76754 Public Debt Management and Macroeconomic Stability: An Overview Peter J. Montiel Recent research suggests that management of the public sector’s debt can have important effects on a country’s macroeconomic performance. This article provides an overview of the factors that the recent literature has identified as important in determining the optimal composition of the public debt. Based on this analysis, it attempts to establish general guidelines for public debt management in emerging economies. To retain market access and promote domestic financial market development, governments should generally finance themselves at market rates using a wide variety of securities. Beyond this general principle, the optimal composition of the public debt involves a tradeoff between enhancing the government’s anti-inflationary credibility and reducing the vulnerability of its budget to macroeconomic shocks. Consequently, the optimal composition of the debt depends on a country’s circumstances. Debt should be heavily weighted toward long-term nominal secu- rities for governments that have anti-inflationary credibility and toward long-term indexed debt for those that do not. Macroeconomists have long been concerned with the economic effects of changes in the allocation of the public sector’s balance sheet between monetary and nonmone- tary liabilities. This is the traditional domain of monetary policy. Analysis of the macroeconomic effects of changes in the structure of the government’s nonmone- tary liabilities—the composition of the public debt—has received much less atten- tion.1 This situation has recently begun to change, however, as issues such as fiscal credibility and the role of balance sheet effects in triggering macroeconomic fluctua- tions have gained prominence in policy discussions. As a result, macroeconomists have increasingly been attracted to investigation of the potential macroeconomic effects of public sector debt management. This article provides an overview of the ways the structure of public debt can affect macroeconomic performance in emerging economies. Thus the article focuses © The Author 2005. Published by Oxford University Press on behalf of the International Bank for Reconstruction and Development / THE WORLD BANK. All rights reserved. For permissions, please e-mail: journals.permissions@oupjournals.org. doi:10.1093/wbro/lki006 20:259–281 on the liability rather than the asset side of the public sector’s comprehensive bal- ance sheet. Asset composition is taken as given. The article considers the factors that the recent macroeconomics literature has identified as relevant for determining the optimal composition of the public debt and synthesizes them into some general guidelines for public debt management in emerging economies. The first section evaluates the view that the objective of debt management should be to minimize the cost of debt service. Although reasonable from a norma- tive perspective, and an objective in the debt management practices of many coun- tries, it is too narrow in scope. Debt management should seek instead to achieve an optimal tradeoff among multiple and competing social objectives. The next section examines the conflict between the objective of minimizing debt servicing costs and the structural objective of promoting domestic financial development. It concludes that promoting financial development requires the government to eschew finan- cial repression—a policy motivated by fiscal considerations—and instead to finance itself by selling securities on market terms, even if that raises debt servic- ing costs. This leads naturally to the issue of the optimal composition of such secu- rities. The next three sections consider the choices between indexed and nominal debt, short-term and long-term debt, and domestic and foreign-currency debt. In all of these cases the government faces a similar fundamental tradeoff: between anti-inflationary credibility and robustness of the government’s budget in the face of shocks. The final section examines how the optimal choice among these tradeoffs is likely to be affected by country circumstances, thereby extracting some simple guidelines for debt management in emerging economies. Minimizing Debt Servicing Costs as an Objective of Public Debt Management A naive view of the optimal determination of the public sector’s debt structure is that the composition of the government’s liabilities should be chosen so as to minimize the costs of debt servicing. Although this objective may be naive, it is clearly not arbitrary, because it can readily be justified on welfare grounds. To see how, consider a government whose objective is to maximize social wel- fare. A useful benchmark is the case in which the government has access to cost- less lump-sum taxation and all public debt is held domestically. In this case, servicing the existing stock of debt simply represents a nondistortionary transfer within the domestic economy from taxpayers to holders of government debt. A welfare-maximizing government would have no incentive to seek to minimize debt servicing costs. 260 The World Bank Research Observer, vol. 20, no. 2 (Fall 2005) However, if taxes are distortionary or costly to collect, raising the revenue required to service the debt imposes an “excess burden� on the economy (in the form of collection costs and deadweight losses associated with distortionary resource allo- cation effects) that can be reduced if debt servicing costs are minimized.2 A reduction in debt service costs would increase social welfare, and minimizing such costs becomes a reasonable government objective.3 Under the assumption of full debt repayment, the incentive to reduce debt servicing costs in this case is further strengthened when at least some of the debt is held by foreigners. In that event the costs imposed by distortionary taxation are even less fully compensated by payments to domestic agents than when all debt is domestically held. However, the single-minded pursuit of this criterion may have harmful conse- quences for the other social objectives that a welfare-maximizing government would tend to value. To see this most simply, note that the most direct way to minimize debt service costs is just to repudiate the entire stock of interest-bearing debt. Debt repudiation is, at least to some extent, (by definition) the only recourse for an insol- vent government. But even a solvent government may have an incentive to repudiate. For example, the welfare-maximizing government described above could minimize the excess burden associated with distortionary taxes simply by opting not to pay its debt service obligations. Because creditors are aware of this incentive, sustaining an equilib- rium with positive levels of debt means that repudiation must entail a cost that induces the government to continue to service its debt. For a welfare-maximizing government, this cost must take the form of a sacrifice of other welfare-enhancing objectives. What might such other objectives be? The costs associated with debt repudiation are both direct and indirect. An obvious source of direct costs is the actions of credi- tors, who will appeal to the legal system to enforce their debt contracts and may be able to penalize the government for reneging on its obligations. Moreover, creditors’ ability to impose such costs may leave the status of the repudiated debt uncertain, creating a “debt overhang� problem that potentially distorts intertemporal relative prices in the domestic economy, to the detriment of macroeconomic performance. Indirect costs of repudiation may arise through reputation effects. If governments that repudiate are subsequently unable to borrow, the cost of repudiation is the loss of market access. This implies the necessity to rely on tax and money finance. Full tax financing of government expenditures may increase the present value of the excess burden associated with the financing of a given program of exhaustive gov- ernment spending (government spending on goods and services) if the excess bur- den associated with each dollar of tax revenue is an increasing function of the tax ratio (Barro 1979). Full tax financing, by causing tax rates to vary intertemporally with government expenditures, may also distort the intertemporal allocation deci- sions of economic agents by inducing them to redistribute their production and spending decisions over time. Moreover, it would involve raising tax rates during cyclical downturns and lowering them during booms, thus causing fiscal policies to Peter J. Montiel 261 behave procyclically. Finally, full tax funding forces the current generation to bear the burden of public sector capital expenditures that will benefit future generations, thus violating the benefit principle of public finance. Full tax funding thus involves unacceptable sacrifices of other valuable social objectives of efficiency, stability, and equity. The alternative of financing deficits by printing money runs the risk of subjecting the economy to high and unstable infla- tion. The increased transaction costs, and the instability of intratemporal and inter- temporal relative prices that this entails, have adverse consequences for efficiency and equity that have always rendered pure money financing unacceptable. In short, one reason that debt is not repudiated is that although it is costly to ser- vice debt, it may be even more costly not to do so. Welfare-maximizing governments do not routinely repudiate debt, despite the excess burdens associated with servicing it, because doing so entails an unacceptable sacrifice of the important economic objective of retaining market access. But nonrepudiation implies that governments will typically find themselves with a stock of financial liabilities that includes some interest-bearing debt. They then face the debt management problem of determining the composition of those liabilities so as to achieve desirable outcomes on a set of multiple—and possibly conflicting—social objectives. Debt Management and Financial Development With debt repudiation likely to be prohibitively costly under normal circumstances, governments typically seek to solve their debt management problem subject to the constraint that debt be serviced on schedule. This does not prevent the government from adopting the objective of minimizing public sector borrowing costs as the over- riding consideration guiding its debt management policies, subject to that constraint. This would suggest choosing the composition of the debt to reduce its carrying cost to the greatest extent possible, leaving all other considerations aside. The incentive for a welfare-maximizing government to do so is the same as the incentive to repudi- ate debt: to minimize tax collection costs and the resource misallocations associated with distortionary taxation. One obvious way to minimize debt servicing costs is to issue debt at below-market interest rates. The question, of course, is why anyone would choose to hold such debt, and the answer is that (by definition!) no one would. Thus the government would have to compel private agents to hold it. Strictly speaking, this is more akin to a form of taxation than to partial repudiation, because creditors know ex ante that they will be receiving below-market returns. Various policies associated with financial repression have precisely this effect, includ- ing ceilings on deposit and lending rates, controls on capital outflows, public ownership and management of commercial banks, and reserve and liquidity requirements imposed 262 The World Bank Research Observer, vol. 20, no. 2 (Fall 2005) on private commercial banks.4 The effect of such policies is to create artificial demand for government securities, enabling the public sector to borrow at rates below those that would prevail if creditors had more choice in the assets they could hold. However, although such policies may carry the benefit of reducing the government’s financing costs, they also carry the important social costs of distorting resource alloca- tion and impeding financial development. Distortions in the allocation of resources arise because the portfolio restrictions that policies of this type place on financial institu- tions compromise their ability to perform the financial intermediation function effi- ciently, preventing financial institutions from allocating the funds that the government absorbs to activities that offer higher returns. Moreover, the implicit tax that these restrictions place on the financial sector reduces the incentives for investment in formal financial intermediation, potentially favoring less efficient types of intermediation. Financial development is impaired not only through this form of taxation but also because government financing through these means, as opposed to selling government securities in an open primary market, deprives the market of a benchmark for pricing risk in the domestic economy (in the form of a market-determined interest rate on rela- tively safe government securities) as well as of other means to aggregate and dissemi- nate information through the pricing of securities (through the term structure, return differentials on nonindexed and indexed bonds, and so on). These observations suggest that reducing debt servicing costs through financial repression is socially costly, and given the important role that recent research suggests that financial development can play in promoting economic growth, these costs may be quite large (Levine 1997).5 The message is that governments that place high value on promoting economic growth are likely to find it optimal to issue liabilities that pro- mote financial development even though that increases their borrowing costs. From the standpoint of promoting the development of an efficient domestic financial system, it is preferable for the government to sell its securities in undistorted primary markets and to issue securities with diverse characteristics, to facilitate market processing of financially relevant information, than to place government obligations with financial institutions that are provided with artificial incentives to hold those obligations. However, this prescription hardly begins to solve the problem of optimal debt management, because it leaves the government with many choices to make. For example, the government can issue securities that are denominated in nominal terms or that are indexed to the domestic price level, that are of short or long maturity, and that are paid in domestic or foreign currency. These choices are examined in turn. Credibility and Hedging: Indexed or Nominal Debt Assuming that government interest-bearing liabilities consist of securities that are issued in open markets and that are expected to be serviced on contractual terms, Peter J. Montiel 263 should the government issue nominal debt (denominated in the domestic currency) or issue at least some share of its debt indexed to the domestic price level? The following section considers the role of optimal debt maturity; for now, all debt is assumed to be long-term debt. Recent research has emphasized that the existence of a large stock of long-term nominal debt may have important macroeconomic consequences, even if the gov- ernment’s solvency can be taken for granted. These consequences help us under- stand the conditions that affect the optimal share of the government debt that should be indexed to the price level. The key issue concerns a tradeoff between credi- bility and robustness: between the government’s anti-inflationary credibility and the avoidance of risks to the government’s solvency or to the economy’s welfare that can be caused by fluctuations in the government’s financing needs. Time Inconsistency A potential consequence of a large stock of long-term nominal debt is that it may aggravate the time-inconsistency problems associated with monetary policy when the central bank lacks a precommitment mechanism. Time inconsistency arises when the sequence of policy choices that a government would make over some time horizon when it exercises discretion in each period are different from those that it would make over the same horizon if it could fully precommit its future behavior at the beginning. Consider the simple framework pioneered by Barro and Gordon (1983). In this framework, if unanticipated inflation tends to increase the level of output and distor- tions in the economy make the “natural� level of output suboptimally high, a welfare-maximizing government that cannot precommit its future behavior (and thus that retains policy discretion in every period) may be tempted to generate “surprise� inflation in each period to move real output closer to its optimal level. That temptation is assumed to be restrained by social costs of inflation that increase nonlinearly. The Barro-Gordon analysis shows that under these circumstances the equilib- rium rate of inflation turns out to be suboptimally high and the level of social welfare suboptimally low compared with the case where the government could precommit its actions. The reason is that the government’s attempt to generate surprise infla- tion must fail in a rational-expectations equilibrium in which the public under- stands the government’s motivation. Even so, the equilibrium rate of inflation must nonetheless be positive in this case, determined such that the marginal social benefits of inflation (which are constant) are equalized to the marginal social costs (which increase nonlinearly from an initial value of zero). Welfare must be lower in the discretionary equilibrium than in the nondiscretionary equilibrium in which the government can precommit itself to producing zero inflation, because the level of real output must be the same in both 264 The World Bank Research Observer, vol. 20, no. 2 (Fall 2005) equilibria, whereas the rate of inflation is higher in the discretionary case. The key point here, however, is that in the discretionary case the equilibrium rate of inflation is higher—and thus the loss of social welfare is greater—the greater the social bene- fit from inflationary surprises and the smaller the social cost of inflation. The presence of long-term nominal debt affects this outcome by reducing the net social costs associated with inflation. Because the contractual interest rate on such debt is fixed, an increase in the rate of inflation that was unanticipated at the time the debt was incurred acts like a capital levy, reducing the real value of such debt. Thus, the existence of long-term nominal debt effectively increases the base of the inflation tax. This creates social gains arising from the fact that the reduced real value of the debt permits the government to reduce the level of taxation, decreasing tax collection costs as well as the resource misallocations associated with distortion- ary taxation. These gains partly offset the social costs of inflation. But because the net social costs of inflation are thus reduced, the government’s incentive to act in a discretionary manner and to engineer an inflationary surprise is magnified. That in turn reduces the government’s anti-inflationary credibility and increases the econ- omy’s equilibrium rate of inflation. Thus, the effect of a large stock of long-term nominal debt is to increase the magnitude of the social loss associated with the discretionary equilibrium (see Calvo 1991). It is worth reiterating that this problem arises only when the government is not prevented from acting in a discretionary fashion by reputational considerations or by institutional mechanisms that prevent it from attempting to generate surprise inflation, such as participation in a currency board or currency union (see Alesina and Barro 2000). In addition, when such restraints are not present, the mechanism suggests that the loss of anti-inflationary credibility arising from the presence of nominal debt should be larger not just the larger the stock of nominal debt but also the greater the tax collection and distortion costs that the government has to incur to raise an additional unit of revenue, because the larger such costs, the greater the gain in social welfare from an inflationary erosion of the real value of nominal debt. In turn, the convexity of the tax collection cost function implies that these costs will be larger the larger the level of government spending to be financed. All of these con- siderations suggest that the magnitude of this credibility effect depends on a specific set of country circumstances. How can this problem be alleviated? Taking as given the absence of institutional constraints on government discretion, the properties of its revenue system, the size of the expenditures to be financed, and the total stock of outstanding debt, debt man- agement policies provide one solution. Calvo (1988), for example, suggests that the loss of credibility can be ameliorated by increasing the share of long-term debt that is indexed to the domestic price level, because the government’s credibility problem is aggravated by the existence of long-term nominal debt. Indeed, the role of debt in aggravating time-inconsistency problems can be eliminated entirely by full indexation Peter J. Montiel 265 of the debt. Notice that a government that is more likely to resist inflation than the public gives it credit for would find the issuance of indexed debt particularly attrac- tive, because in addition to the credibility gains, issuing indexed debt would also reduce its debt servicing costs by lowering the real interest rates it would have to pay. (This assumes, of course, that the government has better information about its future intentions than the private sector does.) Thus, the objectives of increasing the government’s anti-inflationary credibility and reducing its debt servicing costs both suggest that governments lacking anti-inflationary credibility should replace nomi- nal debt with indexed debt. These arguments suggest that the optimal share of long-term nominal debt in the portfolio of a government that retains discretion and lacks anti-inflationary credibil- ity should be zero. Since reliance on indexed debt would eliminate the possibility of a “bad� (high inflation) equilibrium for such governments, the question that arises is whether governments in these conditions should ever choose to issue nominal debt at all. Stabilization of Tax Rates across States of Nature Bohn (1988) offers one reason why governments might still wish to issue nominal debt. Consider a world of uncertainty in which the government’s budget is subject to random shocks. Suppose that markets are incomplete, in the sense that the govern- ment cannot structure its debt as state-contingent contracts with payoffs condi- tioned on the effects of shocks on the government’s budget (such contracts would promise larger payoffs to creditors when shocks are favorable and smaller ones when they are unfavorable). Shocks to the government’s budget would then require offsetting discretionary changes in tax revenues—either in the present or in the future. If, as assumed earlier, raising tax revenue by altering tax rates is subject to convex costs (the marginal cost of raising tax revenue increases with the tax rate), then the variability of tax rates across states of nature associated with these shocks would tend to increase the expected burden of distortionary taxes and thus reduce economic welfare, just as varying tax rates across time would. (Indeed, if the government’s budget is sufficiently vulnerable to shocks, and the variability of potential shocks is sufficiently large, the social costs of raising the required revenue in the face of an adverse shock may be large enough to trigger an optimal default.) Thus organizing the government’s finances so as to equalize expected tax rates across states of nature is a desirable goal. What does this have to do with the structure of the government’s debt? Suppose that in an attempt to minimize time-inconsistency problems, all debt is indexed to the price level, as suggested. Although this might reduce the government’s incentive to try to engineer inflationary surprises, in a world of uncertainty it does not guarantee 266 The World Bank Research Observer, vol. 20, no. 2 (Fall 2005) that unexpected inflation will not occur (for example, as a result of shocks to the demand for money). Suppose that unanticipated inflationary shocks give rise to pos- itive correlations between the price level and the government’s real financing needs. This might happen because the shocks reduce income levels and thus tax revenues, because taxes are defined in nominal terms or paid with a lag so that as the price level rises their real value declines (known as the Olivera-Tanzi effect). Or it might happen because the shocks represent unexpected changes in government spending, as in the case of wars or natural disasters.6 If the shocks do give rise to positive corre- lations between the price level and the government’s real financing needs, the addi- tional revenue that would have to be raised through distortionary taxation would be greater the greater the extent to which the government’s debt is indexed to the domestic price level. Indeed, under such shocks, fully indexed debt would have the effect of maximizing the variability in the government’s financing needs and thus of increasing the average distortionary effects of taxation, relative to a situation in which some portion of the debt is nominal. Bohn’s (1988) observation is therefore that with uncertainty and incomplete markets, nominal debt can provide the government with a valuable hedge and be welfare improving by reducing the excess burden of higher tax revenues required to finance a given plan of exhaustive public spending. If changes in the price level are positively correlated with changes in the government’s financing needs, an increase in the price level would reduce the real costs of servicing the debt just when the government’s financing needs are increasing, whereas a reduction in the price level would increase the real burden of servicing nominal debt just when the government can most afford it. Thus, having some nominal debt helps stabilize tax rates across states of nature, which is welfare improving. Of course, this benefit would need to be offset against the adverse credibility effects described previously, but because such effects would tend to disappear when the stock of nominal debt is zero, in general this argument would justify at least some positive level of nominal debt, even for governments with imperfect anti-inflationary credibility. The Optimality of Nominal Debt These arguments indicate the tradeoffs that the government faces in choosing the optimal composition of debt between indexed and nonindexed liabilities. It can enhance its anti-inflationary credibility by opting for a larger share of indexed debt, but only at the expense of greater budget vulnerability to certain types of shocks, thus making the economy more susceptible to the distortions associated with vari- ability in tax rates across states of nature and possibly increasing the likelihood of default. In short, the tradeoff is between anti-inflationary credibility and the robustness Peter J. Montiel 267 of the government’s budget in the face of shocks. What considerations should gov- ern this choice? First, as already argued, a critically important factor is the extent to which the government can independently precommit its future actions. The greater the amount of precommitment possible, the greater the hedging benefit of nominal debt relative to its credibility cost. Thus, a greater ability to precommit (say, because of participation in a currency board or monetary union) suggests a lower optimal degree of indexation and therefore a larger optimal share of nominal debt. In the limit, if the government can fully precommit, there is no credibility benefit to index- ation. In this case if positive inflation shocks have adverse effects on the budget, hedging considerations favor the issuance of nominal debt. Second, the choice depends on the magnitude of the social costs of inflation. The higher the social costs of inflation, the more important it is to avoid the discretionary outcome and thus the larger the optimal share of indexed debt. Third, given an incomplete ability to precommit, the costs associated with a larger share of nominal debt (a lower degree of indexation) depend on the additional incen- tives that a larger share creates at the margin for the government to turn to infla- tion. Thus the optimal degree of indexation should increase with the level of debt, because a given increment to inflation has a larger positive budgetary impact the larger the stock of nominal debt. Fourth, as mentioned, the tradeoff depends on the relative empirical importance of shocks that give rise to a positive correlation between the government’s financing needs and the price level, compared with those that do not.7 Fifth, it depends on the variability of the shocks just described. The correlation between the government’s financing needs and the price level depends both on the frequency with which such shocks arise relative to other types of shocks and on their magnitude when they do arise. The greater the variance of shocks, the greater the excess burden associated with raising a given expected level of taxation, and the more valuable the hedge provided by nominal debt. Finally, the severity of the excess burden associated with higher tax rates has an ambiguous theoretical effect on the choice between nominal and indexed debt. The greater the excess burden associated with higher tax rates, the greater the incentive for the government to rely on the inflation tax and the smaller the government’s anti-inflationary credibility, implying larger credibility gains from the use of indexed debt. However, a more distortionary tax system also magnifies the costs associated with variations in tax rates across states of nature, which tends to increase the hedg- ing benefit associated with nominal debt. The upshot is that the tradeoff between credibility and robustness that the gov- ernment faces in choosing between nominal and indexed debt depends on charac- teristics of the domestic economy. Theory can pin down what these are, but it does not suggest that either type of debt is superior to the other under all circumstances. 268 The World Bank Research Observer, vol. 20, no. 2 (Fall 2005) Short-Term versus Long-Term Nominal Debt The previous section considered the tradeoff between nominal and indexed debt, holding the maturity of the debt constant. Leaving aside for now the possibility of indexing the debt, and thus taking all debt as nominal, this section turns to the choice between short-term and long-term nominal debt. As will become clear, the choice between short-term and long-term debt involves a tradeoff between the government’s anti-inflationary credibility and the robustness of its budget, just as is true in choosing between nominal and indexed debt. Short- term nominal debt may not be as effective as long-term indexed debt in projecting credibility, and it leaves the government vulnerable to different types of risk than does indexed debt. In particular, although short-term nominal debt may protect the government’s finances from the inflation shocks to which indexed debt makes it vul- nerable, it exposes the government to real interest rate and rollover risks.8 In addi- tion, the existence of a large stock of short-term debt may constrain monetary policy when the government’s solvency is perceived to be precarious. Through that chan- nel it may also render an economy that operates with an officially determined exchange rate vulnerable to self-fulfilling currency crises. Anti-Inflationary Credibility versus Robustness Why do governments borrow in the short term? In a world of asymmetric informa- tion and moral hazard, creditors lend at short maturities to monitor and control bor- rowers. Creditors see the recurrent capacity to repay loan principal as a sign of continued solvency, and they use the threat of withdrawing funds or renegotiating at much higher interest rates as disincentives for borrowers to behave in ways that undermine the creditors’ interests. Thus when information and moral hazard prob- lems are acute, short-term loans will be relatively cheap compared with long-term loans. Under these circumstances, the objective of minimizing borrowing costs will tempt governments to opt for short-term financing. The second reason governments may choose to borrow in the short term is to enhance their anti-inflationary credibility. Unlike the case with long-term debt, with short-term debt the interest rate is continuously renegotiated, so creditors can adjust the contractual interest rate to compensate for the effects of anticipated inflation in reducing the real values of their claims. This means that, unlike with money and long-term nominal debt, the government may not be able to regard its short-term liabilities as part of the inflation tax base, and this would tend to reduce the incentive to renege on inflationary commitments that gives rise to the time-inconsistency prob- lem. Thus, governments may issue short-term debt to “tie their hands� and remove the temptation to inflate that arises from a large inflation tax base. In this way, gov- ernments can enhance the credibility of their anti-inflationary commitments. Peter J. Montiel 269 Short-Term Debt and Time Inconsistency. Just how effective is short-term debt in enhancing credibility? At least two reasons have been proposed for doubting its effectiveness. The first is that government may be able to tax short-term debt with inflation, even if the public can anticipate it. The second is that the existence of short-term debt can make the economy vulnerable to multiple equilibria, with the “bad� equilibrium characterized by low credibility and high inflation. Concerning the first argument, the key point is that the government can use an increase in anticipated inflation to “tax� short-term debt if the increase is associated with a reduction in the equilibrium value of the real interest rate—if the nominal interest rate on such debt does not fully adjust to an increase in expected inflation. In a financially open economy, under officially determined exchange rates, Calvo (1989) notes that the central bank may be able to use monetary policy to engineer an increase in the inflation rate that exceeds its preannounced rate of exchange rate depreciation, causing the real exchange rate to appreciate and the real interest rate to fall. The government would thus be able to erode the real value of short-term debt by increasing the rate of inflation. Turning to the second argument, Calvo (1988) posits that the existence of nominal debt with an interest rate that is responsive to expected inflation (as would be true of short-term debt) makes the government vulnerable to “confidence crises� that can result in high-inflation equilibria. The mechanism is as follows. The government’s incentive to inflate in the absence of precommitment depends on the extent to which it can reduce other forms of distortionary taxation by doing so. That is influenced by the size not only of the inflation tax base (which determines the revenue from infla- tion) but also of the budget gap, which will determine the benefits from inflationary taxation (recall that if the disortionary costs of tax revenues are a convex function of the size of revenues, these costs will increase with the size of the budgetary gap to be filled). Thus the temptation to inflate is an increasing function of the debt service payments to which the government is obligated. That means that it depends both on the interest rate and on the size of the nominal debt. But if creditors believe that some of the real value of short-term debt may be elimi- nated through inflation, they will demand a yield premium that compensates them for expected inflation and the attendant risk. This will itself increase the govern- ment’s debt service obligations and thus its incentive to inflate. The positive depen- dence of the incentive to inflate on the value of the nominal interest rate and of the nominal interest rate on the perceived incentives for the government to inflate cre- ate the possibility of multiple, self-fulfilling equilibria. That is, the high nominal interest rates caused by a lack of confidence may induce the government to inflate the debt away. Thus two equilibria could arise: a good equilibrium with low inflation and low nominal interest rates and a bad equilibrium with high inflation and high nominal interest rates. The implication is that short-term debt maturities may make a high-inflation equilibrium more rather than less likely. 270 The World Bank Research Observer, vol. 20, no. 2 (Fall 2005) Risk Exposures with Short-Term Debt. The credibility problems just examined arise from the government’s actual or perceived ability to generate endogenously an increase in the rate of inflation without a commensurate increase in the interest rate that it has to pay on its short-term debt—or, put differently, from its actual or per- ceived ability to reduce the real interest rate on such debt. But real interest rates on short-term debt may also fluctuate for exogenous reasons (such as from changes in country risk premia associated with international contagion), and such real interest rate movements, like price-level shocks in the case of indexed debt, may induce sub- optimal variability in tax rates. Thus short-term debt may (or may not, as shown) enhance the government’s anti-inflationary credibility. But to the extent that it does so, this benefit comes at the cost of exposing the budget to a new type of risk: the risk associated with market fluctuations in real interest rates. However, matters may be worse than that. Alesina and others (1992) note that a large stock of short-term debt can create vulnerability to self-fulfilling confidence cri- ses in which otherwise solvent governments default on their debt obligations. To see how this can happen, recall that what creates the possibility of a bad equilibrium in the Calvo framework is the government’s reluctance to make the fiscal adjustment required to meet a crisis-driven increase in debt service requirements because of the social costs associated with raising the requisite tax revenue. The Alesina and col- leagues model relies on a similar mechanism: the convexity of tax collection costs. That convexity means that the cost of raising an incremental amount of public reve- nues rises as with the amount of revenues the public sector has to raise. Consider how this affects government decisions on the servicing of short-term debt. As long as creditors are willing to roll over any short-term debt coming due on the same terms, the government does not have to raise additional revenue to service the debt. But if creditors increase the real interest rate on debt they agree to roll over, or if they refuse to roll over the debt, the government can continue to service the debt only by increasing tax revenues, either in the future (if creditors increase the interest rates that they demand to roll over the debt), or in the present (if creditors refuse to roll over any debt). The key question is whether the government will continue to service the debt on schedule when it has to raise more of its own resources to do so. The answer depends on the impact that higher real interest rates or a refusal to roll over would have on the incremental revenues that the government has to raise. Because interest rates are free to adjust and principal payments are higher, the shorter and more bunched debt maturities are, the more sensitive the need for incre- mental revenues will be to any increases in rollover costs or to a refusal of new lend- ing. Thus the shorter and more bunched debt maturities are, the more likely the government is to refuse to raise the revenues required to service the debt in the event of an increase in interest rates required to roll over the debt or a refusal of new lending. Of course, if the government is perceived as unlikely to honor its debt obligations, creditors will be reluctant to take on its short-term liabilities. That means that a Peter J. Montiel 271 reluctance by creditors to roll over debt for fear of default can be self-fulfilling: when creditors become unwilling to roll over short-term debt the government is more likely to default, because it would then be called on to make payments out of resources that would be too costly for it to raise. Thus a short maturity structure of the public debt may increase the likelihood of a confidence crisis on the debt: the shorter and more concentrated the debt maturities, the greater the government’s vulnerability to confidence crises. In this case, such crises take the form of debt runs. The well-known case of Mexico’s tesobono obligations in the first quarter of 1995 provides a dramatic example of such a debt run (see Sachs and others 1996). Giavazzi and Pagano (1990) summarize these results by noting that the likeli- hood of a Calvo-style bad equilibrium depends on three things: the size of the public debt, its maturity structure, and the time pattern of maturing debt. The logic, as already explained, is that if a substantial amount of debt has to be serviced at a point in time, and if a confidence crisis breaks out at that moment (say, fear of a repudia- tion or a devaluation), the treasury would have to refinance a large portion of its debt on unfavorable terms. The welfare cost of doing so would be high, and thus the likelihood that the government will repudiate is greater. This makes the confidence crisis more likely to happen. Giavazzi and Pagano argue that under these circum- stances, good debt management calls for the issuance of long-term indexed debt to push the economy to a good equilibrium because such debt cannot be monetized away and does not create large short-run amortization obligations.9 Short-Term Debt and Monetary Policy When the government maintains a large stock of short-term debt, the sensitivity of the government’s budget to changes in interest rates can affect macroeconomic sta- bility in indirect ways as well as by increasing vulnerability to debt runs. When the stock of interest-sensitive short-term debt is large and the government’s solvency is precarious, the adoption of tight monetary policy carries the risk of triggering fiscal insolvency by increasing the government’s debt servicing costs. The central bank will thus be constrained from adopting policies that it might otherwise have found necessary to stabilize the economy in response to shocks. This is bad enough under floating exchange rates, because one of the virtues of floating is to allow scope for an independent monetary policy in response to shocks that are asymmetric with those of a country’s trading partners. This constraint essentially renders such scope asymmetric, permitting monetary policy to act when expansionary policies are called for, but not (or at least only to a limited extent) when contraction is indicated. But under officially determined exchange rates, this constraint can be a recipe for severe macroeconomic instability by making the economy vulnerable to self-fulfilling currency crises. The logic of second-generation models of currency crises suggests 272 The World Bank Research Observer, vol. 20, no. 2 (Fall 2005) that in assessing the sustainability of an exchange rate peg, speculators evaluate the benefits and costs to the central bank of sustaining a high interest rate defense of the peg. When speculators perceive that the costs to the central bank of sustaining a high interest rate defense exceed the benefits of sustaining the peg, they will judge the prospects for a successful attack to be good, and that will make an attack more likely. Because the vulnerability of the public sector’s solvency to high interest rates is precisely the type of factor that would be perceived by the central bank as making the costs of sustaining a high interest rate defense unbearably high, the combination of a large stock of short-term debt with a precarious fiscal position greatly increases a country’s vulnerability to a successful speculative attack. This issue is of more than academic interest. Observers judge it to have played a key role in some of the more important currency crises of the 1990s. In the 1999 Brazilian crisis, for example, fiscal vulnerabilities associated with short-term govern- ment debt seem to have played an important role in inducing the government to float the real, despite the key role that the exchange rate peg had played in the Real Plan’s exchange rate-based inflation stabilization since 1994 (Razin and Sadka 2004). Domestic-Currency versus Foreign-Currency Debt Choices between nominal and indexed debt and between short-term and long-term nominal debt imply tradeoffs between anti-inflationary credibility and the robust- ness of the government’s budget. In the case of short-term debt, macroeconomic sta- bility is also affected by the potential constraints imposed on an independent stabilization instrument, monetary policy. This section turns to the choice between domestic-currency (nominal) debt and foreign-currency debt. As in the previous cases the choice involves a tradeoff between anti-inflationary credibility and budget robustness. As in the other cases, the type of vulnerability created in the government’s budget is specific to the type of debt instru- ment adopted to increase anti-inflationary credibility. As in the case of short-term debt, the existence of a large stock of foreign-currency debt may impose constraints on the actions of the central bank when the government’s solvency is at risk. But in this case the constraint is on exchange rate policy rather than on monetary policy. Finally, there is an interesting interaction between the analysis of the implications of short- term debt and those of foreign-currency debt. This section takes up each of these issues in turn. Because the basic analysis is now familiar, the discussion will be brief. Credibility versus Robustness As is true of indexed and short-term debt, the government can use the issuance of debt denominated in foreign currency as a tool for gaining credibility, on the Peter J. Montiel 273 assumption that such debt cannot be inflated away. But as with short-term debt, this assumption is questionable. With short-term debt, the government could in effect tax away some of the real value of the debt through inflation if its actions succeeded in reducing the real interest rate. Similarly, inflationary erosion of the real value of foreign-currency debt is possible if nominal shocks succeed in appreciating the real exchange rate. Under officially determined exchange rates, one-time shocks to the domestic price level can indeed succeed in doing so, at least temporarily until the real exchange rate returns to its equilibrium value. But this opportunity to reduce the real value of foreign-currency debt through domestic inflation may not under- mine the government’s anti-inflationary credibility if the public believes that the government will be unwilling to countenance the associated real exchange rate appreciation. Under these circumstances denominating debt in foreign exchange may enhance the government’s anti-inflationary credibility. However, any such gain in credibility comes at a cost: the assumption by the gov- ernment of exchange rate risk. A real exchange rate depreciation triggered by inde- pendent events will increase the government’s debt servicing costs and thus potentially subject the economy to undesirable fluctuations in distortionary taxes or, in the extreme, to the consequences of a government default, as in Argentina in 2002 (Mussa 2002; Serven and Perry 2004). The story, then, is the same as before. What changes in this case is the type of risk exposure undertaken by the govern- ment in its attempt to gain credibility. It is worth noting that as in the case of indexed and short-term debt, foreign- currency debt is likely to be cheaper than domestic-currency debt, precisely because the government, and not its creditors, takes on the foreign-currency exposure and attendant risk. This is particularly important when the government lacks nominal credibility, causing creditors to assess the risk of future exchange rate changes as being high. This situation tends to increase the government’s relative cost of bor- rowing in domestic currency. The supply price of domestic-currency funding is par- ticularly high from creditors with substantial foreign currency exposure, such as external creditors. In the limit, such creditors may demand such a high price for domestic currency lending that no such lending would be forthcoming from them. (Eichengreen and Hausmann 1999 dub this situation “original sin� and consider it a structural characteristic of external borrowing by developing countries.) The net effect is to increase the relative cost of domestic-currency borrowing, because such funds would be forthcoming only from domestic creditors. Other things being equal, this would tend to shift the optimal composition of the government’s liabilities toward foreign-currency debt. Thus, the objective of minimizing borrowing costs is likely to once again align itself with that of enhancing anti-inflationary credibility to induce reliance on foreign-currency debt.10 But how large is the particular type of risk exposure induced by foreign-currency debt likely to be for the government? It is easy to see that the answer would tend to 274 The World Bank Research Observer, vol. 20, no. 2 (Fall 2005) depend on the currency composition of the government’s finances, on the probabili- ties associated with exchange rate changes, and on the expected magnitude of such changes if they occur. Clearly, if the government holds a large stock of foreign- currency assets (such as foreign exchange reserves) or if a large component of its revenues is indexed to the exchange rate (say, if the government derives substantial revenues from the country’s export earnings), the government can sustain a corre- spondingly large stock of foreign-currency debt without exposing itself to exchange rate risk. Similarly, the risk exposure associated with foreign currency debt may not be inordinately high if the risk of exchange rate movements is slight (because the government maintains a credible hard peg, for example, or because the possible con- ditions that would render a self-fulfilling speculative attack more likely to succeed— such as a high degree of capital mobility—do not hold). Foreign-Currency Debt and Exchange Rate Policy As mentioned, the presence of a large stock of foreign-currency debt is also likely to affect macroeconomic stability indirectly, through its effects on the central bank. Under floating exchange rates, this creates an asymmetry in the conduct of mone- tary policy, because the central bank has an incentive to resist currency depreciation but not appreciation. The constraints imposed on exchange rate flexibility in devel- oping economies by the existence of a large stock of foreign currency–denominated debt have been emphasized in the “fear of floating� literature (Calvo and Reinhart 2002). Blanchard (2004) describes a more complicated mechanism through which a large stock of foreign-currency debt can impede the use of monetary policy for infla- tion targeting. This mechanism relies on the interaction between short-term (or variable-interest) debt and foreign-currency debt. When government debt is short- term or variable-interest, a change in domestic interest rates will affect the govern- ment budget, as discussed. An increase in domestic interest rates in response to an anticipated increase in inflation by an inflation-targeting central bank may thus increase default probabilities for government debt. In turn, this requires an exchange rate depreciation to maintain balance of payments equilibrium. However, when the government’s stock of foreign-currency debt is high, this depreciation fur- ther increases the default probability, which magnifies the required exchange rate depreciation. This depreciation has the effect of increasing expected inflation, ren- dering tight money an ineffective means to control inflation. The suggestion is that controlling inflation under these conditions calls for a fiscal response. Under “soft� exchange rate pegs, on the other hand, what is constrained when the stock of foreign-currency debt is high and the government’s solvency is precarious is the central bank’s exchange rate policy, rather than its monetary policy. Under these conditions, the central bank will have a strong incentive to avoid a devaluation or a regime switch that would result in a substantial depreciation of the currency. Peter J. Montiel 275 Again, the issues discussed in this section have been of tremendous practical importance among emerging economies. Not only did fiscal vulnerability to exchange rate movements as a result of large stocks of foreign currency debt contrib- ute to the government insolvencies associated with the Latin American debt crisis of the 1980s, but some observers have also blamed a large stock of contingent govern- ment liabilities that were essentially indexed to the exchange rate for triggering the Asian financial crisis of 1997 (Burnside and others 1998). Less dramatically, Blanchard (2004) provides evidence that the mechanism described in the previous paragraph was operative in constraining the Brazilian Central Bank from raising real interest rates in the face of an increase in expected inflation in late 2002. Short-Term Foreign-Currency Debt It is worth noting an important potential interaction between short-term debt and foreign-currency debt: the debt run outcome as a potential risk incurred by a gov- ernment carrying large amounts of short-term debt may actually be much more likely when this debt is denominated in foreign currency. To see why, notice that the government may always avoid defaulting on short-term debt denominated in domestic currency in the event of a run simply by printing money, as long as it is willing to live with the resulting inflation. In other words, when short-term debt is denominated in domestic currency, the government at least has a choice between default and inflation. No such choice is available when short-term debt is denomi- nated in foreign currency. Because the government cannot print foreign currency, if its liquid reserves are insufficient to pay off the creditors who “run,� it faces only the choice of resorting to distortionary taxation or defaulting. Guidelines for Public Sector Debt Management What do we learn from all of this about management of public debt in emerging econo- mies? This section draws out some general principles suggested by the preceding analysis. The appropriate conduct of fiscal policy implies that there will be times when for tax-smoothing, countercyclical, or intergenerational equity reasons, it will be optimal for the government to borrow. Preserving this option means that the debt can be repudiated only when such an action does not permanently impair the government’s capacity to borrow—in other words, only under exceptional circumstances that are clearly outside the government’s control or in the con- text of a clearly identified and credible regime change. This means that debt service payments will inevitably exert a claim on the resources of most emerging economy governments. 276 The World Bank Research Observer, vol. 20, no. 2 (Fall 2005) Because raising the resources to service debt is costly, and the marginal cost of doing so is likely to increase the larger is the volume of resources that have to be raised, managing the composition of the debt to minimize debt service costs is justifi- able from the perspective of a welfare-maximizing government. However, single- minded pursuit of this objective could be socially harmful if it imperils other, equally worthy social objectives. Reliance on financial repression to reduce debt servicing costs is a clear example. The static and dynamic efficiency gains that are sacrificed when financial repression stunts the development of the domestic financial system suggest that this is a socially extremely harmful way for the government to seek to reduce the costs of meeting its financing needs. The implication is that the govern- ment should finance itself by issuing securities that are sold on market terms. The objective of promoting financial development suggests that the government should issue a diverse set of securities in order to facilitate information aggregation and dissemination in financial markets. These broad principles, however, leave the optimal composition of this diverse set of securities unspecified. In principle, such securities could be of varying maturities and could be denominated in domestic currency or indexed to the domestic price level or to the exchange rate (denominated in foreign currency). An important con- sideration in making these choices concerns their impacts on the government’s anti- inflationary credibility. A government that lacks the ability to precommit its future actions (or those of its successors) will face a time-inconsistency problem that could be aggravated by the issuance of long-term nominal debt, because debt issued in that form essentially increases the base of the inflation tax. Under these conditions, the government may therefore find it advantageous to issue debt in a form that is less susceptible to taxation through inflation— indexed debt, nominal debt with short maturities, or debt denominated in foreign currency. Because these types of debt provide creditors with more information and control (in the case of short-term debt) or protect them from the risks of inflation or devaluation (indexed debt and foreign-currency debt), it is likely to be cheaper than long-term nominal debt. Thus, the objective of enhancing credibility is likely to dovetail with that of reducing the government’s borrowing costs. (Indeed, as shown earlier, when nominal credibility problems are extreme, the implications for the cost of domestic- currency borrowing may even preclude such borrowing entirely, in practice leaving some developing economies no choice but to borrow in foreign currency or not at all.) However, the enhancement of credibility and reduction in borrowing costs associ- ated with eschewing long-term domestic-currency borrowing come at a price: that of increasing the government’s vulnerability to price level increases, real interest rate shocks, or exchange rate shocks. Moreover, when the government’s solvency is at issue, excessive reliance on short-term debt or foreign-currency debt may severely constrain the actions of the central bank. Thus, the question becomes how to optimize these tradeoffs. Peter J. Montiel 277 The answer, of course, requires a careful calculation of the benefits and costs associ- ated with each option under a country’s particular circumstances. For example, the credibility gains to the government from avoiding the issuance of long-term nominal debt may be significant only when the government otherwise lacks the ability to com- mit itself credibly not to inflate such debt away. Such anti-inflationary credibility will be lacking when the government actually retains the discretion to use the inflation tax (under a soft exchange rate peg or a float), when it lacks credibility on other grounds (as when it has not previously invested in a reputation for resisting incentives to act in a discretionary fashion), and when its revenue needs are high and conventional taxa- tion is highly distortionary. In combination with a large stock of nominal long-term debt these factors would make a high-inflation discretionary outcome likely. Thus, a government with these characteristics that wants to achieve a low-inflation outcome in the future would have a strong incentive not to issue long-term nominal debt.11 In other words the existence of long-term nominal debt is only one factor in the government’s decision to devalue or inflate. Creditors can rationally expect the gov- ernment to forgo the option to inflate away the real value of their assets if the gov- ernment is institutionally unable to do so (through participation in a currency board or currency union, for example), if it is perceived as placing a high value on the cred- ibility of its policy announcements, or if inflating creates few net benefits from the government’s perspective, because the conventional taxes that are avoided by using the inflation tax are not highly distortionary. Thus, a government can expect to achieve few credibility benefits from avoiding long-term nominal debt if it has previously created institutions that limit its inflation- ary discretion (for example, by creating an independent central bank that is explicitly committed to an inflation target), if it has established a reputation for nondiscretionary behavior, and if it has chosen levels of expenditure and has mobilized sources of taxa- tion that tend to minimize distortions. Because the additional credibility gains achiev- able by forgoing long-term nominal debt would tend to be small under these circumstances, avoiding vulnerability becomes relatively more important. Thus, opti- mal debt management would suggest heavy reliance on long-term nominal debt. If these conditions do not hold, then optimal debt composition shifts toward indexed, short-term, or foreign-currency debt. How should governments choose among these as credibility-enhancing devices? The arguments presented herein suggest that short-term debt may have important deficiencies because it may be vulnerable to the inflation tax, may give rise to multiple equilibria, and makes the government vulnerable to real interest rate and rollover risk. Thus, it is hard to make a strong case for short-term debt as a credibility-enhancing device when other means to enhance credibility are available. But short-term debt is preferred by creditors who face informational asymmetry and moral hazard prob- lems, so the best case for incorporating short-term debt into the government’s liabil- ity portfolio is to enable the domestic financial market to accumulate and 278 The World Bank Research Observer, vol. 20, no. 2 (Fall 2005) disseminate information about the risk-free interest rate. In this role, however, the stock of short-term debt should be small enough not to jeopardize the government’s solvency should interest rates spike and not to expose the government to excessive rollover risk. Solvency will obviously depend on the strength of the government’s finances, whereas rollover risk will depend on its ability to avoid bunching in short- term debt maturities, its capacity to repay the principal on short-term debt out of liq- uid assets or current revenues, and its access to quickly disbursing nonmarket sources of short-term finance in adequate amounts. These arguments suggest that the brunt of the credibility-enhancing burden (if one exists) should fall on long-term indexed and foreign-currency debt. Assuming them to be equally effective in enhancing credibility, the question becomes which one minimizes vulnerability to unexpected shocks. The currency composition of the government’s financial assets, contingent liabilities, and budget plays an important role. If the government has minimal foreign-currency assets, large contingent liabil- ities in foreign currency, and its revenues are not particularly sensitive to exchange rate changes, then incurring a substantial amount of foreign-currency debt would create a currency mismatch in the government’s comprehensive balance sheet that would leave it heavily exposed to exchange rate risk.12 Coupled with the likelihood that nominal exchange rates will fluctuate more than the average price level, partic- ularly in emerging economies maintaining a floating exchange rate, the possibility of such mismatches is likely to make the government’s net worth substantially more sensitive to changes in the exchange rate than to changes in the price level. Thus, the use of foreign currency–denominated debt instead of indexed debt as a credibility-enhancing device is likely to have relatively larger impacts on the govern- ment’s vulnerability to shocks. If so, then for governments that lack anti-inflationary credibility, long-term indexed debt would appear to dominate foreign-currency debt as a credibility-enhancing device. In short, the optimal composition of the government’s debt depends on a country’s circumstances. Although governments should generally finance themselves at market rates using a variety of securities, the optimal composition of those securities should be heavily weighted toward long-term nominal debt for governments that have anti- inflationary credibility and toward long-term indexed debt for those that do not. Notes Peter J. Montiel is Fred C. Green Third Century Professor of Political Economy at Williams College; his email address is pmontiel@williams.edu. 1. Prominent exceptions are a number of papers written several decades ago by Tobin (1971) and Brunner and Meltzer (1976). 2. Debt servicing can obviously be financed by reducing spending or raising tax revenues. Following the literature, this article takes the level of spending as given and focuses on taxation. As long as public Peter J. Montiel 279 spending is productive and subject to diminishing marginal returns, the welfare arguments in the text would be unchanged if the government responded to changes in its debt servicing needs by adjusting spending rather than revenues. 3. This is essentially the basis for the “tax smoothing� argument that underlies the neoclassical per- spective on optimal debt management (see Barro 1979). 4. The classic descriptions of financial repression are by McKinnon (1973) and Shaw (1973). For some empirical estimates of the implicit taxes that governments have collected through these means see, for example, Giovannini and De Melo (1993). 5. For recent reaffirmations of the empirical strength of the finance–growth link despite the econo- metric challenges in establishing it, see the surveys by Wachtel (2003) and Levine (2004). 6. It is easy to see that the importance of such shocks depends in part on the structure of the govern- ment’s budget as well as on that of the economy, because the structure of expenditures and revenues and the other macroeconomic effects of the shock will jointly determine the sensitivity of the govern- ment’s primary balance to shocks that give rise to changes in the domestic price level. 7. Not all shocks have this effect. For example, as Bohn (1988) points out, monetary shocks would tend to cause unanticipated changes in interest rates and the price level to be negatively correlated. Nominal debt in that case would tend to increase the variability of tax rates across states of nature, because opposite movements in interest rates and price levels would tend to reinforce their effects on the real value of debt service payments. 8. The role of long-maturity variable-interest debt (long-maturity debt with flexible interest rates indexed to market rates) is not considered here. Such debt exposes the government to interest rate risk but not rollover risk. In the case of variable-interest debt, exposure to interest rate risk depends on dura- tion (how frequently the interest rate on the debt is adjusted to market rates) rather than maturity. 9. As shown, however, such a debt management strategy would leave the government vulnerable to inflationary shocks that are associated with increases in the government’s financing needs. 10. In the limit, if the government’s nominal credibility problems or the degree of development of domestic financial markets is such that the government is unable to issue domestic-currency debt to either domestic or foreign creditors, “original sin� would remove the currency composition dimension of the debt management problem: if the government seeks to borrow at all, it must do so in foreign currency. 11. The international evidence suggests that governments are indeed reluctant to issue long-term nominal debt when they lack credibility, presumably reflecting the combined effects of a desire to enhance credibility and to reduce debt-servicing costs. For an application in the context of inflation sta- bilization, see Missale and others (1997). 12. For a discussion of the use of public debt management to hedge the risk associated with contingent public sector liabilities, see Becker (1999). References Alesina, Alberto, and Robert Barro. 2000. Currency Unions. NBER Working Paper 7927. Cambridge, Mass. Alesina, Alberto, Alessandro Prati, and Guido Tabellini. 1990. “Public Confidence and Debt Management: A Model and a Case Study of Italy.� In R. Dornbusch and M. Draghi, eds., Public Debt Management: Theory and History. Cambridge: Cambridge University Press. Barro, Robert J. 1979. “On the Determination of the Public Debt.� Journal of Political Economy 87(4):940–71. Barro, Robert J., and David B. Gordon. 1983. “Rules, Discretion, and Reputation in a Model of Mone- tary Policy.� Journal of Monetary Economics 12(1):101–21. Becker, Torbjorn. 1999. “Public Debt Management and Bailouts.� IMF Working Paper WP/99/103. Washington, D.C. 280 The World Bank Research Observer, vol. 20, no. 2 (Fall 2005) Blanchard, Olivier J. 2004. Fiscal Dominance and Inflation Targeting: Lessons from Brazil. NBER Working Paper 10389. Cambridge, Mass. Bohn, Henning. 1988. “Why Do We Have Nominal Government Debt?� Journal of Monetary Economics 21(1):127–40. Brunner, Karl, and Alan H. Meltzer. 1976. “An Aggregative Theory for a Closed Economy.� In J. L. Stein, ed., Monetarism. Amsterdam: North Holland. Burnside, Craig, Martin Eichenbaum, and Sergio Rebelo. 1998. “Prospective Deficits and the Asian Currency Crisis.� NBER Working Paper 6758. Cambridge, Mass. Calvo, Guillermo. 1988. “Servicing the Public Debt: The Role of Expectations.� American Economic Review 78(4):647–61. ———. 1989. “Is Inflation Effective for Liquidating Short-Term Nominal Debt?� IMF Staff Papers 36(4):950–60. ———. 1991. “The Perils of Sterilization.� IMF Staff Papers 38(4):921–26. Calvo, Guillermo, and Carmen Reinhart. 2002. “A Fear of Floating.� Quarterly Journal of Economics 117(2):379–408. Eichengreen, Barry, and Ricardo Hausmann. 1999. “Exchange Rates and Financial Fragility.� NBER Working Paper 7418. Cambridge, Mass. Giavazzi, Francesco, and Marco Pagano. 1990. “ Confidence Crises and Public Debt Management.� In R. Dornbusch and M. Draghi, eds., Public Debt Management: Theory and History. Cambridge: Cambridge University Press. Giovannini, Alberto, and Martha de Melo. 1993. “Government Revenue from Financial Repression.� American Economic Review 83(4):953–63. Levine, Ross. 1997. “Financial Development and Economic Growth: Views and Agenda.� Journal of Economic Literature 35(2):688–726. ———. 2004. “Finance and Growth: Theory, Evidence, and Mechanisms.� In P. Aghion and S. Durlauf, eds., Handbook of Economic Growth. Amsterdam: North-Holland. McKinnon, Ronald I. 1973. Money and Capital in Economic Development. Washington, D.C.: Brookings Institution. Missale, Alessandro, Francesco Giavazzi, and Piepaolo Benigno. 1997. “Managing the Public Debt in Fiscal Stabilizations: The Evidence.� NBER Working Paper 6311. Cambridge, Mass. Mussa, Michael. 2002. Argentina and the Fund: From Triumph to Tragedy. Washington, D.C.: Institute for International Economics. Razin, Assaf, and Efraim Sadka. 2004. “A Brazilian-Style Debt Crisis.� IMF Staff Papers 51(1):148–53. Sachs, Jeffrey, Aaron Tornell, and Andres Velasco. 1996. “The Collapse of the Mexican Peso: What Have We Learned?� Economic Policy 22(April):13–64. Serven, Luis, and Guillermo Perry. 2004. “Argentina’s Macroeconomic Collapse: Causes and Lessons.� World Bank, Washington, D.C. Shaw, Edward S. 1973. Financial Deepening in Economic Development. New York: Oxford University Press. Tobin, James. 1971. “An Essay on the Principles of debt Management.� In J. Tobin, ed., Essays in Eco- nomics, vol. 1, Macroeconomics. Chicago: Markham. Wachtel, Paul. 2003. “How Much Do We Really Know about Growth and Finance?� Federal Reserve Bank of Atlanta Economic Review (First Quarter):33–47. Peter J. Montiel 281