holding out for a haircut: financial crisis, moral hazard, and interest rate policy

International Journal of Finance and EconomicsInt. J. Fin. Econ. 5: 233–249 (2000)

Holding out for a Haircut: FinancialCrisis, Moral Hazard, and InterestRate Policy

Benjamin Dennisa,* and Simon Kandelba University of the Pacific, Stockton, CA, USAb Harvard Institute for International Development, Cambridge, MA, USA

The International Monetary Fund (IMF)-prescribed introduction of higher in-terest rates in crisis-hit economies has been criticized as being unnecessar-ily contractionary. This criticism ignores the effects of interest rate policy onthe incentives to restructure corporate debt once it has become strategicallyoptimal for domestic firms to declare a state of over-indebtedness. A wide-spread state of debt moratoria constitutes a ‘bad’ equilibrium, whereas thewidespread adoption of a quick debt workout strategy is a potentially ‘good’equilibrium. The latter can be encouraged through various policies, mostcontroversially a short-term policy of high interest rates. Other potential poli-cies include: (i) widespread debt forgiveness; (ii) enhanced implementation ofthe bankruptcy court; and (iii) the enforcement of free market competition.Copyright © 2000 John Wiley & Sons, Ltd.

KEY WORDS: financial markets; corporate debt restructuring; real interest rates; Asian cri-sis; Indonesia

SUMMARY

In times of financial crisis, macroeconomic policymakers in small open economies are faced withconflicting advice regarding the handling of thereal interest rate. One factor that has been over-looked in the interest rate debate is the effect ofinterest rate policy on corporate debt restructur-ing. With a sharply depreciated exchange rate,foreign-denominated debt service becomes muchmore expensive, increasing the incentive to de-fault. One corporation’s decision to default may

reinforce the desire of other corporations to dolikewise, because, with a higher fraction of corpo-rations defaulting, there is a lower probability thatany one of them will be singled out and punished.A form of moral hazard occurs when corporationsdefault regardless of ability to actually servicedebt. Under the equilibrium created by this formof moral hazard, since most corporations are indefault, they would be excluded (at least tempo-rarily) from formal credit markets. Aside from thecatastrophic effect on the domestic financial sector,this exclusion would place a severe drag on theability of the economy to recover. This paperdemonstrates that a policy of initially high realinterest rates is instrumental in reducing a corpo-rate default strategy that we label ‘holding out fora haircut’.

* Correspondence to: Department of Economics, University ofthe Pacific, 3601 Pacific Avenue, Stockton, CA 95211, USA.E-mail: [email protected]

JEL Code: E44, G30.

Copyright © 2000 John Wiley & Sons, Ltd.

B. Dennis and S. Kandel234

In order to clarify the reasons behind the appro-priateness of high real interest rates, two alterna-tive corporate strategies are considered. A firmmay opt to halt debt payments and ‘hold out for ahaircut’ (an HOH strategy) by demanding a large,if not complete, write-down of outstanding debt.Or, a firm may agree to negotiate in good faithwith creditors in order to bring about a ‘quick debtworkout’ (a QDW strategy), allowing it to returnto formal credit markets for fresh funds. Throughthe use of a simple monopolistically competitivemodel, we show that an increase in the real inter-est rate reduces the relative attractiveness of theHOH strategy by (a) allowing the exchange rate toappreciate thereby reducing the overall value ofthe ‘haircut’, (b) increasing the probability that anHOH firm will go bankrupt, and (c) decreasing theprobability that a firm will receive a ‘haircut’.Thus, the interest rate should be increased to thepoint at which the QDW strategy becomes domi-nant. This period of high interest rates is onlytemporary, and depends primarily on the speedwith which the QDW strategy can be adopted. Inaddition to high initial interest rates, other policiesthat might promote the QDW strategy are dis-cussed, including widespread debt forgiveness, theenhanced implementation of a bankruptcy court,and enforcement of free market competition.

Although widely applicable, the model is ap-plied to Indonesia’s recent experience with finan-cial crisis. Despite an exchange rate that had at onepoint plummeted by 525% from its pre-crisis peakcombined with a moribund commercial bankingsystem, corporations continue to function and,amazingly, seem reluctant to engage in debt re-structuring negotiations. This impasse is the singlemost important problem faced in Indonesia andother crisis-struck Asian nations and it directlyhampers efforts to reform financial systems andrestore the credit flows needed for economic re-covery. Therefore, we have outlined an interestrate policy that places the impasse on corporatedebt restructuring at the center.

INTRODUCTION

The economic crisis in South-east Asia has trig-gered a renewed discussion on whether raisinginterest rates is the appropriate policy response

when facing a crisis-induced sharp increase in thecountry risk premium. Numerous economists haveargued forcefully that a policy of low interest rateswould be more appropriate. From an expansionistperspective, lower interest rates, it is argued,would reduce the strain on the financial sector,and the resulting weak exchange rate would pavethe way for an export-led recovery of the econ-omy. However, those who argue in favor of acombination of lower interest rates and a weakerexchange rate ignore a crucial form of moral haz-ard that is currently characteristic of several of thehard hit South-east Asian economies.

With drastically deteriorated balance sheets andan inability to service foreign exchange-denomi-nated debt obligations in particular, many firmshave begun to opt for a ‘holding out for a haircut ’(HOH) strategy. This strategy simply involveswaiting to workout debt until all debt is forgiven,or at least substantially written down. As long asbanks refuse to give new loans to firms that arefollowing a HOH strategy, lowering the interestrates will not lead to increased economic activityunless lower rates simultaneously encourage firmsto settle their debts. In our view, this is unlikely.We argue that attempts to assess the appropriate-ness of interest rate levels must take into accountthe interaction between the interest rate and thedebt strategy choices made by domestic firms. Ourconclusion is that a policy choice of short-termhigh interest rates, combined with an inevitablyappreciated exchange rate, will have a significantimpact on decreasing the number of firms follow-ing a haircut strategy.

To demonstrate this point, a simple monopolis-tic competition framework is developed that leadsto two possible equilibria: one in which all firmsadhere to the HOH strategy and one in whichfirms engage in a ‘quick debt workout’ (QDW) strat-egy. Choosing parameters based on Indonesianeconomic data, simulation of our model demon-strates that the value of each strategy is indeedstrongly affected by the prevailing level of thedomestic real interest rate. We also find that thevalue of the QDW strategy increases with: (i) anincrease in the number of firms following theQDW strategy; (ii) a higher share of dollar-denom-inated debt out of total debt; (iii) a larger financingpremium faced by HOH strategy firms; (iv) lessmarket power (and differentiation) on the part of

Copyright © 2000 John Wiley & Sons, Ltd. Int. J. Fin. Econ. 5: 233–249 (2000)

Financial Crisis, Moral Hazard, and Interest Rate Policy 235

individual firms; (v) a decrease in the country riskpremium; vi) an increase in the probability ofsuccessful bankruptcy cases decided in favor ofthe plaintiff; and (vii) a decrease in the percentageof original debt that a firm must pay off after adebt workout settlement has been achieved.

The reaction of the exchange rate to a change inthe interest rate is also crucial to our model. Thistopic is currently being hotly debated. Someeconomists including Furman and Stiglitz (1998),have argued that lower interest rates willstrengthen the exchange rate as capital inflowsrespond to a perception of greater economicprospects for growth. We take the more traditionalview that a lower interest rate will reduce thecompensation that the investor receives for takingon the risk of keeping capital in the crisis-riddencountry. As a result, the ensuing capital flight willcause the exchange rate to depreciate.

The strategic aspects of debt resolution havebeen dealt with in numerous papers. The literatureon the Latin American debt crisis of the 1980s isimmense, including a large number of studies onthe debt overhang and the conditions for default-ing.1 Some papers, such as Andrabi and di Meana(1994), concentrated on the role that banking regu-lations play in generating a debt crisis. Others,such as Krugman (1988) and Fernandez-Ruiz(1996), evaluated the appropriateness of debt for-giveness in the face of other options. Among thepapers that have addressed corporate debt specifi-cally (e.g., Helpman, 1988; Lamont, 1995), the fo-cus has been on investment and the appropriationof investment returns by creditors.

It is important to note that there is a significantdifference in the effect of macroeconomic changesbetween a strategic game modeled on creditorsversus one modeled on debtors. In the creditor-focused models, an improvement in the macroeco-nomic environment that enables debtors to resumethe servicing of their debt would solve the crisis.However, an improvement in the macroeconomicenvironment would not automatically solve thedebt crisis if the focus was on a strategic gameplayed by debtors. Debtors may find it optimal tobe perceived as insolvent, even when improvedmacroeconomic conditions have in fact renderedthem solvent.

Among the papers that have concentrated onstrategic behavior on the part of creditors, Dooley

(1994) and Cline (1995) discussed the value tocreditor banks of holding out for full repayment.Sachs (1995) argued that a ‘disorderly workout’occurs as creditors prematurely destroy illiquidcompanies by shutting off credit lines and callingin loans, even when the firm is solvent. This themewas followed up in Radelet and Sachs (1998), whoargued that creditor herd behavior as well as do-mestic financial panic are the chief culprits in theAsian crisis. Again, creditors are the primary ac-tors in the strategic game of apportioning lossesfrom the crisis.

In the following section, we develop a simplemodel of the choice between the HOH and theQDW strategies. The ‘Sensitivity of v to ParameterChanges’ section shows how this choice is affectedby various parameters. The ‘Empirical Evidencefrom Indonesia’ section describes evidence in fa-vor of this theory, while the final sectionconcludes.

A SIMPLE MODEL

We begin by describing the macroeconomic shock,or displacement, highlighting the implications foran adjustment in both interest rates and the ex-change rate. We then model the two corporatedebt strategies and examine the relative value ofeach. Finally, we explore the multiple equilibriaimplied by the model.

The Macroeconomic Shock

The macroeconomic shock is modeled as a discreteincrease in the risk premium, l, on investmentswithin the country. We can use the interest rateparity condition to see that this shock will result insome combination of both exchange rate deprecia-tion and higher interest rates. The interest rateparity condition is given by

rd−r*= (eF−e)/e+l, (1)

where rd is the domestic interest rate, r* is thefixed world interest rate, eF is the fundamental andunchanged future exchange rate, e is the currentexchange rate, and l is the country risk premium.If r and e are free to react to a change in l, clearlyinterest rates will rise and the exchange rate willdepreciate. But should the authorities seek to bias



most of the adjustment towards the real exchangerate, or should the real interest rate take the bruntof the adjustment? Because the real interest rate isseen as the policy instrument, the choice of a highreal interest rate will correspond to a relativelyappreciated exchange rate and the choice of a lowreal interest rate will lead to greater real exchangerate depreciation. The endogeneity of the real ex-change rate, i.e., that it is a negative function of thedomestic real interest rate can, be expressed as

e=e(r̄d). (2)

Utility and Profit Maximization

In order to determine the response of the two debtstrategies to changes in macroeconomic policy, wespecify the nature of consumption and productionin a small open economy producing under monop-olistic competition. Krugman (1991), Dixon (1994),and Matsuyama (1995) offer alternatives for thistype of modeling.

Production is modeled as monopolistically com-petitive with each firm producing differentiatedgoods subject to an initial fixed cost, and a con-stant marginal cost. Capital is assumed to be theonly factor in production. The production functionfor output is given by

x=1b

(k− s̄), (3)

where s̄ is a fixed cost and b is the marginal cost.Consumers, both foreign and domestic, substitutebetween these goods based on a constant priceelasticity of substitution (CES), s. Because we as-sume that the economy is small and open, and thatall goods are tradable, demand will not be influ-enced by domestic income. The initial value oftotal capital employed, k, is equal to the initialvaluation of debt, d(e).

The representative consumer’s utility function isgiven by

U=� %

n

j=1c j

rn1/r

, (4)

where U is utility, cj is the consumption of good j,r is equal to (s−1)/s, where s is the elasticity ofsubstitution, and n is the number of variety ofgoods (equal to the number of firms, N).

We assume that the total expenditure on goodsby foreign and domestic consumers, E, is roughlyconstant due to the negligible impact of domesticrecession on total demand. This implies that utilitymaximization is subject to the following budgetconstraint:

E=epici+e %N−1

k=1, k" ipkck. (5)

Assuming that the representative consumer’sconsumption of any distinct variety of a good is aconstant fraction, n, of that good’s output, con-sumer maximization implies the following inversedemand function:

epj=k(CM)(1−r)/rnr−1c jr−1,

where k is a constant, and pj represents the priceof the good.

Using the production function (3), profit maxi-mization by each firm yields the familiar pricemark up over marginal costs. Marginal costs willdiffer for the two types of firms and HOH firmswill face higher costs. There is a constraint on theavailability of working capital for firms engagingin the HOH strategy, although we assume thatthese firms are still able to raise funds through theuse of internally generated funds or offshore ac-counts. However, the use of these funds entails asignificant internal finance premium, o. This pre-mium is not borne by firms following the QDWstrategy. Thus, each firm faces prices:

epQ=rd

b

r

epH= (rd+o)b

r. (6)

From the inverse demand function, the relativequantities demanded of any two goods can becalculated:

xH

xQ

=�rd+o

rd

�−1/(1−r)

. (7)

We now solve for the output of a representativeHOH firm under free entry. We assume that be-cause all firms initially begin with a default HOHstrategy, operating profits are more likely to be bidaway from HOH firms. A zero-profit condition isemployed to determine the output of a representa-tive HOH firm:



pH=epHxH−rk= (rd+o)b

rxH− (rd+o)k=0.

This implies that production of any given HOHfirm will be given by:

xH=s̄b

� r

1−r

�. (8)

Thus, we obtain the familiar result that thequantity of output per firm is constant for HOHfirms. Any reduction in the market for HOH firms’products will be reflected in a reduction of thenumber of surviving HOH firms, not in the outputper firm. On the other hand, a firm opting for theQDW strategy is not allowed to either fail or ceaseproduction. Combining the above equations withEquation (7), we can then determine the demandfor QDW firm output:

xQ=�rd+o

rd

�1/(1−r)

x̄H. (9)

The Comparative Value of Competing DebtStrategies

In order to explore this issue in the context offirms’ choice of debt repayment strategy, we ex-amine the value of each of two strategies open toeach firm under the assumption that the shockrenders all firms illiquid in the short-run. Thereare N firms in the economy, all of them identical interms of debt, technology, and structure. All firmsproduce a differentiated variety of a basic good.We assume that some initially adopt an indefinitedebt moratorium.2 Given that this is tantamount toadoption of the HOH strategy, we seek to deter-mine the conditions under which a firm will aban-don the HOH strategy and undertake debtrestructuring. The position of the creditor is con-sidered to be fixed with respect to the strategicdecisions of the firms.

The value of following the HOH strategy isgiven by the sum of the present discounted value(PDV) of the profit stream plus the value of theone-time capital gain experienced if a 100% haircutis received. We assume that the initial level of debtis equal to the capital input used by the firm, suchthat the value function given below in fact repre-sents a quasi-profit function. However, there is apossibility that the firm will be brought before thebankruptcy court and be forced to shut down. The

owners of the firm, therefore, face the probabilityof zero income from sales revenue if they chooseto follow the HOH strategy. We assume that theprobability of avoiding being brought before thebankruptcy court is positively related to the num-ber of firms following the HOH strategy. If manyfirms are following a HOH strategy, the chances ofany one of them being selected is much lower thanif very few firms were following the strategy.

Thus, the value of the HOH strategy can berepresented by

VH=1rd

f(N)epHxH−(rd+o)

rd

k

+r*rd

[u(N)d(e)− (1−u(N))d(e)]

=1rd

f(N)epHxH−(rd+o)

rd

k+r*rd

(2u(N)−1)d(e),

(10)where f represents the probability of avoidingbankruptcy court, u represents the probability thata full haircut (or complete debt forgiveness) willbe given (i.e., that the HOH strategy will be suc-cessful), and d(e) represents the firm’s debt ex-pressed in local currency terms. Note that u is alsoa positive function of the number of firms follow-ing the haircut strategy. Furthermore, 1/rd is thediscount factor that applies to a stream of futurepayments. This factor is discussed more thor-oughly in Appendix A.

Similarly, the value of following a QDW strat-egy is equal to the PDV of profits minus theamount of debt the firm is required to repay in thedebt workout agreement (which, because it isequal to the firm’s capital, represents the firm’soperating costs). In this case, the firm’s owners donot need to worry about bankruptcy court. Fur-thermore, the repayment requirement need not be100% of the original debt. Thus, the value of theQDW strategy can be represented by

VQ=1rd

epQxQ−rd

rd

k= −r*rd

gd(e), (11)

where the subscript Q denotes variables for firmsfollowing the QDW strategy, and g represents thepercentage of the original debt that firms mustrepay after the debt workout has been settled.

Each firm’s debt can be divided into domesticand foreign debt. For the representative firm, weexpress this as



d(e)=edF+dD, (12)

where edF is the local currency value of the foreigndebt and dD is the value of the domestic debt.Notice that, for small changes in the exchange rate,the elasticity of the debt with respect to the ex-change rate is equal to the share of foreign debtout of total debt:

(d(e)/d(e)(e/e

=dF

ed=

edF

d(e)= f,

where f is the share of foreign debt out of totaldebt.

Multiple Equilibria

The determination of equilibria will proceed intwo parts. First, we specify how the number offirms following a haircut strategy affects the prob-abilities of not being declared bankrupt and re-ceiving a full haircut. Second, we employ theendogeneity of these probabilities in a value func-tion that can be used to determine the number offirms following each type of strategy and howmacroeconomic shocks move the economy fromone equilibria to another.

The Probability of Bankruptcy and the Probabilityof a Full Haircut for HOH Firms

Recall that in Equation (10) we stated that theprobability of being declared bankrupt was nega-tively related to the number of HOH firms, andthat the probability of receiving a full haircut waspositively related to the number of HOH firms. Inorder to determine how these probabilities will beaffected by different variables, we must determinehow the number of HOH firms will be affected.We can use the budget constraint to solve for thisquantity. Rewriting the budget constraint

E= (N−Q)epHxH+Qepqxq, (13)

where Q is the number of firms following theQDW strategy, the total number of firms, N, cannow be expressed as a function of the share oftotal firms following the QDW strategy:

N=m

1−d(1− th−1), (14)

where t equals (rd+o)/rd, h equals 1/(1−r), d

equals Q/N (i.e., the share of QDW firms out of allfirms), and m is the total expenditure on domesticgoods divided by the average expenditure on aHOH firm’s output (a measure of the relativeconcentration of firm size or oligopoly power). Theparameter m will be given as

m̄=E

epHxH

.

It simply remains to specify f and u as functionsof the number of firms following the HOH strat-egy. This will be equal to N−Q, which is equal toN(1−d).

u=u(N(1−d))+

f=f(N(1−d))+

.

The Relative Value Function

Now define a variable v that is equal to the valueof the QDW strategy minus the value of the HOHstrategy:

v=VQ−VH=1rd

(1− th−1− t−1−r*gw−f

−r*(2u−1)w), (15)

where w is the ratio of the value of per firm debtto the value of output of the typical HOH firm:

w=d(e)

epHxH

.

For v\0, it becomes attractive for firms to defectfrom the HOH strategy to the QDW strategy.Applying Equation (15) allows us to generate Fig-ure 1, which shows how v changes for changes ind. Note that equilibria only occur when v=0. Ifv\0, then d will be increasing. On the other hand,if vB0, then d will be decreasing.

Interest Rate and Strategic Choice Dynamics

We now specify dynamics in terms of the funda-mental real interest rate that should prevail in themedium- to long-term (rdF), and the relativeamount of firms opting for the QDW over theHOH strategy (d).

Financial Sector Equilibrium. Setting the exchangerate equal to its fundamental value (e=eF) allowsus to use Equation (1) to determine the fundamen-tal value for rd :



rdF=r*+l.

The dynamics of the interest rate will be given by

r; d= −A(rd−rdF) dt. (16)

If rd\rdF, the interest rate will decrease overtime to restore equilibrium, and vice versa. Wewill not specify the exact mechanism that governsthe speed of adjustment of rd, A, as that is beyondthe scope of the model. However, as long as theinterest rate adjusts less rapidly than the exchangerate, our analysis will still hold. This conditionprovides us with a locus in the (d, rd) space forr; =0, which we will denote FF.

Strategic Choice Equilibrium. We will also denote alocus SS that represents points at which firms willnot change strategy (d: =0), i.e., points at whichv=0. As discussed above, d: is positive whenv\0, and negative if vB0:

d: =B(v) dt. (17)

Figure 2 shows these two loci for a particularcombination of parameters and the implied dy-namics. Note that there is a continuum of potentialand unstable equilibria for values of d \0 butB1. However, the only stable equilibria are ford=0 or d=1. For values of d above a certain

threshold, all domestic firms will invariablychoose to follow the QDW strategy. However, forvalues of d below this threshold and within acertain range of values for rd, all domestic firmswill eventually select the HOH strategy. Thus, themodel implies that not only is the choice of theinitial interest rate essential in determining whichstrategy will prevail, but that the choice of inter-mediate values of rd may be detrimental. Eitherfairly high or fairly low values of rd are required.

SENSITIVITY OF v TO PARAMETERCHANGES

Because v ranges both above and below zero, it isimportant to explore the conditions under whichthe QDW strategy will dominate. Again, our par-ticular interest is in whether it is more appropriateto support high or low real interest rates in light ofthe relative value of the alternative debt strategies.However, we also highlight three other policyalternatives: (i) widespread debt forgiveness; (ii)enforcement of legislature to enhance market com-petition; and (iii) the strengthening and enhancingimplementation of the bankruptcy court.

Differentiating v with respect to rd yields3

Figure 1. The effect of changes in the number of firms following the QDW strategy on v.



Figure 2. Interest rate and strategic choice dynamics.

(v(rd

= −o[(h−1)th+1]

rd2t2

−w

(rd+o)(rd

−1fe erd−1)r*[g+2u−1]

−eN

rd

rd(1−d)[fef

N(1−d)+2r*wueuN(1−d)],

where efN(1−d)\0, eu

N(1−d)\0, eNrdX

?0 and e e

rdB0.The first term represents the elasticity of the

probability of escaping bankruptcy with respect tothe number of HOH firms. The second is theelasticity of the probability of receiving a haircutwith respect to the number of HOH firms. Thethird term is the elasticity of the number of allfirms with respect to the domestic interest rate,and the final term is the elasticity of the exchangerate with respect to the domestic interest rate. Thederivative shown above can be either positive ornegative depending on the condition shown inAppendix B.

To understand the effect of changing the interestrate, notice that as rd increases, the value of eachstrategy is altered. For the QDW strategy, higherinterest rates remove some of the proportional costbenefits of quickly resolving debt, th (assumingthat the internal finance premium remains con-stant), lowering the value of the QDW strategy.

However, higher interest rates also lead to a moreappreciated exchange rate, causing the local cur-rency value of foreign debt to fall. This increasesthe value of the QDW strategy, potentially coun-teracting the first effect.

Changes in the interest rate also have an effecton the probabilities of receiving a full haircut andescaping bankruptcy proceedings for firms follow-ing the HOH strategy. After interest rates haveincreased beyond a threshold level, higher interestrates decrease the number of firms N, causing bothprobabilities to fall and decreasing the value of theHOH strategy.4 Thus, the ambiguity of the elastic-ity of N with respect to rd is resolved as beingnegative in the range of parameters in which weare interested. The initial effect of the decline in Non the probabilities is not very large due to ourassumptions about the concavity of the functionalforms of f(N) and u(N).5 However, the higher therd value becomes, the more profound its impactwill be on lessening the value of the HOHstrategy.

Thus, the ambiguity of the sign on this deriva-tive is based on the result that, for small values ofrd, increases in rd decrease the value of the QDWstrategy faster than they decrease the value of theHOH strategy. In other words, the decline in the



cost advantages of the QDW strategy (which isinitially quite steep) more than offsets the deflationof the value of the HOH strategy. However, atsome point, the concavity of the probability func-tions (steeper ef

N, euN), combined with the relative

stabilization of the cost advantage function(smaller t) and the steady deflation of the localcurrency value of the foreign debt (if f is largeenough) all combine to result in a positive sign onthe derivative. This is illustrated by the changingslope of the v(rd) function in Figure 3.

This result suggests that a policy of high initialinterest rates would succeed in pushing firms to-wards a QDW strategy.

Next, we turn to the effect of changes in themarket power of firms on the value of v. Marketpower is inversely related to the elasticity of sub-stitution or s.6 For very small values of r (lowvalues of s), market power is high. However, as r

approaches one (i.e., s rises), the economy ap-proaches perfect competition. Increases in marketpower tend to decrease the value of the QDWstrategy relative to the HOH strategy. Marketpower greatly reduces the importance of the costadvantage to QDW firms (in other words, thevalue of not having to bear the internal finance

premium), enabling more of the HOH firms tosurvive for a given level of d. At the same time,the increased number of HOH firms increases theprobability of the HOH firms’ survival and thatthey will receive a full haircut. Thus, v falls as s

increases. See Appendix D for a formal discussionon the effect of market power on v.

We can conclude that policies designed to in-crease inter-firm competition and reduce market powerwould support the QDW strategy choice (Figure4).

Other parameters affect v unambiguously. Thederivative of v with respect to the percentage ofdebt that has to be repaid under the QDW strategy(g) is negative:

(v(g

=−r*w

rd

B0.

We can model a policy of widespread debt forgive-ness as a reduction in g. Thus, for sufficiently highvalues of g, the QDW strategy will alwaysdominate.

Now suppose we model the probability ofavoiding bankruptcy as having an autonomouscomponent:

f=f( + (N(1−d)).

Figure 3. The effect of interest rate changes on the value of v.



Figure 4. The effect of changes in firm’s market power on the value of v.

As expected, a decrease in the autonomous com-ponent of the probability of avoiding bankruptcywill unambiguously increase the relative prof-itability of the HOH strategy:

(v(f( =

−1rd

.

Thus, improving the implementation of bankruptcyproceedings should increase the incentives for se-lecting the QDW strategy.

Other derivatives are given in Appendix E,where we show that the derivatives of v withrespect to the internal finance premium o and theshare of firms following the QDW strategy d areboth positive, while the derivative of v with re-spect to the country risk premium l is negative, ascould have been surmised from our previousdiscussion.

EMPIRICAL EVIDENCE FROM INDONESIA

How well does the model fit the facts? In order toshed light on the applicability of the model and todiscuss sensible parameter values for it, we exam-ine the recent experience of Indonesia. Of themany Asian countries that suffered contagion ef-

fects of the regionwide currency crisis, it wasIndonesia that faced the most devastating collapsein the value of its currency. At its nadir, the rupiahhad plummeted by almost 525% from its pre-crisislevel. The resulting shock to corporations withlarge dollar-denominated debts was immense. In-deed, it was partly due to the intense clamoringfor foreign exchange by these corporations to meettheir short-term dollar obligations that the ex-change rate depreciated so sharply. From an aver-age growth rate of 7.5% per annum in the early1990s, the economy has been estimated as con-tracting by 15% or more during 1998.

It is widely felt that the Indonesian economyfaces two major interlocking economic problems(in addition to a complex political situation). First,the banking sector must be restored to health inorder to restart broad-based financial intermedia-tion and, second, corporate debt must be restruc-tured. It is often a working assumption thatthese two problems must be resolved simulta-neously or, at least, few imagine that corporaterestructuring can proceed without banking sectorrecapitalization.

Nonetheless, contrary to the assumption of abreakdown in the economy inherent in this theory,corporations continue to function even without the



resolution of the banking system’s negative networth. This is attributed to two factors: (i) theavailability of sufficient cash reserves (untouchableby creditors and sometimes held offshore); and (ii)the ability to meet working capital demands withcurrent revenues once debt moratoria are put inplace. In many ways, the resolution of corporatedebt should be seen as more important than bank-ing sector recapitalization. A revitalized corporatesector could resume repayment of some of itsoutstanding debt to the domestic banking sectorand, moreover, it will be difficult to attract theprivate capital inflows needed to restore growthwithout the resolution of the corporate debt.7

Indonesian official data show that the country’stotal private sector debt had reached US$68.3 bil-lion as of December 31 1997. Of this total, non-bank corporations were responsible for US$58.9billion of the debt. Approximately US$53.6 billionof this non-bank debt was bank credit of whichprivate national firms owed US$30.1 billion. An-other US$23.5 billion in bank credit was owed byforeign firms and foreign joint ventures, while theremaining US$5.3 billion in non-bank private debtwas composed of commercial paper, medium-termnotes and floating rate notes.8 Taken together withIndonesian public sector debt, then, Indonesia’stotal external debt adds up to US$136.09 billion or63% of 1997 nominal gross domestic product(GDP) (which was roughly US$214.99 billion mea-sured using the average 1997 nominal exchangerate).

The ability of corporations to continue opera-tions, albeit at an inefficient and diminished level,is partly the result of the relatively inexperiencedbankruptcy court as well as the result of reputedlylarge hidden stocks of liquid assets. As of mid-December 1998, only three corporations had beendeclared bankrupt; two of these cases are underappeal and one (the only case in which a listedcompany was declared bankrupt) was reversed onappeal.9 Many other bankruptcy cases have beendismissed out-of-hand.

Although at the time of writing recent progresshas been announced on debt restructuring, to dateall such deals amount to between US$1 and US$2billion. Some attempts have been made to furthercorporate debt restructuring in Indonesia, mostnotably the Jakarta Initiative, which was partlysponsored by the World Bank. While there are

indications that this initiative may succeed, it hadyet to produce its first concrete agreement at thetime of writing.10

It is important to note, however, that there arealso important stumbling blocks on the side ofcreditors. The most significant of these is the in-ability of Japanese banks to write off largeamounts of debt.11 Japanese banks are reported tohold 38% of the US$58.4 billion of foreign corpo-rate debt in Indonesia.12 However, many Japanesebanks are currently on the edge of bankruptcy dueto devastating losses abroad and to a deep reces-sion in the country. As a result, they are reluctantto write down losses on their balance sheets. Whilethis issue has been receiving some attention as amajor stumbling block to debt restructuring, thereare also reports that Japanese banks are selling offloans extended in Asian economies at a discount.13

With ownership of debt changing hands fromJapanese banks to other creditors, the Japanese bankimpasse may decline in relevance.

We can use data on Indonesia to proxy for someof the parameter values used in our simulations.Beginning with the probability of not being suc-cessfully hauled into bankruptcy court, f, it isimportant to note that to date there has not beenone ruling in favor of bankruptcy that has sur-vived the entire appeals process. However, therehave been rulings in favor of bankruptcy that havebeen overturned on appeal. Although f representsa variable in our model (dependent on the share offirms following a HOH strategy), in practice it isnot likely to be significantly different from one.

For the variable w, i.e., the ratio of the localcurrency value of total debt to the average salesrevenue of the HOH firm, we employ net debt-to-equity (NDE) estimates for the top 20 listed firms.With efficient markets, equity should reflect thePDV of the stream of future profits. Assuming thatprofits are proportionally related to revenues onaverage, a discount rate can be used to link rev-enue to equity.

The range for the elasticity of substitution (s)used in these simulations is 2–2.5, which fallsslightly below the value of s used in other monop-olistic competition papers.14 However, lower val-ues of s are associated with higher levels ofmarket power per firm, and Indonesia is mostlikely less competitive than the US and the Eu-ropean nations that have been the subject of other



studies. Although arguing that Indonesia is not asoligopolistic as it was in the past, Bird (1996)calculated that the average concentration ratio ofthe top four firms in manufacturing sectors in 1993was roughly 61%. Furthermore, a full 30% of totalmanufacturing industries had a four-firm concen-tration ratio of above 75%. Given that these indica-tors are roughly in line with higher levels ofmarket power, our estimates of s appear to beappropriate.

The elasticity of the exchange rate with respectto the interest rate is another important aspect ofour model. This parameter is difficult to estimatefor several reasons. First, prior to the crisis, In-donesia was able to maintain an exchange ratetarget while simultaneously managing to exercisean independent monetary policy. This has pro-vided a paradox for those who maintain that it isimpossible to have an open capital account, a fixedexchange rate target, and set interest rates inde-pendently (the ‘impossible trinity’). However, forreasons discussed in Bond (1996), Indonesia wasable to do just this. Thus, the relationship betweenthe exchange rate and the domestic interest rateshould be non-existent over this period. Second,once the crisis began, sharp movements in the

country risk premium dominated movements inthe exchange rate. Estimation of this country riskpremium is difficult and is beyond the scope ofthis paper. Nonetheless, few observers would dis-agree that the high interest rates prevailing herehave dampened capital outflow and have helpedto stabilize the exchange rate, regardless of theirviews towards the appropriateness of high interestrate policies. For this reason, we feel justified inusing a fairly high elasticity of the exchange ratewith respect to the interest rate (Table 1).

CONCLUSION

This paper argues that, among other policies, ini-tially high interest rates are necessary to addressthe impasse on corporate debt restructuring. Thisimpasse is the single most important problem cur-rently facing the crisis-hit Asian nations and di-rectly hampers efforts to reform financial systemsand restore capital inflows. However, in order tosimplify the analysis, several interesting aspects ofthe current financial crisis were ignored.

One interesting extension of this model may beto examine the effects of a high interest rate

Table 1. Parameterization for Indonesia

Value (%)ParameterItem

Number of bankruptcy cases resulting in liquidation relative to the number of cases 1−f 0concluded

NDE ratio of top 20 listed non-financial firmsa w 55Foreign funding real interest rate prevailing just prior to crisisb r* 5

rdPost-crisis domestic interest ratec 55Average concentration ratio in manufacturing industryd s 61

30Percentage of industries with a four-firm concentration ratio above 75%a, b so, l 14Country risk premiume

1−g 40Discount at which the Widjaja family was able to repurchase outstanding bondsbelonging to Asian Pulp and Paper (APP)

fShare of external debt out of total debt 55f 72Share of external debt out of total debt for the top 20 non-financial institution listed

companies

a Prasetijo (1998).b A nominal interest rate of 15% prevailed on a successful bond issue by PT Astra International (the dominant vehicle producer inIndonesia) just prior to the crisis. Subtracting a value for expected inflation of 10% (the government’s target) leaves us with a realinterest rate of 5%.c Bank Indonesia was required to pay a nominal interest rate of 71% on its open market instrument, the SBI, in September 1998. Atthis time, an assumption of 20% for future annualized inflation would not have been unreasonable. Most observers attributed thehigh real interest rate prevailing to the excessive risk inherent in politics at that time. We use the SBI rate because, apart from theguarantee of exporters’ letters of credit (LCs) and some export credit lending, domestic lending has effectively ceased in Indonesia.d Bird (1996).e The Jakarta Bulletin, Credit Lyonnaise CLSA Global Emerging Division, 15 September 1998.



regime on the financial sector. Many have arguedthat the apparent weakness of financial intermedi-aries in the crisis-hit countries constitutes a con-straint for policy-makers. A substantial increase inthe real interest rate, it is argued, would causeextreme damage to the banking system. However,the fate of the banking sector is closely tied to thatof the corporate sector. Thus, if high interest ratesencourage quick debt workouts, banks can onlygain. Furthermore, many banks also have largeforeign debt exposure that is inflated (in localcurrency terms) by a weak exchange rate.15 Thus, amore appreciated exchange rate could benefitbanks directly as well.

In a recent study by Claessens et al. (1998) on theimpact of currency and interest rate shocks be-tween early 1997 and September 1998, the massiveexchange rate depreciations where considered suf-ficient to drive almost two-thirds of Indonesianfirms into insolvency, and 72% into illiquidity. Bycontrast, the interest rate shock was found to bemuch less damaging, pushing only 2–5% of In-donesian firms into insolvency, and 15–25% intoilliquidity. Of course, interest rates and exchangerates are inextricably linked. Nonetheless, the factthat high interest rates were found to have amilder effect than an over-depreciated currency issuggestive.

Another useful extension might focus on poli-cies that are designed to stimulate the economyout of a recession. Another useful extension mightfocus on policies that are designed to stimulate theeconomy out of a recession. Krugman (1999) ar-gues forcefully that efforts to stimulate aggregatedemand are an essential policy response to exter-nal shocks. The model presented here ignoresemployment and considers the amount ofexpenditure on domestic goods, E, to be fixed.While it would be difficult to integrate employ-ment into the model, it may be possible to endog-enize E by tying it to a competitive exchange ratemodel. In such a model, E would represent domes-tic demand for goods and services (instead ofworld demand), and the balance of paymentscould be modeled explicitly. Furthermore, itwould prove a valuable extension if the effects ofchanges in the interest rate on national incomecould be incorporated into the relative value of thetwo strategies. However, it should be rememberedthat we are not advocating a policy of permanently

higher interest rates. Rather, we feel that with highinitial interest rates, the swift resolution of corpo-rate debt may paradoxically allow for a fasterreturn to actual lending at lower interest rates, asopposed to having low interest rates but no effec-tive lending taking place.

We should also state that a model that allowsfor strategic behavior by both creditors anddebtors would be preferable. Our model is open tothe criticism that it lays the blame for the currentdebt impasse primarily on the debtors. It shouldbe noted that we do argue that creditors shouldprovide some debt forgiveness, represented by g,to companies that are willing to engage in a QDWstrategy. Indeed, the higher the level of debt for-giveness by creditors for QDW firms, the moreattractive the QDW strategy becomes. However, asis stated above, there is evidence that some credi-tors, mainly Japanese banks, are refusing to negoti-ate debt workouts due to their own fragilefinancial condition. An improved model might beable to incorporate strategic behavior by creditorsas well, perhaps through endogenizing g as theoutcome of some strategic process used by credi-tors, although that is beyond the scope of thispaper. However, as stated in our literature review,we feel that this paper is an important first step inexamining an aspect of strategic behavior bydebtors that has been mainly overlooked.

Finally, there is the empirical issue of the rela-tionship between the interest rate and the ex-change rate. Furman and Stiglitz (1998) arguedthat this relationship could be positive in times ofcrisis (i.e., low interest rates could lead to a low—appreciated—exchange rate and vice versa). Thishypothesis is based on the assumption that bankswould be willing to lend at the current interestrate, whatever it is. However, the moral hazard ofcorporate non-payment of debt may prevent banksfrom lending, even at low interest rates, making thequestion of the level of interest rates less important fordomestic lending than for short-term money marketcapital flows. The adoption of a high domestic realinterest rate that stabilizes short-term capital flowsand encourages the rapid workout of corporatedebt, therefore, represents a starkly different pol-icy recommendation than might be drawn fromFurman and Stiglitz’s argument. Unfortunately, asFurman and Stiglitz also acknowledged, given thelarge structural shifts in parameters that have



likely taken place, econometric work on the linkbetween interest rates and exchange rates in timesof crisis remains inconclusive, and the debateabout post-crisis interest rate policy is likely tocontinue for some time.16

ACKNOWLEDGEMENTS

The authors wish to thank Joseph J. Stern, NancyAllen, L. Peter Rosner, Susan Fiorella, Susan Baker,and the team of the Economic Analysis Project ofthe Harvard Institute for International Develop-ment, Jakarta, Indonesia for their valuable com-ments and suggestions. None of these individualsare responsible for the views expressed in thispaper and all remaining errors and omissions re-main our own.

APPENDIX A

We assume that the sales revenue represents aflow over a period t equal to the period betweendebt payments. Thus, we can discount both thesales and the debt payments as follows.

Consider a stream of debt payments:

PDV (payments)

=r*d(e)

(1+rd)+

r*d(e)(1+rd)2+

r*d(e)(1+rd)3+

r*d(e)(1+rd)4+ . . .

=r*d(e)� 1

(1+rd)+

1(1+rd)2+

1(1+rd)3

+1

(1+rd)4+ . . .�

.

This is a geometric series that converges to:

PDV (payments)=r*d(e)

rd

.

The same is true for the PDV of revenue flows.

APPENDIX B

The derivative of v with respect to rd will bepositive if and only if

−w(rd+o)

(rd−1fe e

rd−1)r*[g+2u−1]

−eN

rd

rd(1−d)[fef

N(1−d)+2r*wueuN(1−d)]

\)−o

[(h−1)th+1]rd

2t2

).

APPENDIX C

The elasticity of N with respect to rd is given by

eNrd= t−1! od(h−1)th−1

rd(1−d(1− th−1))−1

".

Whether this is negative depends on whether thefollowing condition is true:

th−1� o

rd

(h−1)−1nB�1−d

d

�.

Taking the derivative of the left-hand side withrespect to rd, we find that the left-hand side de-creases as rd increases as long as

h\1+rdo

rdo,

which is a condition that is sure to be met for highenough values of the domestic interest rate andthe internal finance premium, and low values ofmarket power. Figure 5 shows the elasticity of Nwith respect to rd for a range of values of thedomestic interest rate.

APPENDIX D

The appropriate derivative is

(v(r

= th−1 ln th2−r*wr−1(g+2u−1)

−f %(N(1−d))(1−d)N %(r)

−2r*wu %(N(1−d))(1−d)N %(r).

This derivative is positive as long as N %(r) isnegative, which will be true if

rdth−1 ln th2

(1−d(1− th−1))\1.

In Figure 6, we graph the derivative of v withrespect to r for a range of values for the domesticreal interest rate.



Figure 5. The elasticity of N with respect to the domestic interest rate.

Figure 6. Derivative of v with respect to r.

APPENDIX E

The derivative of v with respect to the internalfinance premium o is given by

(v(o

= (h−1)th−2rd−1+ t−2rd

−1−r*gw %(o)

−f %(N(1−d))(1−d)N %(o)



−2r*wu %(N(1−d))(1−d)N %(o)

−r*(2u−1)w %(o).

As the derivatives of both f and u with respect toN are positive by assumption, (and it is easilyverified that the derivatives of both N and w withrespect to o are negative) this derivative is greaterthan zero.

N %(o)= −� Nd(h−1)th−2

(1−d(1− th−1))rd

+N

(rd+o)nB0.

w %(o)=−w

(rd+o)B0.

An increase in the share of firms following theQDW strategy, d, increases v :

(v(d

= −f %(N(1−d))(−N)N %(d)

−2r*wu %(N(1−d))(−N)N %(d).

This derivative is positive as it is easily verifiedthat the derivative of N with respect to d ispositive.

The derivative of v with respect to the countryrisk premium l is negative:

(v(l

= −r*w %(l)(g+2u−1)B0,

because w is easily verified to be a negative func-tion of l (via the effect of the risk premium on theexchange rate e).

The same is true for the initial level of debtitself, d(e):

(v(d

= −r*(g+2u−1)w %(d)B0.

NOTES

1. See, for example, the overview presented in Sachs(1989).

2. While stylized, this assumption is hardly unrealistic.Examples of similar debt moratoria occurred in Rus-sia (summer 1998) and in the Latin American debtcrisis of the 1980s.

3. Note that we have excluded the impact of a changein the discount rate as both strategies are discountedidentically in this comparative statics exercise.

4. See Appendix C for a more formal discussion of theelasticity of N with respect to rd.

5. We assume that for small values of N, each addi-tional firm has a large but decreasing influence onthe probabilities. For high values of N, one addi-tional firm does not have a significant effect. Notethat all of this is based on the assumption of aconstant proportion of firms following the QDWstrategy d. Clearly, this proportion will change overtime according to the economy’s location on thephase diagram.

6. See, for example, Krugman (1991).7. An article in the 14 October 1998 Asian Wall Street

Journal entitled ‘Asia’s Chances of Attracting CapitalFade’ concludes ‘. . . [the] debt pileup is creating afresh barrier to new capital coming into the region.That’s because the creditors chasing after their olddebts want a piece—or sometimes all—of anymoney their debtors obtain. This has blocked notonly new equity investment, but also the structuredfinancial deals, such as borrowing against futureincome streams, that helped get Mexico out of itscrisis.’

8. Bank Indonesia press release in the Jakarta Post, 25February 1998.

9. In mid-October 1998, PT Modernland Realty, aproperty development firm, was declared bankrupton the basis of a US$12400 claim. On 7 December,this claim was thrown out on appeal. See, ‘RIThrows Out More Bankruptcy Suits’, Indonesian Ob-server, 8 December 1998. Another striking article inthe September 1997 issue of The Economist describeshow PT Steady Safe, a company primarily engagedin taxi services, bankrupted it primary creditor,Peregrine Ltd. of Hong Kong, by defaulting on itsdebt. PT Steady Safe, however, has still (to this day)avoided bankruptcy.

10. In an article in the Indonesian Observer on 15 Decem-ber 1998, it was reported that 63 companies withcombined debts of US$7.5 billion (or roughly 13% ofnon-bank corporate debt) had joined the initiative.This simply means that they can engage in a struc-tured dialogue with their creditors and, in somecases, receive streamlined approval from govern-ment ministries for restructuring deals.

11. See, for example, the 4 November 1998 Asian WallStreet Journal article, ‘Why Indonesia Never Got aDebt Deal’. On 13 December, the debt restructuringplans of PT Astra International were derailed by thereluctance of Japanese banks to accept a temporarymoratorium on interest payments. See ‘Astra Im-passe in Debt Talks may be Lengthy’, Asian WallStreet Journal, 14 December 1998.

12. Ibid.13. ‘Japanese Banks’ Exodus from Overseas Escalates’,

Asian Wall Street Journal, 16 November 1998.14. For example, Krugman (1991) used a value of 4 for

s.15. Lane et al. (1999) stated, ‘. . . high debt-equity ratios

in the corporate sectors as well as systemic andstructural problems . . . together with the prevalenceof unhedged foreign currency liabilities of these



countries financial and corporate sectors, meant thatexchange rate depreciation could also have a sub-stantial effect on the real economy’ (p. 55).

16. The literature on this debate has grown with thepublication of several recent papers in addition toFurman and Stiglitz (1998), including Goldfajn andGupta (1998), Goldfajn and Baig (1998) and Kraay(1998).

REFERENCES

Andrabi T, di Meana A-R. 1994. Bank regulation anddebt overhang. Journal of Money, Credit, and Banking26(3): 460–478.

Bird K. 1996. Industrial concentration and competitionin manufacturing. The Indonesian Quarterly 24(2): 140–147.

Bond T. 1996. Monetary independence with an exchangerate target: the ‘impossible trinity’ in Indonesia.Mimeo, Harvard University.

Claessens S, Djankov S, Ferri G. 1998. Corporate distressin East Asia: assessing the impact of interest andexchange rate shocks. Mimeo, World Bank.

Cline W. 1995. International Debt Reexamined. Institute forInternational Economics: Washington, DC.

Dixon H. 1994. Imperfect competition and open econ-omy macroeconornics. In Handbook of InternationalMacroeconomics, van der Ploeg F (ed.). Blackwell:Oxford.

Dooley M. 1994. A retrospective on the debt crisis.National Bureau of Economic Research Working Pa-per No. 4963, Cambridge, MA.

Fernandez-Ruiz J. 1996. Debt and incentives in a dy-namic context. Journal of International Economics 41(1–2): 139–151.

Furman J, Stiglitz J. 1998. Economic crises: evidence andinsights from East Asia. Brookings Papers on EconomicActivity 2: 1–135.

Goldfajn I, Baig T. 1998. Monetary policy in the after-math of currency crises: the case of Asia. IMF Work-

ing Paper 98/170, International Monetary Fund,Washington, DC.

Goldfajn I, Gupta P. 1998. Does tight monetary policystabilize the exchange rate?. Mimeo, InternationalMonetary Fund, Washington, DC.

Helpman E. 1988. Voluntary debt reduction: incentivesand welfare. National Bureau of Economic ResearchWorking Paper No. 2692, Cambridge, MA.

Kraay A. 1998. Do high interest rates defend currenciesduring speculative attacks?. Mimeo, World Bank,Washington, DC.

Krugman P. 1988. Financing vs. forgiving a debt over-hang. National Bureau of Economic Research Work-ing Paper No. 2486, Cambridge, MA.

Krugman P. 1991. Increasing returns and economic ge-ography. Journal of Political Economy 99(3): 483–499.

Krugman P. 1999. Depression economics returns. ForeignAffairs 78(1): 56–74.

Lamont O. 1995. Corporate-debt overhang and macroe-conomic expectations. American Economic Review 85(5):1106–1117.

Lane T, Ghosh AR, Hamann J, Phillips S, Schulze-Ghat-tas M, Tsikata T. 1999. IMF-supported programs inIndonesia, Korea, and Thailand: a preliminary assess-ment. Mimeo, International Monetary Fund, Washing-ton, DC.

Matsuyama K. 1995. Complementarities and cumulativeprocesses in models of monopolistic competition.Journal of Economic Literature 33: 701–729.

Prasetijo RB. 1998. Deleveraging ASEAN-3Jardine Flem-ing Research, Regional Strategy Paper, PT JardineFleming Nusantara, Jakarta, Indonesia.

Radelet S, Sachs J. 1998. The East Asian financial crisis:diagnosis, remedies, prospects. Brookings Papers onEconomic Activity 1: 1–74.

Sachs J. 1989. The debt overhang of developing coun-tries. In Debt, Stabilization and Development: EssMs inMemory of Carlos Diaz-Alejandro, Guillermo C (ed.).Blackwell: Cambridge, MA.

Sachs J. 1995. Do we need an international lender of lastresort? Frank D. Graham Lecture, Princeton Univer-sity, April 20 1995.


holding out for a haircut: financial crisis, moral hazard, and interest rate policy

Documents