rates and economic growth should be endogenous, because they are clearly affected by debt relief and by lenders behaviour. 8

9 Even if sustainability is not explicitly considered, this model allows for a broad macroeconomic evaluation of debt reduction. The evolution of the debt stock and the government budget constraint consents to assess the likelihood of a future debt distress. The main limitations of this model concern: (1) the disconnection between the monetary and the real part of the model, given that inflation, completely determined in the money market, is not allowed to impact on output, and (2) the exclusion of domestic debt, of the current account and of real exchange rates. A comprehensive framework that addresses these issues requires a fully-fledged government budget constraint that includes both the external and domestic position, and the exchange rate, in order to have a more realistic assessment of debt relief and of its impact on the economy. 4. The Macroeconomic Effects of External Debt on the Economy The most important determinant for debt sustainability and poverty reduction is economic growth. Whichever is the definition of sustainability, an increase in GDP growth increases the level of sustainable debt. However, the debt-growth nexus is a two-way relationship, since a large external debt could be a symptom of low growth, but also a cause of it: this section presents the theoretical arguments and empirical evidence of the effect of debt on growth. Claessens et al. (1996) provide an accurate definition of the linkages between debt and economic performance: Disincentive effects Debt overhang theory: when the debt burden is too high (the debt exceeds the future repayment ability), the expected debt service is a positive function of the aggregate output. This squeezes investment, because the returns are taxed away by foreign creditors. The stock of debt has another negative effect on economic performance, due to the degree of uncertainty on the fraction of debt service that will be repaid with the country s own resources as outcome of debt rescheduling negotiations. Cash flow effects Debt reduction could facilitate access to the international financial markets and repatriation of flight capital (increasing private investment and the efficiency of the investment selection). The liquidity constraint 9 : a reduction in current debt service increases the current level of investments, for any given level of future indebtedness. It is different from debt overhang, which is focused on the stock of debt. If there is no debt overhang, the same increase in investment can be reached with a new loan or with a reduction in current debt service. Moral hazard effect 9 The adverse effect of external debt service on economic performance and investment is widely called crowding out effect, especially in the empirical literature (see, among others, Clements et al. (2003) and Pattillo et al. (2002)), since repayments soak up resources and reduce public investment. 9

10 With debt relief, debtors might believe that creditors have eased their position and that they will be more willing to forgive any future debt, when the likelihood to obtain a repayment will become low. Conditionality through refinancing and rescheduling could reduce moral hazard: this is an advantage over upfront debt reduction. Economic literature has generally investigated and tested empirically three main channels through which external debt affects growth: (1) the debt overhang effect, (2) the uncertainty effect, and (3) the liquidity constraint effect 10. The negative effect of debt on growth works not only through the impact of the stock of debt, according to the debt overhang hypothesis, but also via the flows of service payments, which are likely to crowd out public investment. It is worth noting that this strand of literature (Krugman, 1988, Sachs, 1989 and Cohen, 1993) was developed as a response to the Latin American debt crisis of the 1980s, which affected Middle Income countries, whose debt was contracted mainly with private creditors. Conversely, the current macroeconomic environment and the debt structure of many indebted countries is nowadays very different, so that the validity of these theories should be carefully questioned and the theoretical arguments have to be adapted to the specific HIPC experience. The HIPCs, contrary to the indebted countries in the 1980s, are all Low Income countries, mainly located in Sub-Saharan African, without market access and highly dependant on concessional external lending. Net positive resource transfers, the lack of sudden stops in external assistance and the continuous process of debt rescheduling and restructuring are likely to reduce the disincentives to invest embedded in the debt overhang hypothesis. Therefore, a decreasing level of debt service payments is the crucial determinant for investment and growth 11. The theoretical background of the debt overhang hypothesis can be used for a broader interpretation of the negative effects of debt on growth ( extended debt overhang) - namely, the disincentive effect of a high stock of debt on other type of investments (i.e. human capital) and on the government s willingness to undertake structural reforms and fiscal adjustments. Sachs (2002) supports an extensive interpretation of debt overhang effect in a model that shows how an excessive debt burden could lock low income countries in a poverty trap 12. This model supports the idea of a fresh start for the HIPCs and a more comprehensive idea of debt sustainability, which should explicitly consider the investment requirements in health, education and infrastructures. A third channel through which large external debt could affect economic performance concerns the uncertainty about future resource inflows and about debt service payments, together with their effects on macroeconomic stability. 10 See, among others, Addison, Hansen and Tarp (2004), Bhattacharya and Clements (2004), Hanmer and Shelton (2001) and Elbadawi et al. (1997), who add the lack of access to international financial markets. 11 Hansen (2004) calls for more additionality because both investment and growth rates are significantly explained by debt service and effective aid. His empirical results show that if aid and debt payments are reduced one for one, there is no effect on economic growth and investment could fall. 12 A poverty trap could derive from the fact that, in absence of public investment in human capital and basic infrastructure, also private investment are unlikely to be made. 10

11 Risk of default, rescheduling and arrears are likely to increase the volatility of future inflows and additional lending, while access to capital markets crucially depends on the perceived sustainability (Gunning and Mash, 1998). This generates an uncertain environment, where government policies and reforms depend on conditional lending and rescheduling. In this context, domestic and foreign investors are likely to exercise the waiting option, even when fundamentals are improving (Serven, 1996). Moreover, investment decisions under uncertainty may not be forward-looking: short-term, lowrisk investments might be the preferred option. This misallocation of resources reduces efficiency and productivity of capital, leading to a slowdown of economic growth, because of the extended debt overhang. Thus, debt relief could reduce uncertainty and increases confidence in the debtor country s government and its policies, fostering growth. This argument is similar to the debt overhang hypothesis, but here the focus is on the general uncertainty that dominates the economy and biases the investment choices, in terms of misallocation and withdrawals. Claessens et al. (1996) argue that one of the conclusions that could be drawn from the experience of the Brady Plan is the crucial importance of reduced uncertainty. The removal of continual ongoing rescheduling was the main channel through which debt reduction affected growth. Debt reduction has the effect of increasing capital inflows by strengthening confidence in the government. This happens because investors look at debt reduction as an endorsement by the international financial community that the country is successfully pursuing sound macroeconomic policies and structural reforms (Claessens et al. 1996: 34). Presbitero (2006) argues that if debt reduction goes hand in hand with selectivity and structural and economic adjustments, it could even be viewed by the international community as a positive signal, so that it might stimulate foreign investment, macroeconomic stability and growth. Notwithstanding these theoretical arguments, the empirical evidence on debt overhang is not conclusive, especially because econometric results lack robustness and do not address explicitly the direction of causality. In particular, empirical works should focus on a more accurate investigations of the real effects of indebtedness on economic performance in the HIPCs (or LICs), instead of dealing with all developing countries, since high debt is likely to affect growth through different channels, depending on the specific macroeconomic environment. Furthermore, in order to draw a more realistic estimate of the impact of debt dynamics on economic performance, there should be a careful analysis of the debt effects on investment (and not only on economic growth), and also of the relevance of a policy or institutional variable, which is often neglected and could be a common determinant of both low growth and high debt 13. Some empirical results (Elbadawi et al., 1997, Pattillo et al., 2002 and 2004, Clements et al., 2003) are consistent with the presence of non-linearities in the debt-growth nexus, supporting a so-called Debt-Laffer curve, while others disagree (Cohen, 1993, Chowdhury, 2004). Two very recent papers, Imbs and Ranciere (2005) and Cordella et al. (2005) deepen the analysis of the debt-growth nexus. The former paper uses a non parametric technique to support the bell shaped curve, arguing that better institutions 13 A more comprehensive review of the empirical literature on the debt-growth nexus is in Arnone et al. (2005). 11

12 reduce the magnitude of the debt overhang. The latter suggests that the relation between debt and growth is a modified Debt-Laffer curve because, over a certain threshold, the debt effect on growth is nil, creating a sort of debt irrelevance zone, so that the debt overhang is a valid hypothesis only for non-hipc countries. In this way, these findings support the hypothesis that positive net transfers reduce the disincentive to invest, so that the debt overhang theory has to be reconsidered or extended, in order to include the role of instability. Presbitero (2006) argues that the basic relation between debt and growth is linear and negative 14 and shows that external debt does not reduce the level of investment. However, he argues for the presence of an extended debt overhang effect, which works in terms of misallocation of capital, short-termism, lack of structural reforms, and subsequent lower efficiency. Furthermore, the positive and strong impact of the institutional indicator requires to give great emphasis on economic policies, governance and structural reforms in poor countries. The empirical findings on the liquidity constraint 15 are instead less controversial and they corroborate the effectiveness of the crowding out effect, even if Pattillo et al. (2002, 2004) find debt service not significant as an explanatory variable in their growth regression 16 and Clements et al. (2003) estimate that the magnitude of this effect is quite weak, suggesting that debt relief should not be expected to provide a sharp increase in public investment. 5. Deficit Financing and the Role of Domestic Debt A relevant element, ignored by the HIPC Initiative, is the role played by domestic debt in poor countries. A full-fledged macroeconomic analysis has to include domestic debt dynamics. Considering that participation in the HIPC Initiative forbids borrowing on commercial terms in the international capital markets, and that monetary financing of the deficit is normally prohibited in IMF programs, internal financing is becoming more and more important in many countries. This trend is not likely to be reverted any time soon. At least in the medium term, large primary deficits in HIPCs cannot be promptly reduced because of low revenues and the inherent political difficulties in reducing public spending in countries where most of the population is already below the poverty line. The creation of a domestic market for government securities could stimulate the development of deep and liquid internal financial markets and protect countries from adverse external shocks. On the other hand, this has to come after the achievement of a certain degree of macroeconomic stability, in terms of credible fiscal and monetary policies, and financial markets liberalization (Arnone and Ugolini, 2005; Del Valle and Ugolini, 2003). If this is not the case, the benefits of domestic debt are likely to be offset by adverse effects in terms of higher interest rates, and crowding out of private 14 However, he recognizes that this link could become less strong or even not significant when debt is too large, so that there might be a debt irrelevance zone, coherently with the findings of Cordella et al. (2005). 15 See Cohen (1993), Chowdhury (2004) Hansen (2004) and Presbitero (2006). 16 This result, contrary to the expectations (as admitted by the authors themselves), could be due to the fact that the data do not measure the actual payments, but rather the scheduled ones. 12

13 investments. The unfavorable evolution of domestic debt could be worsened by a reduction in foreign lending. This scenario could be made at all possible by the recent pronuncements of the US administratios in favor of a move towards grants instead of loans, and endorsed by the new World Bank presidency. With respect to African countries, Beaugrand et al. (2002) look at the choice between domestic and external financing and stress the advantages of internal borrowing in terms of financial costs and fewer risks, even considering the possibility of currency devaluation. Thus, a limited recourse to domestic market is still recommended, because it could help mobilize domestic savings and develop financial markets. Christensen (2004), instead, underlines that: 1. Notwithstanding its relative small dimension, domestic debt is a severe burden to the government budget because of high interest payments; 2. The stock of domestic debt is increasing over time; 3. Domestic debt could squeeze private investment because credit constraints; 4. The short maturity structure poses high rollover risks and increases macroeconomic instability. The analysis of the recent trend of domestic debt 17 and interest rates in the last decade in 14 HIPC countries confirms this scenario: (1) domestic debt, even if it is still a small fraction of total debt, is increasing and, more important, (2) interest payments on domestic debt show a more dramatic upward trend. In the last decade, domestic debt has almost doubled in HIPCs, reaching more than 18 percent of GDP. A similar pattern is observable in the ratio of interest payments over GDP (Figure 1), which followed a pretty stable path in the nineties and then increased dramatically in the last three years, reaching almost 3 percent of GDP: this reflects the fact that, in order to raise money on domestic markets, governments have to offer high interest rates. As a matter of fact, the comparison between the domestic and external implicit interest rates 18 shows that domestic financing is very costly compared to multilateral concessional loans. After 1997 the cost of internal borrowing stabilized at around 20 percent, almost ten folds higher that the implicit interest rate on foreign lending. A stable cost of external financing, associated with a rising recourse to domestic debt, results in the observed dramatic increase in the expenditure for interest payments. In the countries that have reached Completion Point 19, however, the cost of domestic financing declined also after 1997, even if it is still at 12 percent in The great 17 A more detailed analysis of domestic debt in HIPCs is in Arnone and Presbitero (2006). For reasons of comparability, we define domestic debt as the central government securitized debt, excluding loans, advances, local Government debt and contingent liabilities - which are relevant in a number of countries - so that we are underestimating the real burden of domestic debt. Data on domestic debt are drawn from a database developed at The World Bank by JF. Perrault, A. Presbitero and V. Tulin. A brief description of the dataset and of its sources is provided in Annex B. 18 As indicator of the cost of debt financing, the implicit interest rate is calculated as the ratio of the interest payments in the year T over the stock of debt in T An HIPC country reaches Completion Point after having maintained macroeconomic stability under a PGRF-supported program, carry out key structural and social reforms, and implement a PRSP satisfactorily for one year. Once a country reaches Completion Point it receives the full amount of debt relief which now becomes irrevocable. 13

14 concern is about four other countries at Decision or Pre-Decision Point (Burundi, Gambia, Malawi and Sierra Leone), whose interest rates are above 30 percent. Even considering the real cost of financing 20, the real interest rate on domestic debt has ranged between 5 and 10 percent from However, if we split the sample between the Completion Point and the other HIPC countries, we see that, starting from 2000, interest rates diverged dramatically, with the Completion point countries approaching zero in 2003 and rising to a small 2 percent in 2004, while Gambia, Malawi, Sierra Leone and Burundi reached, on average, 27 percent in 2002 and 26 percent in Figure 1: Central Government Securitized Domestic Debt in HIPCs Domestic debt (%GDP) Interest payments (%GDP) CG securitized domestic debt (%GDP) Interest payments on domestic debt (%GDP) Source: The World Bank The comparison of domestic and PPG external debt shows that: (1) interest payments on domestic public debt moved from a minor share to more than 60 percent of total interest expenditure, and (2) the share of the stock of debt is still small, but it is increasing quickly, especially for Completion Point countries, where domestic debt is reaching almost 20 percent of total public debt. Moreover, the maturity structure is severely biased towards short-term instruments (mainly Treasury Bills), and the banking sector in the main holder of government securities, without the existence of a broad investor base 21. Furthermore, there is the real risk of a credit constraint, since the banking system will prefer government securities instead of private sector lending. 20 Since the maturity structure is generally really short and there is a large prevalence of one and three months securities, deflating by the annual rate of inflation (measured by the GDP deflator) could provide biased estimate. However, taking into account this limitation, the real interest rate could be taken as a good approximation of the real cost of domestic financing. 21 The lack of a broader of the investor base could have negative effects on the maturity structure, since potential institutional and foreign investor could prefer long term securities. 14

15 As a consequence, internal debt service can end up absorbing more resources than external debt, undermining pro-poor investment and social expenditures. The raising cost of internal financing is therefore likely to undermine overall debt sustainability: the impact on the external financing gap of possible default on domestic debt should be properly accounted for; as the case of Argentina has shown, a default on domestic debt could seriously impact on the balance of payments if a proportion of debt is held abroad. Therefore, an assessment of the sustainability of domestic public debt is also necessary to assess external debt sustainability Domestic Debt and Sustainability Using the accounting approach (equation (1)), a policy maker should consider the real interest rate on the total public debt and not the one on external debt only (Arnone and Presbitero, 2006). By doing so, one can observe the critical role played by the inclusion of domestic debt on the solvency condition the difference between the real GDP growth (g) and the real interest rate on total public debt (r); the latter is defined as: (4) r = i π α ( πus π ) ( + π 100)( 1+ π 100) i απus ( 1 α ) π ( 1+ π 100)( 1+ π 100) 1 US US = where i is the nominal interest rate on the total public debt, π is the country s inflation rate, π US is the US inflation rate (since external debt is denominated in US dollar), and α is the fraction of external on total debt. As highlighted by data above, implicit nominal interest rate on domestic debt are much higher than the concessional rate applied on external debt, so that the resulting nominal interest rate on total public debt is larger than the one used in the standard sustainability analysis. Furthermore, we observe how an increase in inflation (both domestic and foreign) reduces the real interest rate: the relative weight given by the proportion of external debt, so that the larger is the stock of domestic debt (the smaller α), the stronger the effect of domestic inflation. This is a standard results, because the higher the inflation rate, the lower will be the real debt service payments. Central governments, therefore, especially in presence of fiscal dominance, will be tempted to raise inflation to reduce their future payments in exchange of capital loss for debt holders (inflation tax). However, higher inflation could: 1) trigger further increase in nominal interest rates to keep domestic debt attractive, thus crowding out investment, and 2) have an adverse effect on external debt sustainability, through the exchange rate, because the real depreciation will make debt service on external debt more expensive in terms of national currency. Thus, the advantages of a domestic price increase might easily be offset by its costs (a reduction of overall debt sustainability). Eventually, the simple accounting approach, if extended to take into account the overall public debt, points to the necessity for a comprehensive DSA to look at the fullyfledged government budget constraint, without ignoring the feedback effect of deficit financing on expected inflation, domestic nominal interest rates and exchange rates. Higher inflation, rising interest rates and a real depreciation are likely to have adverse impacts on the macroeconomic environment, reducing GDP growth, which is the other key variable of the solvency condition (1). 15

16 6. A Comprehensive Approach to Debt Sustainability It should now be clear that the basic accounting approach to debt sustainability, as embedded in the standard IMF-WB DSA, will not necessarily achieve its targets. The dominant financial framework represents a rigorous tool to project the evolution of public debt, but it does not internalise the feedback effects of debt on the functioning of the economy nor does it consider the developmental needs of LICs. Thus, we argue that a comprehensive framework requires a fully-fledged government budget constraint, which includes not only the external position, but also domestic debt and exchange rates. The theory of debt is, indeed, generally concerned with the government budget constraint and the evolution of debt (internal and external) indicators along with other relevant macroeconomic variables. An appropriate extended DSA should be based on a government budget constraint that links fiscal deficit, public debt, output growth, inflation and the balance of payments 22. The IMF and the WB have developed a new debt sustainability framework (Annex A), which includes only some of the proposed suggestions. In particular, large part of the literature underlines the necessity to consider the specificity of a single country s economic and institutional environment. Even if a general procedure that ends with some general threshold has the advantage of being simple and transparent, empirical works and theoretical considerations show that different countries are able to bear different level of indebtedness 23. The relevance of institutions is generally confirmed by the literature, which stresses the importance that institutional quality has for a country s capability of managing debt and fostering growth. Kraay and Nehru (2004) show that institutional quality, exogenous shocks and the debt burden are the three main structural factors that affect the probability of debt distress across countries. On the ground of the previous discussion, a comprehensive debt sustainability framework should include the following five blocks, paying attention to the necessity of evaluating feedback effects: 1. Country-specific debt thresholds, which define the need and magnitude of debt relief (or grant financing), such that the reduction in debt burden has a positive effect on economic growth, according to the findings of the extended debt overhang theory and of the relevance of the institutions and economic policies for debt management. 2. Debt service thresholds, in order to avoid the temporary liquidity constraint of servicing public debt, which crowds out public or private investment. Even if the stock of debt could not be a real disincentive, as predicted by the debt overhang effect, debt service could be an actual constraint to investment, as confirmed also by large part of the empirical literature. 3. The evaluation of the risk of default, rescheduling and arrears, which are likely to increase volatility of future inflows and uncertainty about future policies. In 22 Goldstein (2003) points out that even private sector liabilities should be considered, because they could become government debt (i.e. in case of a banking crisis, as happened in the Asian financial crisis). 23 Sun (2004) stresses the relevance of four elements which affect fundamentally the HIPCs capability of achieving long-term sustainability: (1) the quality of policies and institutions, (2) an adequate debt management, (3) diversified export base, and (4) fiscal revenues mobilization. 16