Proposal for a Contingency Debt Sustainability Framework

World Development Vol. 36, No. 2, pp. 2547–2565, 2008 Ó 2008 Elsevier Ltd. All rights reserved 0305-750X/$ - see front matter www.elsevier.com/locate/worlddev

doi:10.1016/j.worlddev.2008.02.002

Proposal for a Contingency Debt Sustainability Framework BENNO FERRARINI * World Trade Institute, Bern, Switzerland Summary. — We argue that the New Debt Sustainability Framework (NDSF) of the World Bank and IMF is centered on the Country Policy and Institutional Assessment (CPIA) to suit the aid allocation mechanism of the International Development Association (IDA), but fails to deal effectively with the economic vulnerability of low-income countries. Instead, we propose a Contingency Debt Sustainability Framework (CDSF), which identifies the sources of vulnerability and compensates for exogenous shock and trend factors. Without giving rise to significant moral hazard implications, the CDSF is thus suitable to effectively shield low-income countries from the main external causes undermining their achievement of debt sustainability. Ó 2008 Elsevier Ltd. All rights reserved. Key words — debt sustainability, aid allocation, low-income countries, HIPC, Sub-Saharan Africa

1. INTRODUCTION Apart from natural disasters, volatility of primary commodity prices has long been singled out as the main causal factor underlying low-income countries’ (LICs) economic vulnerability (Collier & Gunning, 1999; Deaton & Miller, 1996; Maizels, 1992). Commodity price volatility has increased both in magnitude in and frequency over time (Cashin & McDermott, 2001), and there is evidence that it has had a negative impact on LICs’ economic growth (Collier & Dehn, 2001; Deaton, 1999) as well as their ability to macro-manage their economies and to achieve external debt sustainability (Edwards, 2003; Maizels, 1994; Williamson, 2005). Repeated debt relief efforts by the international donor community, including the ongoing Heavily Indebted Poor Country (HIPC) Initiative, have been undermined by extreme commodity price volatility, in combination with LICs’ limited capacity to counter the economic impact of such shocks (Brooks, 1998; Gautam, 2003). In the case of many low-income countries, weak resilience to external shocks was exacerbated by policy deficiencies and economic mismanagement, mainly in terms of over-borrowing and excessive profligacy during good times, as well as the lack of counter-cyclical fiscal policies, and misconceived monetary policies and diversification efforts (IMF, 2003a).

Multilateral aid allocation operates an incentive structure of ex post rewards for countries scoring favorably in terms of the World Bank Country Policy and Institutional Assessment (CPIA), assigning most weight to governance and policy criteria. Although multilateral donors increasingly have acknowledged the impact of exogenous shocks on LICs’ debt sustainability, the CPIA-centered approach to aid mobilization broadly disregards the effects of external shocks on economic outcomes and the achievement of development targets by low-income countries (Kanbur, 2005). Similarly, debt relief initiatives have increased in scope and depth, but have failed to factor in vulnerability concerns of LICs to any appropriate degree. Under the auspices of the Multilateral Debt Relief Initiative (MDRI), most recent international efforts have entailed the full cancellation * The author is particularly grateful for comments from two anonymous referees on an earlier draft, and from Tony Addison, Thomas Cottier, Susan Kaplan, George Mavrotas, Machiko Nissanke, in relation to the Ph.D. monograph from which this article draws. Financial support from the Swiss National Science Foundation, through the National Centre of Competence in Research—Trade Regulation, is gratefully acknowledged. The usual disclaimer applies. Final revision accepted: February 22, 2008..

2547

2548

WORLD DEVELOPMENT

of the external debt burden of countries graduating from the HIPC Initiative (IDA, 2006). However, the linkage of MDRI debt relief to HIPC eligibility raises equity concerns in relation to the heavily indebted poor countries that have been excluded from the HIPC Initiative (Bird & Milne, 2003; Killick, 2004). Furthermore, even full debt cancellation falls short of representing a long-term solution in the event of a renewed build-up of unsustainable debt burdens, as the HIPC Initiative has amply demonstrated. In an attempt to avert the failure of the HIPC Initiative to address the specific vulnerabilities and external price shocks experienced by recipient countries, the World Bank and International Monetary Fund (IMF) have recently introduced the New Debt Sustainability Framework (NDSF) to guide LICs’ borrowing according to their current and prospective ability to service debt. Marking a crucial step in the multilateral approach to debt sustainability, the NDSF introduces country specificity into its assessment of LICs’ debt-carrying capacity, with important implications for the grant share of aid mobilization by the International Development Association (IDA). In this paper, we ask whether the NDSF is suitable to achieve its goal of promoting LIC debt sustainability, in light of these countries’ pronounced vulnerability to external shocks. Our assessment highlights a range of shortcomings of the new framework, which appear not only to make it unsuitable to effectively insulate debtor countries against exogenous shocks, but also actually show it to have perverse implications on a debtor country’s capacity to deal with the disruptive liquidity effects of the latter. We thus advance an alternative proposal, for a Contingency Debt Sustainability Framework (CDSF), which would modulate the amount and grant element of aid in compensation for the exogenous shocks actually experienced by recipient countries. In contrast to the existing international mechanisms dealing with exogenous shocks, most notably the IMF Exogenous Shocks Facility (ESF) and Poverty Reduction Growth Facility (PRGF), or the European Union FLEX system, the CDSF offers a more comprehensive protection against exogenous shocks, which can extend to any line of LIC’s balance of payments. 1 Moreover, its capacity to distinguish and quantify the exogenous sources of balance of payments shocks makes the proposed framework suitable to operate as an ex ante, contin-

gent debt facility. In contrast, the ex post, judgmental assessment of balance of payments needs involving IMF emergency lending that confers upon these facilities a great deal of uncertainty and arbitrariness with regard to the amount and timing of compensation, largely depending on perceived compliance of borrowers with given conditionalities. With its focus on the external balance of payments aspects of debt sustainability, the framework outlined in this paper does not address the fiscal dimension of foreign aid. Nevertheless, the proposed debt sustainability framework can be readily expanded to reflect also the fiscal implications of exogenous shocks. It could thus be envisaged that debt service adjustment by the CDSF be made more reflective of the social and human development prerogatives, for example, as advocated by the proponents of the human development approach to debt sustainability. 2 The remainder of this paper is organized as follows: Section 2 assesses the NDSF and its implications for aid allocation by the International Development Association (IDA). Section 3 outlines our alternative proposal for a contingency debt sustainability framework, and describes the credit and debt relief instruments to dampen the impact of exogenous shocks to debtor countries’ balance of payments and debt sustainability. Section 4 summarizes the main results from country studies simulating the cost and benefit implications of the CDSF. Section 5 concludes. 2. THE NEW DEBT SUSTAINABILITY FRAMEWORK In March 2005, the IDA concluded the negotiations concerning its 14th replenishment period and established the guidelines for the allocation of aid among eligible member countries during 2005–08 (IDA, 2005a). IDA14 focuses on the debt sustainability implications of IDA lending, with particular emphasis on a progressive shift toward multilateral grant financing. Concomitantly with the IDA14 endorsement, in April 2005 the executive boards of the IMF and the World Bank jointly endorsed the NDSF for low-income countries. The stated aim of the NDSF is to ‘‘guide borrowing decisions of low-income countries in a way that matches their need for funds with their current and prospective ability to service debt, tailored to their specific circumstances’’ (IDA,

PROPOSAL FOR A CONTINGENCY DEBT SUSTAINABILITY FRAMEWORK

& IMF, 2004a, p. 4—emphasis added). In order to reconcile the NDSF’s quest for country specificity in the assessment of debt sustainability with the CPIA-centered incentive structure of the IDA allocation framework, its architects needed to create the missing link between LIC debt sustainability and policy and institutional environment, along the World Bank’s CPIA. Empirical evidence in support of such a link had been readily available at an early stage of IDA14 negotiations. In 2004, the World Bank and IMF issued two empirical studies independently conducted by their research departments, which found three factors to explain most of the historical incidence of debt distress among developing and LICs: the external debt burden, the CPIA, and shocks (IDA, & IMF, 2004a; Kraay & Nehru, 2004). The policy-relevant corollary of the empirical finding was that ‘‘countries operating in a weaker institutional and policy environment are likely to experience debt distress at significantly lower debt ratios, as such countries tend to be more prone to misuse and mismanagement of funds and less capable of using their resources productively’’ (IDA, & IMF, 2004a, p. 19). On the basis of their shared CPIA centeredness, and notwithstanding the ascertained role of shocks in causing debt distress, the debt sustainability and IDA allocation frameworks could thus be integrated into what we choose to call the combined NDSF-IDA14 scheme. Chart 1 puts the scheme’s main elements in relation to each other. As both NDSF and IDA14 have been extensively described in a series of official reports by the IDA and the IMF, 3 we limit our overview to the features which are most relevant for the subsequent discussion. (a) The combined NDSF-IDA14 framework As the central junction between the combined frameworks, Chart 1 shows the NDSF to define indicative debt thresholds, which serve as a key input to the IDA14 traffic light system. Breaking with the HIPC Initiative’s practice of applying uniform thresholds across all eligible countries, the NDSF defines that maximum debt threshold LICs are supposedly able to sustain in relation to their specific CPIA ranking. Thresholds are thereby derived from the estimated coefficients of the IMF and World Bank probabilistic panel analyses across countries and time, and translated into a trade-off between the rankings of countries along their policy environment (CPIA) and external debt

2549

stock burden. Categorizing countries as poor, medium, or strong quality, according to the distribution of CPIA scores across the sample of LICs, the DSF thresholds establish the upper limits of the debt burden ratios each category of countries can sustain with a 75% likelihood of not falling into debt distress in any given year. The debt burden indicators are relatively higher for those countries scoring more favorably in terms of the CPIA. For example, at the same probability of falling into distress (25%), strong performers are deemed capable of bearing three times as much debt in relation to their total yearly exports value (300%), than countries with poor CPIA performance (100%) (IDA, & IMF, 2004a, p. 21). The traffic-light system maps debt burdens and CPIA-dependent thresholds into a ranking of countries according to categories of risk of distress (see Chart 1). It calculates the average distance of a country’s actual debt burden indicators from the indicative thresholds associated with its performance category. The distance of the actual debt burden ratios from the CPIA-specific threshold values classifies LICs as having a high, moderate, or low risk of distress, which, in turn, determines the specific country share of IDA grant financing. The grant share is set to zero in the case of low risk, to 50% for moderate risk, and to 100% for high-risk countries. That is, according to the traffic light system, high-risk countries are to receive their entire IDA entitlement in the form of grants, being deemed unable to carry any further loans without posing serious risks to the future sustainability of their external debt exposure. Apart from introducing the novelty of the traffic light system in relation to the grant share of aid, IDA14 remains in line with previous IDA replenishments with regard to the application of the Performance Based Allocation (PBA) system. Simply put, the PBA determines the country volume share of aid disbursement, rewarding with a relatively larger proportion of aid to those countries achieving higher CPIA scores, and which, more generally, are found to be in broad compliance with the typical set of policy conditions attached to multilateral lending. 4 Once volume and grant share of IDA country allocations are determined, Chart 1 shows the final step in the NDSF-IDA14 allocation process to involve the application of an upfront volume discount of 20% to the grant component. 5 It will be further argued below that this reflects an over-simplistic approach to control

2550

WORLD DEVELOPMENT

Chart 1. Schematic illustration of the building blocks of the aid allocation and debt sustainability frameworks of IDA and IMF. Note: —-, implemented in IDA14; –Æ–Æ–, Envisaged for future IDA replenishments. Ferrarini (2007:87)— Chart 3.1.

for the potential risk of perverse incentive implications from increased grant financing to poorly performing recipient countries. It should be noted that the core modules of the NDSF-IDA14 make no reference to the role of exogenous shocks in causing debt distress. 6 Instead, the NDSF addresses shocks mostly outside its juncture with the IDA allocation process, by adding to the static threshold analysis a more dynamic, forward-looking, debt sustainability analysis (DSA) component. The DSA projects LICs’ debt burden indicators and the main macroeconomic, fiscal, and external factors determining external debt dynamics over the longer term and under alternative scenarios. In relation to exogenous shocks, the most relevant scenarios involve the so-called ‘‘stress tests,’’ which simulate variations in some of the key variables affecting debt sustainability, such as debtor countries’ export revenues or GDP. ‘‘Plausible shocks’’ are thereby defined as variations occurring with an average 25% probability over a simulation period of 10 years, and their likely effects are assessed on the basis of so-called ‘‘bound tests,’’ simulating a two-year one-standard deviation from historical averages of key macroeconomic and fiscal variables.

On future implementation of DSAs for IDA countries, the IDA14 framework envisages integrating, and later replacing, the current debt distress classification based on the traffic light system and indicative thresholds with a ranking system based on the forward-looking templates of the DSA. Thereby, a country’s risk of debt distress would be assessed according to a comparative analysis of debt indicators and thresholds over time and under the baseline, alternative, and shock scenarios. The overall severity of the breach of thresholds would then feed into the traffic light system, determining grant eligibility. At the same time, the insights from dynamic sustainability assessments would inform the IDA’s decision regarding the adjustment/financing mix deemed appropriate for a specific debtor country, as an input to the overall IDA country allocation strategy. The contemplated integration of the currently static DSF modules with those representing the more dynamic forecasting exercises under alternative simulation scenarios is indicated in Chart 1 by broken lines, connecting debt sustainability analysis, debt distress classification, the traffic light system, as well as the judgment-based policy response informing country allocation strategy.

PROPOSAL FOR A CONTINGENCY DEBT SUSTAINABILITY FRAMEWORK

The DSA bound tests represent 25% likelihood of the average, representative, low-income country experiencing shocks to its key macroeconomic variables over a specific period of time. By adopting this narrow statistical definition of shocks, the NDSF-IDA14 fails to identify any large exogenous shocks occurring with low probability. Moreover, it lacks any mechanism by which the severe liquidity disruptions resulting from such shocks could effectively be countered. In partial recognition of these shortcomings, in 2005 the World Bank circulated a report evaluating alternative debt service modulation schemes for the adjustment of low-income countries’ yearly debt service payments to some broad proxy of repayment capacity, including real GDP growth, the real exchange rate, and the barter terms of trade (IDA, 2005b). Chart 1 displays the modulation schemes as a third, potential, pillar of the combined NDSF-IDA14 scheme, should the donor community agree to their implementation. 7 (b) Why the NDSF is unlikely to achieve LICs’ debt sustainability We identify at least three major shortcomings that are likely to undermine the effectiveness of the new debt sustainability framework. Our first concern relates to the robustness of the empirical analysis upon which the NDSF cements the central role of the CPIA in determining policy-dependent debt thresholds. Indeed, in a re-assessment of the World Bank and IMF analyses underlying the NDSF, we find evidence against the postulated existence of a causal relationship between policy (e.g., as measured by the CPIA) and the occurrence of debt distress (Ferrarini, 2007). Such a relationship turns out to lack robustness in the case of changes to the definition of debt distress episodes as the predicted variable in the probit regressions, and to the use of alternative indicators of countries’ policy environment and exposure to shocks among the distress predictors, other than the CPIA and deviations in GDP growth. For example, by re-estimating the same set of probit models underlying the World Bank and IMF studies on the basis of similar data, we find robust evidence that the United Nation’s Economic Vulnerability Index (EVI) explains a large proportion of distress events across LICs. Moreover, when jointly entering the probabilistic estimations as distress predictors, the EVI causes the CPIA, as well as alternative policy and institutional variables, 8 to

2551

lose statistical significance and predictive power. The robustness of these results casts doubts about the reliability of the central finding and the implications of the empirical analyses underlying the NDSF. Indeed, if vulnerability explains the occurrence of debt distress episodes across LICs, while policy does not, consistency with the NDSF’s logic would require that debt burden indicators relate to countries’ EVI and not their CPIA. However, in the absence of the crucial link provided by the CPIA between selectivity-based aid allocation and debt sustainability analysis, the entire logic of the NDSF-IDA14 collapses, and so does its potential to facilitate LICs’ debt sustainability through the country-specific adjustment of the IDA grant component. Secondly, the NDSF-IDA14 appears to suffer from a pronounced bias against the liquidity dimension of debt sustainability. For the framework increases the grant element to poorly performing debtor countries, but decreases the liquidity injection from new disbursements by virtue of the CPIA-centered PBA regime itself and the 20% volume discount on grant financing introduced by IDA14. To the extent that low-income countries’ risk of debt distress depends on their vulnerability to exogenous shocks, rather than CPIA performance, the NDSF introduces an allocation bias against those countries with the greatest need of liquidity to counter the immediate effects from such shocks. Furthermore, IDA loan disbursements entail a 10-year grace period before repayments start falling due, delaying the benefits to LICs’ debt sustainability from debt service reductions through higher grant financing and ruling out any significant short-term liquidity impact. Thirdly, the potential benefits of debt service modulation schemes on LICs’ sustainability are likely to be undermined by the strict limitations the World Bank has envisaged in relation to their function as a third pillar of the DSF. In order not to entail any pre-commitments for additional funding beyond the amount determined by the core pillars of the NDSFIDA14, these instruments are not to change the overall amount and grant share of IDA financing, and would be limited to adjusting only marginally the yearly amount of debt service according to some narrowly defined repayment capacity index. However, by ruling out additional resource transfers to LICs in any given period, these instruments would provide insufficient coverage to counter the sharp

2552

WORLD DEVELOPMENT

liquidity shortages typically caused by exogenous shocks. Furthermore, the simple indexing instruments envisaged by the World Bank are bound to either capture exclusively shocks from a relatively limited and well-defined source (as is the case with the terms of trade or specific export price indices) and thereby ignore all other important sources of shocks to an economy, or to fail to identify the source of shocks when indexing to some broader proxy of repayment capacity (e.g., to countries’ nominal GDP growth). As a result, the indexing mechanisms envisaged by the NDSF-IDA14 are either unreflective of the broad spectrum of shocks to which the low-income countries are typically subjected, or unsuitable to address the incentive distortions arising from insuring countries against an unidentified source of liquidity shortage, ensuing from a mix of policy-driven and exogenous factors. In sum, we argue that the major shortcoming of the NDSF-IDA14 approach is reflected in its reliance on a presupposed inverse relationship between CPIA performance and the occurrence of debt distress across LICs and time, which is insufficiently supported by empirical evidence. Consequently, the integration of the NDSF into the pre-existing ex post reward structure of IDA lending fails to provide a framework to effectively guide sustainable debt allocation across LICs, and is merely instrumental in accommodating the shift from loan to grant financing of multilateral aid. Ultimately, even after the introduction of debt service modulation instruments, the new debt sustainability framework would resemble an inconsistent patchwork of measures, which would be largely inappropriate to address the external debt effects of LICs’ pronounced vulnerability to exogenous shocks. 3. PROPOSAL FOR A CONTINGENCY DEBT SUSTAINABILITY FRAMEWORK Our proposal for a Contingency Debt Sustainability Framework (CDSF) aims at providing effective protection against the disruptive effects of exogenous shocks on LICs’ balance of payments (BOP) and debt sustainability. In contrast to the CPIA-centered NDSF, our approach for a contingency scheme is grounded on the evidence that external shocks have played a central role in undermining LICs’ debt sustainability, and that the efficacy of aid is maximized when it is targeted toward alleviat-

ing the disruptive effects of such events (Collier & Dehn, 2001; Guillaumont & Chauvet, 2001). The CDSF builds on an established body of sovereign debt literature, which demonstrates that the overall efficiency of loan contracts is broadly enhanced when the ensuing liabilities are linked to the repayment capacity of the recipient country as determined exogenously, that is, by the state of nature (Krugman, 1988). Seminal contributions to this literature have also shown that any distinction between a borrower’s situation of temporary illiquidity and one of insolvency is largely a theoretical artifact, with little bearing on policy decisions. For not only is the infinite time horizon associated with a sovereign entity unsuitable for establishing its net worth in the presence of uncertain future outcomes, but so is the sustainability of its foreign debt—both in terms of flows and stock—essentially determined by the lending decisions according to creditors’ own perceptions about a borrower’s creditworthiness, considered along other lenders’ response functions (Grossman & Van Huyck, 1988; Huberman and Kahn, 1988; Froot et al., 1989). Particularly in the context of official aid-dependent LICs, also largely spurious is the distinction between temporary balance of payments crises and longer term debt sustainability. There, adjustment policies are enshrined as pre-commitments in the form of aid conditionality, and liquidity concerns arise mainly out of lenders’ refusal to adjust disbursement schedules according to recipient countries’ needs instead of being guided by their own perception about a recipient’s past and future compliance with the policy strings attached to extant and new lending. Ultimately, the concept of debt sustainability in the context of LICs is intimately linked to the analytical framework and the reciprocal perceptions through which a recipient country’s sustainability is perceived among the donors, and the main issue of concern is one of illiquidity, not insolvency. 9 In line with these insights, the proposal for a CDSF outlined here seeks to depart from both the narrow and the partial focus of the World Bank indexing mechanisms 10 and the judgmental, ex post, nature of the IMF emergency lending facilities discussed above. Instead, the CDSF endeavors to define a comprehensive accounting mechanism to distinguish policy-induced from state-contingent factors determining repayment capacity. By providing the accounting basis on which the state of nature

PROPOSAL FOR A CONTINGENCY DEBT SUSTAINABILITY FRAMEWORK

can be identified and measured with regard to its effects on debtor countries’ BOP and external debt sustainability, the CDSF is suitable to operate as a state-contingent contract. The incentive structure of the CDSF is optimized by locking in ex ante lenders’ optimal response functions to the states of nature contemplated by the contract underlying the scheme, and by providing an enabling environment for contract renegotiation in relation to all the factors falling outside the realm of ex ante regulation. Therefore, we argue that the contingency scheme we propose is better able to compensate LICs for a broad range of shocks, and yet avoid the perverse incentive implications associated with modulation schemes that link debt service to broad indicators of repayment capacity, but do not explicitly account for the effect of the state of nature on the latter. Our proposal for a CDSF entails an array of closely inter-related instruments, which are outlined in Chart 2. Before discussing the instruments of the CDSF in detail, it should be noted that Chart 2 shows the framework to distinguish between two time periods, in relation to its main implications. Period t should be thought of as referring to a time span long enough to allow for the collection and analysis of the relevant information about the factors affecting a debtor country’s BOP during its course. It may correspond to the yearly auditing of the country’s BOP statistics, or to the periodic conclusion of broader economic assessments such as those that most LICs un-

2553

dergo every six months by virtue of their involvement in three-year arrangements with the IMF Poverty Reduction Growth Facility (PRGF). Whatever its duration, period t constitutes the time frame within which the CDSF operates as an ex ante contingency mechanism, that is, as a contractual arrangement for compensation in the event that pre-defined criteria are met. In contrast, period t + 1 refers to the CDSF implications relating to subsequent periods, which are not based on an ex ante arrangement, and thus require either negotiations between the donors and the recipient, or are left to discretion of the donor community alone. The CDSF presented here envisages a typical sovereign borrower–sovereign lender relationship. If this relationship is assumed to involve only one multilateral lender, the main implications of the CDSF are suitable for direct comparison with the IDA aid allocation and debt sustainability frameworks outlined above. However, the sovereign lender party to the contract would more appropriately be viewed as representing the broader donor community, so as to confer on the scheme the potential to effectively address LICs’ broader debt sustainability concerns in relation to their entire official external debt. Similarly to the HIPC Initiative, the donor community can thus be thought of as involving mainly the multilateral and bilateral entities, to whom the bulk of LICs’ debt stock is owed. Here, we conveniently assume that the donor community manages to overcome all coordination problems,

Chart 2. The contingency debt sustainability framework. Ferrarini (2007:188)—Chart 5.1.

2554

WORLD DEVELOPMENT

and that a representative body (e.g., the IDA, Paris Club, or an entity constituted ad hoc) be invested with the power to act on behalf of all its members. Finally, for ease of exposition the focus of the CDSF is limited to the BOP aspects of external debt sustainability. Although the fiscal aspects relating to foreign aid dependency are thus not directly addressed by the scheme outlined here, a more encompassing CDSF should be thought of as comprising two interrelated assessment processes addressing both a borrower’s external and fiscal sustainability aspects in relation to its overall indebtedness. With these qualifications in mind, the key components of our proposal for a CDSF are now discussed in turn. (a) Endogenous and exogenous balance of payments determinants The central analytical building block of the CDSF relies on the separation of endogenous from exogenous determinants of illiquidity. Exogenous factors are defined as being beyond the influence of the debtor country, such as world demand and prices facing a small country’s exports, while endogenous factors are subject to the country’s control or partial influence. The lower left-side of Chart 2 shows the CDSF to further disaggregate exogenous factors into trend and shock components. While both components are externally determined variables, trend effects are assumed to be internalized in the country’s expectations regarding future BOP realizations, while shocks are not. That is, the country is assumed to formulate its economic policies according to trend expectations, thereby reflecting historical trends of exogenous BOP effects in its policy decisions. 11 Exogenous shocks, in contrast, are defined as random realizations around trend and are, as such, unforeseeable. As will be argued below, in relation to the CDSF performance assessment, it is crucial for such a distinction to be made in the context of a contingency scheme, for it not to distort a LIC’s incentives toward emancipation from a condition of vulnerability to exogenous shocks. The upper left-side box of Chart 2 represents all non-exogenous determinants of the BOP. These include fully endogenous factors, which are entirely under government control, as well as the variations in BOP flows resulting from the complex interrelations between external shocks and endogenous policy reactions to the latter. We label these mixed, or indeterminate,

effects. In order to circumvent the indeterminacy problem in relation to the strongly entangled forces affecting the BOP, the CDSF relies on an accounting methodology that allows for a sufficiently accurate extrapolation of all the BOP determinants that can be clearly qualified as exogenous price shock or trend factors on the basis of consolidated BOP data observable by the lenders at the end of period t. By contrast, all the other BOP effects are dealt with as a residual. To derive the CDSF accounting method, it should be noted that in relation to any period t a country’s external demand (ex post) for new concessional loans is reflected in its balance of payments, which in simplifying notation can be expressed in the following terms: LSt P LDt  it Dt1 þ pt Dt1 þ IMP t  EXP t  GRt  WRt  FDI t þ DRt þ Z t

ð1Þ

Identity (1) shows the demand for loans (LDt , with inverted sign) to derive from the sum of the total trade balance (IMPt  EXPt ), the payments of interest (itDt1) and principal (pt Dt1) on existing debt, and net changes in the country’s international reserves (DRt); minus the sum of non-debt creating financial inflows, that is, official grants (GRt), net foreign direct investment inflows (FDIt), and workers’ remittances (WRt); plus a residual factor (Zt), aggregating all other BOP flows including errors and omissions. As the net sum of all these flows, LDt represents a low-income country’s financial gap before concessional lending. The supply of concessional loans is a function of the development lenders’ judgment about a country’s performance and needs, relative to those of the other recipient countries. However, aid supply also reflects other important factors, such as political considerations and coordination failures, which make it largely unsuitable for deterministic modeling. Assuming that demand for new loans is insatiable, at least to the extent that it always outstrips supply, LSt 6 LDt , identity (1) represents the supply-constrained demand identity for observed loan disbursements in any period t. To circumvent the indeterminacy problem relating to the identification and measurement of exogenous shocks, we follow a methodology introduced by Balassa (1982), and further developed by Solis and Zedillo (1985), to derive a counter-factual balance of payments that would have prevailed in the absence of certain relevant exogenous events. 12 We first proceed

PROPOSAL FOR A CONTINGENCY DEBT SUSTAINABILITY FRAMEWORK

by expressing net official transfers (NTRt) as the difference between disbursements of new loans ðLSt Þ and the service of existing debt (it + pt)Dt1 in period t: NTRt ¼ LSt  ðit þ pt ÞDt1 : ð2Þ Separating price ðpmt ; pxt Þ and volume (Mt, Xt) components of total exports in goods and services, the balance of payments identity of period t can be stated as: NTRt  pmt M t  pxt X t  GRt  WRt  FDI t þ DRt þ Z t ;

ð3Þ

Or, in terms of trend values, signed by over bars: NTRt  pmt M t  pxt X t  GRt  WRt  FDI t þ DRt þ Z t

ð4Þ

Trend values serve as benchmarks against which actual realizations of BOP items are measured, and are calculated as moving average in relation to period t. For example, the export price trend during the period t  k is computed as k 1X  pxt ¼ px : ð5Þ k i¼1 tk1þi The deviations of BOP items from trend levels are computed by subtracting identity Eqn. (4) from Eqn. (3). For example, exports price trend deviations are expressed as ðpxt   pxt Þ, and real exports deviations as ðX t  X t Þ. In the case of

prices, deviations of actual values from trend can be clearly classified as exogenous factors, to the extent that LICs are usually not in a position to influence world prices. In contrast, for the case of trade volumes, the identification of the exogenous component requires the introduction of the so-called ‘‘hypothetical’’ values, 13 to account for the scenario of unchanged external circumstances in relation to a country’s real trade balance. We define as external factors those affecting a debtor country’s real trade balance through variations in world demand, explaining real changes in exports, and via fluctuations in GDP growth, affecting import volumes. More specifically, following Solis and Zedillo (1985), we compute trend and hypothetical exports volume as X jt ¼ X jo ð1 þ gjX Þt t ¼ 1; 2; . . . ; T bj ¼Xj X t o

trend export volume ð6Þ

T Y ð1 þ gjX Þ hypothetical export volume 1

t ¼ 1; 2; . . . ; T

ð7Þ

where index j denotes a specific commodity or category of commodities, 14 gjX is the growth rate in period t of world demand for the country’s export item j, and gjX is the trend of world demand. Assuming that a country’s share in the world market of item j remains unchanged

Export Volume

X

Xˆ

X

t+1

gX

t

)

t+3

t+2

(X

(X

2555

t

t

)

− Xˆ t > 0

Increase in export volumes due to domestic factors (mixed endogenous-exogenous)

− Xˆ t > 0

Negative impact from an exogenous change in world demand

Figure 1. Actual, trend, and hypothetical export volumes. Source: Ferrarini (2007:202)—Figure 5.1.

2556

WORLD DEVELOPMENT

between subsequent periods of assessment, the difference between hypothetical and trend export values represents a measure of the real effect of exogenous changes in world demand on a debtor country’s balance of payments. In conjunction with observed changes in export prices, it gauges the degree to which exogenous shocks to export volume and prices affect a country’s actual export receipts over the chosen period of observation. Figure 1 further clarifies the conceptual distinction between the actual, trend, and hypothetical export volume in relation to the CDSF. Trend volume of exports is shown to be computed on the basis of average growth in world demand over the three years prior to t. The hypothetical volume reflects the country’s potential export quantity, determined by the world demand for its exports, as observed in each period. The difference between trend and hypothetical exports measures the real impact of trend deviations in the growth rate of world demand. For instance, in period t + 1 the growth of world demand is shown to fall short of average growth, and the vertical distance between trend and hypothetical volume measures the shortfall of export quantity due to this negative exogenous shock. In contrast, the difference between actual and hypothetical volume in period t + 1 would only be explainable by internal forces affecting real exports, such as a policy-induced increase in the productivity of the export sector, usually in complex interaction with unidentifiable exogenous events. To the extent that the latter cannot be clearly distinguished, the combined effect is to be allocated within the mixed-endogenous category of the CDSF. Only if it were possible to clearly identify and measure further factors that systematically determine a country’s export volume, besides changes in world demand, should these be accounted for in the right-hand side of Eqns. (6) and (7), and thus automatically qualify as being exogenous. Turning to the imports side of the trade balance, trend, and hypothetical import series are calculated as follows (Solis & Zedillo, 1985): M jt ¼ M jo ð1 þ ejM gGDP Þt b j ¼ Mj M t o

trend imports

ð8Þ

T Y ð1 þ ejM gGDP Þ hypothetical imports 1

t ¼ 1; 2; . . . ; T ejM

ð9Þ

is the trend income elasticity of imwhere port item j, gGDP is the debtor country’s actual

GDP growth rate between periods t  1 and t, and gGDP is the trend rate of the GDP growth. Similarly to exports, any measurable factor that should be observed to be systematically affecting a specific country’s import volume overtime, for example, official grant disbursements in support of imports, ought to be accounted for by inclusion in the right-hand side of Eqns. (8) and (9). 15 Assuming that import elasticity remains constant over the period of analysis, the difference between trend and hypothetical imports measures the volume effects of the deviation of GDP growth from its trend value. It should be noted that the difference in actual, trend, and hypothetical imports is attributable to changes in import elasticity and/or GDP growth, which are both factors that are, at least in principle, susceptible to influence by the country authorities. Therefore, the corresponding BOP effects cannot be qualified as exogenous, as will be further clarified below, in relation to identity (10). The CDSF accounting method can be expanded to distinguish between trend and hypothetical values in relation to any of the other BOP items on the right-hand side of identity Eqn. (3). For example, changes to the net flow of FDI against hypothetical FDI could be defined as the flow consistent with the country’s share of FDI inflows in relation to its own specific FDI determinants and those of neighboring countries at a certain point in time. Discrepancies between donor commitments and actual disbursements of official grants could enter the accounting identity in the form of hypothetical grants, defined as the proportion of the overall shortfall that is not attributable to debtors’ policy actions or compliance with policy conditionality more generally. Ultimately, the actual setup of the CDSF accounting mechanism will depend on the specific characteristics of the debtor country involved, and the availability of a sufficiently comprehensive and reliable set of information relating to the BOP determinants. Limiting the distinction of hypothetical values to the trade balance alone, the full specification of BOP effects is obtained by subtraction of identity Eqn. (4) from Eqn. (3), substitution for trend and hypothetical imports and exports, and disaggregation of value terms to reflect price and trend components. We thereby obtain the following set of expressions, constituting the central accounting framework underlying the CDSF NTRt  NTRt 

ð1016 Þ

PROPOSAL FOR A CONTINGENCY DEBT SUSTAINABILITY FRAMEWORK

(I) Exogenous shock effects on the trade balance b tÞ  IðX t  X b tÞ þðpXt  IÞðX t  X þðpm pm t  t ÞM t x x ðpt  pt ÞX t b tÞ ðpxt  pxt ÞðX t  X

Volume change (a) in exports demand Value effect of (a) (b) Terms of trade effect

Price trend deviation of (g) b ðpm pm t  t Þð M t  M t Þ Price trend deviation of (i) b ðpm pm t  t ÞðM t  M t Þ Price trend deviation of (k)

(c) (d) (e) (f)

(II) Trade balance mixed effects: volume and price trend effects b tÞ IðX t  X b tÞ ð pxt  IÞðX t  X b t  M tÞ Ið M ð pm t  IÞ ^ t  M tÞ ðM b tÞ IðM t  M b ð pm t  IÞðM t  M t Þ

Internal change in export volume Price trend effect of (g) Internal change in import volume Price trend effect of (i) Import volume effect of GDP Price trend effect of (k)

(g) (h) (i) (j) (k) (l)

(III) Financial flows mixed effects: IðGRt  GRt Þ IðWRt  WRt Þ

Deviation in grants Deviation in workers’ remittances IðFDI t  FDI t Þ Deviation in FDI inflows IðDRt  DRt Þ Deviation in reserves accumulation þIðZ t  Z t Þ Other net debt-creating flows

(m) (n) (o) (p) (q)

Identity (10) distinguishes three categories of BOP effects. Whether individual items enter the identity with positive or negative sign depends on their effect on the country’s demand for additional net transfers. Items with positive sign are debt inducing, since they explain a proportion of the positive deviation in net official transfers to the country. Conversely, items taking negative sign are accounted for as having

2557

had a debt-deducing effect during the period of observation, reducing the BOP gap to be filled by net official transfers. The first category of items contains all the purely exogenous debt-inducing trade balance effects. With the exception of item (a), measuring the effect of volume changes in world demand, these are all price effects. Item (b) captures the price impact of item (a). Together with item (a), it adds up to a value measure of differences between actual and hypothetical export volume in the trade balance. Item (c) measures terms of trade shocks as price variations against imports and exports volume trends, thus capturing the purely exogenous shock element of the terms of trade factor affecting a country’s BOP in period t. Items (d), (e), and (f) represent the price trend deviation of exports volume, imports volume, and GDP growth, respectively. According to our definition, only these price variation components represent true shocks to a country’s BOP, while their trend effects are mixed and thus belong to category (II). The overall sum of items (a) to (f) constitutes the exogenously determined proportion of the overall trend deviation in the country’s BOP during period t, net of official transfers. Depending on whether it takes a positive or a negative sign, the overall shock component in t is classified as unfavorable or favorable to the country, respectively. As will be discussed below, the contingent credit line of the CDSF triggers automatic credit adjustments at the end of period t, to compensate for the overall BOP shock component. All the volume and trend effects of the trade balance, which cannot be classified as exogenous shocks, are included in the second category of identity (10). Item (g) measures the difference between actual and hypothetical exports. Similarly to the other items in category (II), it enters identity (10) with a negative sign because a positive difference between actual and hypothetical export volume reduces the country’s need for net official transfers in relation to the overall BOP deviations. Item (h) represents the price trend effect of the deviation in export volume. In contrast to item (d), neither (g) nor (h) qualify as purely exogenous effects, since change in export volume is deemed to be at least in part determined by factors under the control of the debtor country. All remaining items of category (II) are import effects. Items (i) and (j) compute the volb  M t Þ, which ume and price trend effects of ð M t is the deviation of actual imports volume from

2558

WORLD DEVELOPMENT

the amount of imports explained by the country’s rate of growth of real GDP and the income elasticity of imports. Items (k) and (l)compute the volume and price trend effects b t Þ, thus quantifying the imports efof ðM t  M fect of deviations in the GDP growth rate from its own trend. The items of category (II) typically pose difficulties in trying to fully distinguish their exogenous and endogenous origins. Nevertheless, as will be further discussed below, the CDSF entails that, conditional upon a country’s policy performance, the price trend components of category (II)—that is, items (h), (j), and (l)— qualify for an equivalent grant conversion of debt disbursed in period t. In contrast, the volume effects (g), (i), and (k) do not qualify for grant conversion, but serve as indicators of the country’s need for official assistance in relation to the determination of future aid commitments. Finally, the third category of effects includes the trend deviations of non-debt financial flows. They take a negative sign, because positive trend deviations have a debt-deducing effect on the BOP. Item (m) measures variation in official grant disbursements; item (n) in workers’ remittances, or private current account transfers more generally; 17 item (o) in net FDI inflows to the country; item (p) in changes to the country’s international reserve holdings. It should be noted that the reason for including disbursements of official grants (item m) in category (III) is that, from the perspective of an ex post assessment, the CDSF’s emphasis is mainly placed on identifying the debt-inducing effect of specific BOP items. Since a positive trend deviation of official grant transfers has a lowering effect on the period’s balance of payment gap, it explains part of the net deviation in official net transfers, similar to the other effects. However, in relation to the CDSF’s determination of aid commitments along expected BOP trend developments during period t + 1, the grants item would be more correctly shifted to the left-hand side of identities (3), (4), and (10), reflecting total net official development assistance planned for period t + 1, jointly with net loan transfers. Based upon the distinction between exogenous, trend, and indeterminate BOP effects according to the accounting method outlined in this section, the CDSF is able to define and operate its main instruments and assessment functions. The contingent credit line, policy performance assessment, debt relief mecha-

nism, and debt sustainability assessment are now discussed in turn. (b) The contingent credit line The contingent credit line constitutes the central CDSF instrument to adjust a LIC’s BOP cash flow to the occurrence of exogenous shocks. Upon identification of the BOP effects of shocks, the contingency mechanism involves the automatic disbursement—or amortization, in the case of favorable shocks—of interest-free top-up funds in proportion to the net sign and magnitude of these effects. In order to effectively fill the liquidity gap ensuing from negative shocks, such disbursements are to occur periodically, at the end of period t, or with higher frequency, depending on the feasibility of conducting an immediate impact assessment. It should be noted that since the CDSF defines shocks as random realizations around trend, the balance of the contingent credit line should revert to zero over the long term. Therefore, there is no particular justification for the contingency mechanism to disburse grants, instead of credits, except for the case of substantial imbalances in relation to the occurrence of large real shocks such as natural disasters. 18 (c) Performance assessment against contractual obligations The performance assessment constitutes the central element of the CDSF in relation to the treatment of non-shock factors determining a beneficiary country’s balance of payments during period t. According to the above definition, both trend and mixed-endogenous components are to some degree under the control of the debtor country, or are at least in principle amenable to the effect of deliberate policy choices. Therefore, a compensatory mechanism that extended to these factors the same treatment as accorded to fully exogenous shocks would introduce the potential for incentive distortions affecting a recipient’s effort aimed at reinforcing its BOP position over time. 19 To circumvent moral hazard problems, the CDSF assesses domestic policy performance against pre-defined country-specific benchmarks. Performance expectations are set in relation to each period t, and a detailed country policy agenda in lieu of an underlying contract between the country authorities and the donor community is defined. 20At the end of each period t, a country is assessed in terms of the enactment

PROPOSAL FOR A CONTINGENCY DEBT SUSTAINABILITY FRAMEWORK

of planned policies during the period under analysis, but not on the basis of outcomes. Therefore, to the extent that domestic policy implementation relates to observable actions, such as decrees and laws, the CDSF avoids both a performance assessment being distorted by exogenous factors affecting outcomes, and borrowers being unduly held accountable for either exogenous factors or the actual development effectiveness of agreed-upon policies. Apart from establishing the pre-condition for the scheme to be able to operate on an ex ante basis, in accordance with the terms set out in the underlying contract, the CDSF performance assessment provides the donor community with an effective instrument for the enforcement of countries’ policy commitments. In practice, the performance assessment could be conducted in a similar way to the periodic reviews of LICs’ IMF PRGF arrangements, which already include close monitoring of governments’ compliance with IMF policy conditionality. 21 However, in contrast to the PRGF review, which makes the periodic disbursement of IMF credit tranches conditional upon recipient countries meeting quantitative performance criteria and benchmarks relating to policy results, we propose the CDSF to assert countries’ qualification for continued support on the basis of policy enactment alone. The performance assessment has different implications with regard to the various components of the CDSF. In relation to LICs’ overall qualification for the CDSF, including the contingent credit line, a severe breach of policy commitments, or incompatible actions such as fraudulent reporting, would necessarily have to lead to suspension or indefinite exclusion from the scheme’s benefits. The exact procedure and criteria underlying such a decision would have to be anchored in the terms of the contractual agreement, and would optimally involve a panel of representatives from both the lending and the borrowing communities. In relation to both the category of mixedendogenous and trend BOP factors, the performance assessment ascertains a LIC’s fulfillment of policy conditions during period t, as one input informing the lenders’ decision concerning aid commitment for period t + 1. Similarly to the current IDA allocation process, such a decision would ultimately have to rely on a broader assessment of any LIC’s specific needs for official development assistance, also in reflection of the fiscal accounts. However, in contrast to the CPIA-based allocation process of NDSF-

2559

IDA14, a LIC’s qualification for continuing aid disbursements or new commitments would depend on actual policy compliance, not ex post reward on the basis of CPIA performance, and with balance of payments volatility as a central criterion for aid allocation across eligible countries. (d) The debt relief mechanism The CDSF categorizes BOP trend factors as originating from causes exogenous to the country, but recognizes that their magnitude is susceptible to influence by deliberate policy decisions. However, by classifying policy actions according to their contractual legitimacy, the performance assessment validates the component of trend factors that is (potentially) under government control. Hence, to the extent that a LIC is found to be in compliance with its policy obligations, the CDSF is suitable to provide distinct treatment for the category of trend effects, without introducing incentive distortions. More specifically, we envisage the CDSF to convert official credit flows automatically into grants, thus effectively relieving existing debt in proportion to the unfavorable trend factors faced by the country during period t. The rationale for such a relief operation is that LICs frequently face prolonged negative trend factors as a reflection of underlying structural characteristics, which tend to undermine LICs’ debt sustainability, but which cannot be overcome in the short- and medium-term by domestic policy alone. It should be noted that the provision of ex ante debt relief according to a country’s observed degree of exposure to exogenous trend factors puts the CDSF in strong contrast to the NDSF-IDA14, which predetermines a country’s grant share according to the perceived risk of debt distress and thus lacks the necessary flexibility to adjust the grant share of ODA to the actual circumstances affecting a debtor’s BOP. (e) Debt sustainability assessment The compensatory functions of its contingency instruments make the CDSF unsuitable for debt sustainability assessments along the NDSF’s static benchmark approach across LICs. Moreover, contingent compensation limits the concern of debt sustainability assessment to the projection of the residual mixed and endogenous factors, in relation to their effects on external debt stock and flow sustainability. 22

2560

WORLD DEVELOPMENT

100 200 300 400 500 600

Therefore, CDSF debt sustainability assessments would have to be formulated along alternative scenario forecasts and stress tests, involving the mixed and endogenous items other than trend effects, as identified by identity (10). In combination with a broader needs assessment in relation to LICs’ demand for aid beyond the grounds of vulnerability, as well as the outcome of the performance assessment in relation to period t, CDSF debt sustainability analysis would serve as an input to the lender community’s decision with regard to the pre-determined volume and grant mix of aid commitments during t + 1. Falling outside the realm of the CDSF operating as an ex ante contract-based mechanism, the predetermination of aid mobilization would thus remain a largely discretionary process, ensuring the scheme’s compatibility with the lender community’s prerogative to

Disbursement, Actual

1989

50 100 150 200 250

1988

1990

1991

1992

keep broad control over their commitment of funds. Unavoidably, debtor countries’ debt sustainability would ultimately have to depend on lenders’ lending decisions. However, within the limits of the latter, LIC’s debt sustainability prospects would be greatly enhanced by the CDSF’s role of re-adjusting the relative amount and terms of aid flows across participating LICs’ according to their relative exposure to exogenous shock and trend factors. 4. AN APPLICATION OF THE CDSF We assessed the implications of the CDSF for a sample of low-income countries in a related study (Ferrarini, 2007). Adopting a historical, backward-looking, simulation approach we simulate the CDSF accounting methodology

Disbursement, Adjusted

1993

1994

1996

1997

1998

1999

2000

2001

2002

1998

1999

2000

2001

2002

1998

1999

2000

2001

2002

Amortization, Adjusted

0

Amortization, Actual

1995

1989

100 200 300 400 500

1988

1990

1991

1992

1993

1995

1996

1997

Net Flows, Adjusted

0

Net Flows, Actual

1994

1988

1989

1990

1991

1992

1993

1994

1995

1996

1997

Graph 1. DIS, AMT, and NFL: actual vs. adjusted, US$ millions, 1988–2002. Source: Ferrarini (2007:267)—Graph 6.25.

PROPOSAL FOR A CONTINGENCY DEBT SUSTAINABILITY FRAMEWORK

and compensatory instruments for the cases of Uganda, Ghana, Mozambique, and Zambia over the period 1988–2002. Accounting for the observed exogenous shock and trend factors these countries have actually experienced during the period of assessment, it is possible to directly compare the actual development of debt stock and flow indicators with the simulated effects after introducing the CDSF regime. Moreover, this comparative assessment allows for a basic insight into the likely costs incurred by the CDSF, compared to those of the debt relief instruments that were actually implemented during the period of assessment, mainly in the form of Paris Club relief operations and the HIPC Initiative. Due to space constraints, it is not possible to report here the simulation results in any detail. We thus limit the focus to a summary of the most salient results from one representative case study, applying the CDSF to Uganda. 23 The upper panel of Graph 1 shows the contingent credit line to play a crucial role in balancing the country’s BOP liquidity, through automatic compensation in the face of exogenous shocks. For example, the CDSF regime is shown to reduce aid disbursement during the coffee price boom around 1995, and to significantly increase aid in its aftermath, when Uganda’s terms of trade were worsening. The middle panel of Graph 1 illustrates the liquidity effects of the debt relief facility. Here, CDSF grant conversion is accounted for as equivalent amortization of existing debt, in order to allow for direct comparison with the

2561

country’s actual amortization pattern. The CDSF is shown to significantly accelerate Uganda’s debt amortization (grant conversion) during years of unfavorable trend exposure. In terms of net flows, together with the modulation of aid disbursements, the lower panel of the graph shows the CDSF to be highly effective in countering the disruptive shock and trend factors that Uganda has been exposed to over the period of assessment. Turning to the debt stock implications of the CDSF, Graph 2 shows a comparison of the observed outcome of the Paris Club and HIPC initiatives over the period 1988–2002 with those implied by the simulated CDSF. 24 Particularly since 1998, the CDSF is shown to induce a falling trend in Uganda’s external debt stock, against rising stocks associated with the HIPC Initiative. Contingent debt relief triggered by unfavorable trend factors more than offset the substantial increase of net loan transfers to Uganda by the CDSF in compensation of negative shocks during the second half of the 1990s. In contrast, the lack of any contingent support system by the HIPC Initiative undermined its massive relief operations. In the face of unfavorable shock and trend factors, Uganda had to resort to new borrowing to finance its BOP gap, causing a renewed buildup of its external debt stock. The benefits of the CDSF in terms of its liquidity and stock implications clearly emerge from the Uganda country study. Assessing the cost implications of the CDSF over the entire simulation period and assuming full relief of

Net debt stock Net debt stock. after HIPC/PC relief

0

1.000

2.000

3.000

4.000

5.000

Net debt stock. after CDSF relief

1988

1989

1990

1991

1992

1993

1994

1995

1996

1997

1998

1999

2000

2001

2002

Graph 2. CDSF external debt stock effects, Uganda, 1988–2002, US$ millions. Source: Ferrarini (2007:270)—Graph 6.26.

2562

WORLD DEVELOPMENT

the final net balance on the contingent credit line, we find that with a total cost of US $1,216 million, the CDSF would have only slightly exceeded the combined costs of the Paris Club and HIPC relief initiatives, which amounted to US $1,161 million (IDA, & IMF, 2003). In sum, we find that with similar cost implications to those the international donor community has actually sustained in the case of Uganda, the CDSF would have achieved its primary goal of providing substantial compensatory cash flows in response to the multitude of shocks experienced by the country. Moreover, it would also have led to a reversal in the country’s accumulation of external debt at a time when all the existing relief mechanisms were failing to offer appropriate support to the country. 5. CONCLUDING REMARKS Further research will be necessary in order to make the CDSF a fully operational framework, and a feasible alternative to the NDSFIDA14. Among the main future developments of the proposed framework, we envisage a further refinement of the CDSF accounting framework, particularly with regard to its specification of hypothetical values. For example, in relation to trade volumes, simple indexing to world demand and GDP fails to capture accurately all the main factors determining hypothetical levels of real exports and imports flow, and a viable method needs to be defined

by which the NDSF could include countryspecific factors influencing these flows. Furthermore, the accounting method is to be fully integrated with the fiscal balance, adding to the BOP dimension of LICs’ debt sustainability considered in this paper. Finally, the question of equity and fairness of any country-specific compensatory scheme, across LICs and the broader aid-recipient community, needs to be further addressed in the context of state-contingent financing. In particular, questions arise in relation to the definition of countries’ qualification and graduation procedures from the contingency scheme, and the adverse selection implications of alternative rules. Similar concerns relate to the coordination problem among the donor community, particularly with regard to incompatibility issues between the specific conditionalities and assessment methods applied by different donors. Upon future solution of these important caveats, we take the view that a contingency scheme along the lines of the CDSF could represent a useful basis for a renewed international effort to overcome the shortcomings of the current aid allocation and debt sustainability paradigm. Although the donor community might find it problematic to agree on the CDSF in its proposed form, the main elements of the scheme—in particular, the contingent credit line—would also be suitable for implementation within the current DSF, as an alternative to the more narrow debt service modulation schemes envisaged by the World Bank.

NOTES

1. For a description of international instruments to address terms of trade shocks, see Varangis, Varma, dePlaa, and Nehru (2004). For an overview of IMF assistance for countries facing exogenous shocks, see IMF (2003b). The ESF facility is outlined in IMF (2006).

tarities and compatibility issues between the two approaches, which would mostly relate to the fiscal dimension of an enhanced CDSF. For a human development assessment of the new debt sustainability framework, see Caliari (2006). 3. IDA & IDA, and IMF (2004a, 2004b, 2005).

2. Similar to the so-called human development approach to debt sustainability, the CDSF modulates debt service payments to countries actual ability to pay. However, while the former approach emphasizes the precedence of human development imperatives over debt payments mainly in terms of fiscal outlays, the CDSF is mainly concerned with the liquidity implications and macroeconomic effects of balance of payments shocks. In this paper, we do not explore possible complemen-

4. More precisely, the upper left-hand panel of Chart 1 shows the country’s performance-based rating (PBR) to be derived from a LIC’s combined score in the Country Policy and Institutional Assessment (CPIA), the Annual Report on Portfolio Performance (ARPP) relating to World Bank lending, and a special governance factor which itself is mostly derived from the CPIA. To a lesser extent, IDA allocation is also determined by recipient

PROPOSAL FOR A CONTINGENCY DEBT SUSTAINABILITY FRAMEWORK

2563

countries’ population and per capita gross national income, reflecting their need for aid. Both measures of performance and needs are then entered into a simplistic PBA allocation formula, which assigns the highest weights to CPIA and governance performance. In exceptional circumstances, the overall IDA Country Allocation Strategy (CAS) is adjusted in light of countries’ access to alternative World Bank lending, or their emergence from conflict or natural disaster.

12. Balassa (1982) applies this method to the analysis of the differences in policy responses to external shocks by outward- versus inward-oriented developing countries during the 1970s. Solis and Zedillo (1985) analyze the causes underlying Mexico’s debt build up and subsequent crisis during the early 1980s.

5. Nine percent of the volume discount is then redistributed among all eligible countries, through the PBA allotment mechanism.

14. The degree of agglomeration by export sectors in Eqns. (6)and (7), and by import sectors in Eqns. (8) and (9), is reflected in the number of subcategories included in index j. Ideally, the categories would range from the largest aggregate of the country’s total exports of goods and services, to the single most important export sectors and items. In practice, the informational requirement for a detailed disaggregation of trade items is too large for most LICs’ current institutional capacity to cope with.

6. More precisely, some measure of cross-country average vulnerability to external shocks is implicit in the indicative thresholds, since they are derived from probabilistic regressions including both CPIA and shock variables. However, the degree to which shocks are reflected by indicative thresholds is negligible, and depends on the actual model fitted for the estimation of the CPIA coefficient. 7. To date, the introduction of debt service modulation does not seem to be part of the ongoing negotiations in relation to the next, 15th, replenishment of IDA (IDA, 2007). 8. Among the alternative predictors, we tested for the World Bank’s Kaufmann–Kraay–Mastruzzi indicators (see Kaufmann, Kraay, & Mastruzzi, 2004). 9. To illustrate the point, it should be sufficient to note that any debt sustainability assessment along debt thresholds defined across country groups would be largely meaningless in the context of a contingency mechanism that would modulate the amount and grantshare of aid according to recipient countries realization of exogenous events. Put differently, country specificity of aid allocation would invalidate any debt sustainability assessment method that failed to internalize its benefits in terms of recipient countries sustainability. 10. Also see Combes and Guillaumont (2002) and Gilbert and Tabova (2005) for proposals similar to those envisaged by the World Bank. 11. For example, the amounts of a particular crop produced and exported may be adjusted as a deliberate policy choice in the face of an observable price trend. Therefore, export earnings from that particular crop are exogenous only with regard to the actual price trend and the effects of natural factors on yields, but not with regard to volume adjustments made in response to those external forces.

13. Adopting the terminology introduced by Balassa (1982).

15. Furthermore, if estimations of the price elasticity of specific import commodity items are available, these should be used to complement income elasticity in the calculation of overall import elasticity. 16. The basic structure of identity (10) is borrowed from Solis and Zedillo (1985), and has been substantially modified by the author. The symbol I represents a sum vector across the total number of export and import agglomerates, j. It should be noted that the sum of all exports effects is equal to the simple trend deviation: pm pXt X t  ðaÞ þ ðbÞ þ ðdÞ þ ðgÞ þ ðhÞ þ ½ðcÞ  ðpm t  t ÞM t  ¼ ð X pt X t Þ. Similarly, with regard to import items: ðeÞ þ ðf Þ þ ðiÞ þ ðjÞ þ ðkÞ þ ðlÞ þ ½ðcÞ þ ðpXt  pXt ÞX t  ¼ pM ðpM t Mt   t M t Þ. 17. Workers’ remittances is typically the most important item of private unrequited transfers to LICs, and explains our choice of labeling. The inclusion of specific items and exclusion of others from identity (10) will vary from country to country. 18. It is unlikely that the occurrence of real volume shocks due to natural disasters would be offset by the occurrence of positive real shocks. Of course, the events referred to in the text relate to the exceptional occurrence of disasters with significant impact, and not to the experience of more or less favorable climatic conditions, which should, at least in principle, average out. 19. It should be noted that, as is implicit in any insurance mechanism, the contingent credit line may also reduce a country’s incentive to improve its vulnerability to exogenous shocks over time. However, in the case of trend and other BOP factors, the risk factor of

2564

WORLD DEVELOPMENT

moral hazard mainly relates to the debtor’s attempt to increase the actual amount of compensation in any period t to the detriment of its longer-term development plans. For example, it cannot be excluded that the country authorities would have an incentive to cheat and reap higher compensation in period t, if, for some reason, they were to place a higher value on the maximization of short-term benefits than on the discounted value of the long-term benefits resulting from alternative policy actions leading to emancipation from vulnerability. 20. Ideally, but not necessarily for the CDSF to avoid the specific moral hazard implications addressed here, the contract would be the expression of a genuinely cooperative approach between the parties to the contract, rather than the imposition of a particular set of policies by donors. 21. For a detailed description of the IMF PRGF, see http://www.imf.org/external/np/exr/facts/prgf.htm.

22. We abstract here from any time-limits that may be attached to the application of the contingency mechanisms, including termination as a result of non-compliance. Otherwise, debt sustainability beyond the expiration of the contingency regime would also have to be assessed with regard to the country’s capacity to cope with shock and trend factors. 23. The results reported in this section rely on trend deviations calculated as simple differences between subsequent years (one-year moving averages), as a special case of the longer-term moving averages envisaged in identities (5) to (9). For a more detailed discussion of the CDSF application in relation to selected country studies the reader is referred to Ferrarini (2007) and related publications. 24. Net debt stock is computed as the total external debt stock minus stock/flow reconciliations and changes in cross-currency valuations.

REFERENCES Balassa, B. (1982). Structural adjustment policies in developing economies. World Development, 10(1), 23–38. Bird, G., & Milne, A. (2003). Debt relief for low income countries: Is it effective and efficient?. The World Economy, 26(1), 43–59. Brooks, R. (1998). External debt histories of ten lowincome developing countries: Lessons from their experience. Washington, DC: International Monetary Fund Report. Caliari, A. (2006). The New World Bank/IMF Debt Sustainability Framework—A Human Development Assessment. Background paper to the CIDSE/Caritas Internationalis Working Group on Resources for Development. Cashin, P., & McDermott, C. J. (2001). The long-run behavior of commodity prices: Small trends and big volatility. International Monetary Fund Working Paper WP/01/68, Washington, DC. Collier, P., & Dehn, J. (2001). Aid, shocks, and growth. Development Research Group Policy Research Paper No. 2688. The World Bank, Washington, DC. Collier, P., & Gunning, J. W. (1999). Trade shocks in developing countries. Oxford: Oxford University Press. Combes, J. L., & Guillaumont, P. (2002). Commodity price volatility, vulnerability and development. Development Policy Review, 20(1), 25–40. Deaton, A. (1999). Commodity prices and growth in Africa. Journal of Economic Perspectives, 13(3), 23–40. Deaton, A., & Miller, R. (1996). International commodity prices, macroeconomic performance and politics in Sub-Saharan Africa. Journal of African Economies, 5(3), 99–191.

Edwards, S. (2003). Debt relief and the current account: An analysis of the HIPC initiative. World Economy, 26(4), 513–532. Ferrarini, B. (2007). External debt and macroeconomic vulnerability: A proposal for state-contingent debt contracts to achieve low-income country debt sustainability. London. Ph.D. Thesis, SOAS, University of London. Froot, K. A., Scharfstein, D. S., & Stein, J. C. (1989). LDC debt: Forgiveness, indexation, and investment incentives. Journal of Finance, 44(5), 1335–1350. Gautam, M. (2003). Debt relief for the poorest: An OED review of the HIPC initiative. Washington, DC: The World Bank. Gilbert, C. L., & Tabova, A. (2005). Can we link concessional debt service to commodity prices? Working Paper No. 0508, Department of Economics, Universita` degli Studi di Trento, Italy. Grossman, H. I., & Huyck, J. B. V. (1988). Sovereign debt as a contingent claim: Excusable default, repudiation, and reputation. The American Economic Review, 78(5), 1088–1097. Guillaumont, P., & Chauvet, L. (2001). Aid and performance: A reassessment. Journal of Development Studies, 37(6), 66–92. Huberman, G., & Kahn, C. (1988). Limited contract enforcement and strategic renegotiation. The American Economic Review, 78(3), 471–484. IDA (2005a). Additions to IDA resources: Fourteenth replenishment. The World Bank, Washington DC: International Development Association. IDA (2005b). Managing the risk of exogenous shocks in low-income countries. The World Bank, Washington DC: International Development Association.

PROPOSAL FOR A CONTINGENCY DEBT SUSTAINABILITY FRAMEWORK IDA (2006). IDA’s implementation of the multilateral debt initiative. The World Bank, Washington DC: International Development Association. IDA (2007). IDA 15 second replenishment meeting—Key issues. The World Bank, Washington DC: International Development Association. IDA, & IMF (2003). Heavily indebted poor countries (HIPC) initiative—statistical update (April 2003). Washington DC: International Monetary Fund and The World Bank. IDA & IMF (2004a). Debt sustainability in low-income countries—Proposal for an operational framework and policy implications. Washington DC: International Monetary Fund and International Development Association. IDA & IMF (2004b). Debt sustainability in low-income countries: Further considerations on an operational framework and policy. Washington DC: International Monetary Fund and International Development Association. IDA, & IMF (2005). Operational framework for debt sustainability assessments in low-income countries— Further considerations. Washington DC: International Monetary Fund and International Development Association. IMF (2003a). Debt sustainability in low-income countries—Toward a forward-looking strategy. Washington, DC: International Monetary Fund. IMF (2003b). Fund assistance for countries facing exogenous shocks. Washington, DC: International Monetary Fund. IMF (2006). Guidance note on the exogenous shocks facility. Washington, DC: International Monetary Fund. Kanbur, R. (2005). Reforming the formula: A modest proposal for introducing development outcomes in Ida allocation procedures. Centre for Economic

2565

Policy Research (CEPR) Discussion Paper No.4971, London, UK. Kaufmann, D., Kraay, A., & Mastruzzi, M. (2004). Governance matters III: Governance indicators for 1996, 1998, 2000, and 2002. World Bank Economic Review, 18(2), 253–287. Killick, T. (2004). Politics, evidence and the new aid agenda. Development Policy Review, 22(1), 5–29. Kraay, A., & Nehru, V. (2004). When is external debt sustainable? World Bank Policy Research Working Paper, No. 3200, Washington, DC. Krugman, P. (1988). Financing vs. forgiving a debt overhang. Journal of Development Economics, 29(3), 253–268. Maizels, A. (1992). Commodities in crisis: The commodity crisis of the 1980s and the political economy of international commodity policies. Oxford: Clarendon. Maizels, A. (1994). Continuing commodity crisis of developing countries. World Development, 22(11), 1685–1695. Solis, L., & Zedillo, E. (1985). The foreign debt of Mexico. In G. W. Smith, & J. T. Cuddington (Eds.), International debt and the developing countries (pp. 258–288). The World Bank, Washington DC: International Bank for Reconstruction and Development. Varangis, P., Varma, S., dePlaa, A., & Nehru, V., (2004). Exogenous shocks in low income countries: Economic policy issues and the role of the international community. Background paper prepared for the Report: Managing the Debt Risk of Exogenous Shocks in Low-Income Countries. The World Bank. Williamson, J. (2005). Policies to reduce the vulnerability of low-income countries. In J. J. Teunissen, & A. Akkerman (Eds.), Protecting the poor: Global financial institutions and the vulnerability of low-income countries (pp. 14–34). The Hague: FONDAD.

Available online at www.sciencedirect.com