LIBRARIES qfTE^V, HB31 .M415 Massachusetts Institute of Technology Department of Economics Working Paper Series DOLLARIZATION OF LIABILITIES: UNDERINSURANCE AND DOMESTIC FINANCIAL UNDERDEVELOPMENT Ricardo Caballero, J. MIT Dept of Economics Arvind Krishnamurthy, Northwestern University Working Paper 00-14 July 2000 Room E52-251 50 Memorial Drive Cambridge, MA 02142 This paper can be downloaded without charge from the Network Paper Collection at com/paper taf?abstract_id=XXXXXX Social Science Research http //papers ssrn : . . . MASSACHUSETTS INSTITUTE OF TECHNOLOGY LIBRARIES Massachusetts Institute of Technology Department of Economics Working Paper Series DOLLARIZATION OF LIABILITIES: UNDERINSURANCE AND DOMESTIC FINANCIAL UNDERDEVELOPMENT Ricardo J. Caballero, MIT Dept of Economics Arvind Krishnamurthy, Northwestern University Working Paper 00-14 July 2000 Room E52-251 50 Memorial Drive Cambridge, MA 02142 downloaded without charge from the Social Science Research Network Paper Collection at http://papers.ssrn. com/paper. taf?abstract_id=XXXXXX This paper can be — Dollarization of Liabilities: Underinsurance and Domestic Financial Underdevelopment Ricardo J. Arvind Krishnamurthy* Caballero June 27, 2000 Abstract While there kets' crises, much disagreement on is still the causes underlying recent emerging mar- one factor that most observers have agreed upon (foreign currency) denominated external debt is that contracting "dollar" —as opposed to domestic currency debt created balance sheet mismatches that led to bankruptcies and dislocations that amplified downturns. Much of the analysis of the "currency-balance sheet channel" hinges on the assumption that companies contract dollar denominated debt. Yet there has been little this systematic inquiry into why companies must or choose to take on dollar debt. In paper we cast the problem as one of microeconomic underinsurance with respect to country-wide aggregate shocks. Denominating external debt in domestic currency equivalent to contracting the same amount of dollar-debt, against shocks that depreciate the equilibrium exchange rate. level international financial constraints justify the agents are risk neutral. However, will if is complemented with insurance The presence of country- purchase of such insurance even if all domestic financial constraints also exist, domestics undervalue the social contribution of contracting insurance against country-wide shocks. Foreign lenders will reinforce the underinsurance problem by reducing their participation in domestic financial markets. JEL Numbers: F300, F310, F340, G150, G380 Classification Keywords: Currency mismatch, balance sheets, financial constraints, international liquidity, insurance, contingent credit lines, put options, capital flows, thin markets, limited participation. 'Respectively: sions. Caballero MIT and NBER; Northwestern University. thanks the NSF We thank Adriano Rampini for financial support. E-mails: caball@mit.edu, for useful discus- a-krishnamurthy@nwu.edu. Digitized by the Internet Archive in 2011 with funding from Boston Library Consortium Member Libraries http://www.archive.org/details/dollarizationoflOOcaba Introduction 1 While there is still much disagreement on the causes underlying recent emerging market one factor that most observers have agreed upon crises, currency) denominated debt — as opposed is that contracting "dollar" (foreign to domestic currency debt — created balance sheet mismatches that led to bankruptcies and dislocations that amplified downturns. Much of the analysis of the "currency-balance sheet channel" hinges on the assumption that companies contract dollar denominated debt. 1 Yet there has been into why companies choose little Companies to take on this type of debt. systematic inquiry certainly recognize that contracting dollar debt brings with it the cost of a balance sheet mismatch. other hand, the attraction to dollar debt is that a nominally - at lower interest rates. risk for a company in Is company much take on too 2 world. - at least an emerging market? Do prices allocate the risk efficiently? risk of dollar Should a debt and therefore of it? At an abstract denominated debt able to fund itself the the low price of foreign debt worth the balance sheet maker be concerned that companies underprice the policy is On level, is the decision to take on dollar debt as opposed to domestically a decision to opt out of insurance against bad aggregate states of the Imagine a Thai company that faces a currency denomination choice in taking on one period debt. Suppose that next period there are two (aggregate) states of the world: in the good state the baht/dollar exchange rate purchase a dollar and the exchange rate will have assets that are worth of debt, its value next period A is, Good While if the company took on Normalizing by setting D^, cases is in the = that in the former the one, in the bad it takes more baht two. In either state of the world, the baht. Then, if the company chooses to take on A - Dd D b : or baht debt, A - Db its Bad A - 2D d : value to company D d dollars or Bad . is, : A - Db . Db, we can see that the only difference between these two company is required to make an additional payment of Dd bad state of the world. Thus, one can see the benefit due to issuing baht debt as that of purchasing insurance against the bad aggregate state. 1 state 3 : Good is is For models of the balance sheet channel in emerging markets, see The e.g., firm buys a put option Caballero and Krishnamurthy Krugman (1999), Aghion, Bacchetta and Banerjee (1999), Chang and Velasco (1999). From general equilibrium asset pricing results, it may seem odd that we refer to insuring against (1999), 2 gate shocks. 3 We With heterogeneity such insurance have assumed there is no bankrupty is desirable, will be traded, here, or that A> 2Dd- and can affect real aggre- outcomes. paying which Da is in the bad state of the world. The cost of this strategy the option-premium, is reflected in a higher ex-ante interest rate vis-a-vis the dollar-interest rate. model In our agents are risk neutral but domestics value and all 4 demand insurance because they face a risk of liquidation (or production interruptions) in bad states of the a result of the binding for the country.5 may become world as international financial constraints If, as agents expect the exchange rate to depreciate sharply during bad latter, aggregate states of the world, they will provision international liquidity for these states - meaning they will "purchase" insurance that pays them dollars in the bad states of the world. The question we ask Domestics surance. view only if whether domestics purchase a demand — mestic firm can pledge exchange rate indeed its financial an extra unit of international liquidity markets are well developed, of in- in (assets). the sense that a do- from investment to domestic lenders, the entire marginal product reflects this amount —in our model, the relative price of foreign to reflects the social value of show that when domestic socially efficient the right amount of insurance from the social point of the equilibrium exchange rate domestic assets We will is marginal product and microeconomic insurance decisions support the second best. But when domestic financial markets are underdeveloped, domes- demand tic for international liquidity is determined by the limited availability is by marginal product. The empirical counterpart of of domestic collateral rather than scenario during crises this a wedge between the internal return on investment for a domestic firm and the external return that is promised to lenders. As a result, international liquidity is underval- ued, and insurance to guard against a lack of international liquidity is eschewed by domestic firms. While also driven by an insurance mispricing mechanism, our explanation dollarization is quite distinct from ones that point out that fixed exchange rates offer free insurance and creates moral hazard that distorts investment choices (see and Burnside, Rebelo and Eichenbaum (1999)). In these models, offers free insurance to firms that borrow rowing. In our model, on the other hand, in dollars it is e.g., fixing the Equivalently, the benefit due to issuing dollar debt paying Da option premium it if the bad state of the world arises. in is Dooley (1997), exchange rate and therefore encourages dollar bor- not government misbehavior but financial underdevelopment that creates the private underinsurance problem. This 4 for liability the A company premium on may explain why selling a foreign currency option that chooses to issue dollar debt receives this terms of a lower nominal interest cost, but faces the risk that in bad states of the world must make a higher payment. 5 The hedging motive, as in Froot, Scharfstein constraints create hedging demand in a dynamic and Stein (1993), stems from the observation that setting. However, in our case, the hedging from from an equilibrium-aggregate rather than a microeconomic constraint. The distinction since we address a set of macroeconomic issues surrounding hedging. financial demand stems is important, the dollarization of liabilities change rate systems. problem extends across emerging markets, regardless of ex- 6 In addition to domestic demand problems, under-insurance may be driven by external supply problems. This has been emphasized in the policy literature. For example, Haus- mann, from foreign investors and Guidotti (1990) is emerging markets as the inability to borrow et al (1999) refer to the "original sin" of in A domestic currency. in the context of public formalization of a supply effect is in Calvo debt (see also Calvo (1996)). The argument based on time-inconsistency of the government: ex-post governments will always devalue denominated debt. Seeing to reduce the value of local currency from domestic currency debt. 7 lems. We ' 8 Our framework adds this, foreigners naturally to the list shy away of supply prob- extend our model and show that domestic financial underdevelopment can affect foreign lending supply and insurance foreign lenders find that it is in two ways. less profitable to their entry into domestic financial markets is development First, as financial falls, lend to firms with poor collateral, and hence cut back. Second, when interacted with a thin-market supply externality, we show that this withdrawal raises the risks that the fewer suppliers are exposed to The next and hence can feedback into higher lending and insurance premia. section of this paper lays out the model, while section 3 describes the un- derinsurance problem. Section 4 explores in depth the connection between underinsurance and domestic 5, financial underdevelopment. This section also serves as a transition to section where we discuss external supply problems that contains final remarks. An appendix may follows. See, e.g., the evidence of dollarization of liabilities in fixed as well as flexible Hausmann 7 Section 6 arise in this context. exchange rate systems in et al. (1999). See also Allen and Gale (2000) for a similar argument. In their model, banks optimally invest in foreign bonds and issue domestic bonds. They argue that the risk of inflation may dissuade foreigners to purchase the domestic bonds. These explanations seem most compelling for high inflation countries (Latin America), rather than, for example, the Asian countries where chronic inflation Calvo (2000) points out, connection between Moreover, as hard to extend this argument to private sector debt if one The problem is that the private sector liability dollarization has incentives to choose indeed costly. it is was not a problem. its liabilities to Our explanation accounts and financial difficulties. hedge its assets for private sector is interested in the and avoid balance sheet mismatches, debt choices, and is if they are designed to address precisely this difficulty. Constraints on domestic currency external borrowing may have a domestic policy origin as well. Until recently in Chile, for example, external borrowing in domestic currency borrowing. was taxed more heavily than dollar The Model 2 we In this section present the core model. has three basic ingredients, It necessary to address the issues highlighted in the introduction: First, international insurance shocks. Second, third, it on the presence of external them roots the need for financial constraints and country-wide denomination of external debt to the insurance problem. links the it it of all And on exchange rate highlights the role of imperfections in domestic financial markets determination and insurance demand. Preferences and Technology 2.1 Consider a three period world indexed by = t entrepreneurs/firms and foreign investors. Assume that there is 0, 1, 2. There are two sets of agents: domestic and competitive. All agents are risk neutral a unit measure of each type of agent. Domestics borrow from foreign and investors, contract debt repayments, invest in production at date Then 0. at date 1, there are idiosyncratic and an aggregate shock that determine the injection of funds required to continue production. The aggregate shock only takes on two values: we between the high (aggregate) state and the low state depending on the debts are fully repaid and 2, Production. All production requires imported or dollar goods or baht goods. At date a firm borrows c(k) The function c(k) solution. Once fc < created, the capital is "baht" - b it date 1 2, if all production We things go well, capital generates may be in order to generate an interior generates local currency revenue, and value as collateral will vary with the exchange rate. At date this for at a cost of c(k). Thus, assumed to be convex and increasing is and produces domestic from a foreigner, exchanges 6o dollars imported raw materials and creates capital of size of the shock. agents consume. Finally at date all shall distinguish Rk formalize this below its 9 units of baht goods. However, at interrupted by an idiosyncratic liquidity shock, and the firm is required to import an additional one unit of foreign goods per unit of capital in order to realize output of Rk on the capital that of its is baht. If a firm chooses not to, then capital units, then date 2 output for this firm this firm 9 is tor<R = r + A entirely real, is < 1 is, must import 0k units of goods to compensate the aggregate state Our model falls + 0Rk, We assume that when the aggregate state is good, When output not saved. Thus, suppose that a firm chooses to save a fraction (l-0)rk and its no firm for the shock. is affected by a liquidity shock. bad, on the other hand, half of the firms need to reinvest. hence any allusion to the exchange rate refers to the real exchange The rate. shock bad is country- wide in the sense that a positive measure of firms are affected by state, while affected by good state is 1 At date date it is in the it 2, bad state. 10 The probability of the bad state is n, while that of the — 7r. n the domestic entrepreneurs/firms pay off Ud = where c B is c B + cD c consumption of baht goods, and c D may be Foreigners in the idiosyncratic in that an individual firm has a probability of I2 of being all debts incurred at date from production proceeds and consume the excess. Their preferences 1 it •l called B ,c D are, >0 consumption of dollar goods. is and date on at date and 1 to supply dollars to domestics to finance production. Their preferences are over consumption of dollars (foreign goods) at all three dates, These preferences pin down the dollar risk free interest rate at one. International and Domestic Financial Markets 2.2 Foreigners do not consume any of the output from domestic capital. This is the sense in which capital generates only local currency revenues. Goods produced using domestic capital are non-traded, so that the value of capital falls as the exchange rate between these baht goods and dollars depreciates. Note also that because of in baht goods. We shall assume that domestics have an exogenously of foreign currency revenues arriving at date 2 given willing to accept baht exchange for denoted as e this, foreigners specified endowment by w. Thus, while foreigners may be will immediately swap these in The exchange rate for this transaction is goods as repayment on debts, they the foreign currency endowment. cannot be repaid (baht per dollar). However, for expositional purposes it is easier to proceed by assuming that foreigner lend only against the foreign currency revenue stream of w, so that any swapping of baht for dollars 10 The fact that heterogeneity rises during is only in the background. bad states is not important. We need heterogeneity state for domestic financial markets to matter, but adding heterogeneity in the in the bad good aggregate state would not add any substantive insight and would complicate the expressions. 11 R We also need to make an assumption that date investment in capital is sufficiently profitable. Thus, and r are high enough that, (1 where w is the firm's date 2 endowment - n)R + k^y1 >c(w), of foreign currency (see below). This expression just says that despite the possibility of the production shock at date profitable. Finally to guarantee an interior solution (1 - n)R + (i.e. fc < l c~ (w)) we assume that, n^^- <(R~ r)c'(w). 1, investment is Assumption 1 (Foreign Lending) All foreign lending takes the form of debt contracts currency collateral of w. Lending is is that are fully secured by the foreign default free, so that the maximum repayment on debt w. By debt we mean contracts that are not contingent on purely idiosyncratic performance. As one might imagine, flexibility in specifying Of firm could provide greater insurance. debt repayments contingent on the type of course, in practice, these shocks are idiosyncratic and may be hard to observe so that contingent contracts are not enforceable. Assumption the date 1 2 (Non-observability of Production Shock) The identity of firms receiving production shock in the low state is and private information of the firm, is not observable by any lender. Debt repayments aggregate state, are, however, assumed to be To be more to. u £ Definition 1 Let made contingent on the (observable) precise: {l,h} be the state of the world at date A 1. date dollar debt contract between a domestic firm and a foreign investor will specify date 2 repayments, f", and date funding of bo dollars. Since foreign investors are risk neutral, competitive, and the dollar interest rate is one, b = l 7Tf + (1 - n)f h Additionally, since all debt is fully secured, f As we argued is equilibrium, e liabilities is making <w in the introduction, the question of dollar liabilities vis-a-vis baht liabilities simply a question of l w > e h . how high / is relative to f h . We shall shortly prove that, in Since the effective baht repayment for the firm contracting dollar e^/", equal dollar repayments in high and low states would have the firm larger baht denominated, then repayments /' in the would be lower low state. for On the same f the other hand, if liabilities were baht h . Unlike foreigners, domestics do value baht revenues, so they are willing to lend to each other against domestic capital. However, the domestic financial market so that not all of the output from capital is is (l—9)rk+9Rk would be domestic clearly larger. In section 4, we shall underdeveloped, pledgeable to a domestic lender. Only rk of the date 2 goods serves as domestic collateral. developed, then is Instead, if collateral. work through the financial market was fully For 6 > 0, the latter expression this case to clarify the model. Assumption 3 (Domestic Domestic firms may lend is fully Lending) each other at either date to secured by baht revenues. or date However, the domestic credit 1. All domestic lending market is underdeveloped, so that only rk of the date 2 baht revenues can be used to secure financing from another domestic. The limited domestic lending part of this assumption underinsurance result, for it is is the central ingredient for our responsible for the emergence of a gap between social and private valuation of external insurance. This be clear after we describe will determining the equilibrium exchange rate during external its role in crises. Decisions and Equilibrium 2.3 Let us suppose that a firm borrows &o funds at date and creates capital of 0, k. At date there are two possible states of the world. In the high state, there are no shocks, and firms continue to produce Rk. Net of the repayment of f in the h , 1, all this implies that entrepreneurs high state consume, V h = Rk + w - h f . In the low state, firms are divided into two groups: those that receive the shock are distressed, while those that do not are intact. Consider the raising funds to alleviate its production shock. 2 of (1 — 6)rk + 9Rk raise finance and A problem for a distressed firm in choice of 9k will result in output at date goods. In order to save a fraction 6 of distressed capital, the firm must reinvest 6k imported goods. It can do this in can go to foreigners to raise additional funds. That is two ways. First, the firm the firm can always raise directly, h < w - f. The latter quantity will firms, date A which also 1 is also w— always be positive in equilibrium. 12 The rest must come from intact have access to foreign investors since their capacity to borrow abroad at l f . distressed firm can use its baht collateral of rk to Since the exchange rate from intact firms. is e Through markets in our framework economy and pledge l , the firm can borrow a this "credit chain" — the distressed firm borrow dollars from intact maximum amount of 6f < firms. ^r dollars — which represents the domestic financial is able to aggregate the resources of the this to foreigners to raise funds for date 1 reinvestment. We can then write the problem of a distressed firm as, Since firms are ex-ante homogeneous w> /', otherwise there would be no international liquidity the aggregate in the low state. But since reinvestment must be sufficiently profitable for external be interesting, we rule out such corner. left in crisis to f V (PI) l s s.t. =ma,x gbib r> w + (l-6)rk + 6Rk- (t) h (n) £ + h<w-f + $ (Hi) Ok than and l = h +^ < < 1. (iv) (i) (9 are balance sheet constraints (net marketable assets greater (ii) the resources received by the firm at date (iv) is An new investment must be while constraint (Hi) reflects that liabilities), intaking on debts of 1 b\ fully paid and bf with Constraint . purely technological. intact firm at date has only one decision: 1 how much finance will distressed firm. Suppose that the firm accepts claims at date baht) in return for making a date 1 it extend to the of x\ (face value of date 2 l 1 V} (P2) contribution of x\^/e dollars, then, w + Rk + = max Xl x1 - ^ f<w-f s-t. The -h-bf <w- f b Constraints l constraint reflects the fact that an intact firm can, at maximum, lend w—f l dollars to the distressed firm. Date At date problem. 0, a firm looking forward to date and either distressed or intact, 1 can expect to find itself as low or the high state. Thus the decision at date in either the is, maxMoJ . (F3) - *)V h + tt(FJ + ^)/2 (1 s.t. f b h J <w l = c(k) An Equilibrium. (fc,6o,/^) and (9, equilibrium of this 6f ,&f, xf ), = economy respectively, problems (PI), (P2), and (P3) given firms' irf + (1 - n)f h bo- consists of date and prices prices. e". and date 1 decisions, Decisions are solutions to the At these prices, the exchange market clears. The only equilibrium price following must hold Lemma 1 Let c in the domestic • If c and 0, the exchange rate. Given the preferences of domestics, the true. denote the equilibrium consumption of any intact entrepreneur c economy B ,c D > is at date 2. thene = 1. Then, • Ifc The B > 0, but c case of c B D = = then e 0, and c D > > 0, 1. can never occur in our model since production always generates at least some baht, and the domestics must consume this baht. The exchange rate is as long as the solution 1 both baht as well as dollar goods. However, exchange rate we shall the economy runs out of dollar goods, the will depreciate further to reflect this scarcity. Since this be interested the low state. if with domestics consuming interior, is 13 in, we shall construct The equilibrium exchange an equilibrium where rate, e l is this precisely the case is happens at date in 1 determined in the domestic financial market, where intact firms lend dollars to distressed firms against baht: 14 It? = 1*? Let us now study equilibrium in faces the choice of restructuring more in I- state (i) Consider a distressed firm that detail. an additional unit of capital at date 1. Let, A = R-r be the baht return to saving one unit of baht and one dollar equally A> 1 , for capital. remember that a firm values one D + c B ). As a result, if (i.e. U = c First, consumption at date 2 then the distressed firm would choose to borrow as much as it can against its foreign A currency revenues at the dollar interest rate of one, and reinvest these funds to return baht goods at date 1. h=w-f. If the amount raised from international investors, the shortfall. e It will . It can sell up to rk date choose to do this as long as the exchange rate. w— /', is less than the funds needed have to access the foreign exchange market to make up for restructuring, k, the firm will l (2) 2 baht to another domestic at the exchange rate of A> e The maximum amount , or the baht return of funds raised on restructuring exceeds is, r b? As long k, as the the firm 13 is sum of ^y and the right unconstrained in 1 but not at date 2 (liquidity the qualitative aspects of the issues The (3) side of (2) is reinvestment at date 1 more than the borrowing need and all of production units will be See Caballero and Krishnamurthy (1999) for a variant of this model where the shortage of dollars occurs at date 14 its hand < 4- More precisely, what we we "fire sale" dimension of the problem the exchange rate at date is 1, one and interest parity must hold. is The model has domestic interest rate. Choosing This i\ = not important for is are addressing in this paper. refer to as international interest rate the domestic interest rate. crisis). 0, a free parameter in that a date 2 forward exchange rate. Thus, ei(l we do allows us to call this the date 1 + = ii) e2, where not, nor need to, pin exchange rate as well. i\ down is the saved. In this case, the firm will borrow less than ^f (and perhaps less than the international debt capacity). most Intact firms can tender at their excess international debt capacity of They return for purchasing domestic debt. > 1. moment that A > rate exceeds one, or e Assume for a will w— all f goods, which choose to do this as long as the exchange e > l is borrow as much as so that distressed firms 1 Then, directed to the distressed firms. is production units to be saved the economy can in total A necessary condition for that, <*»-"?; (4) f If this constraint is violated, then some 9- = 2 capital units are not restructured at date w-f shall refer to this constraint as the international liquidity constraint. nor (4) binds, all production units are saved. Since in equilibrium is 1, 0<1. l J We the exchange rate in f l they can, and intact firms lend as much as they can. import w— c > When 0, it neither (3) must be that one. The other extreme case when both is (3) and Equilibrium in the exchange (4) bind. market requires that, rk -r el Since (3) binds, distressed firms sell all A = W-f. of their baht revenues to intact firms. many intact firms purchase this by borrowing as distressed firms. Solving for e l , (4) binds, and paying this to yields '-^ho In this case, the exchange rate dollars as they can As is 1 - depreciated since there < is 5» a scarcity of dollar assets relative to baht assets. Now, depending on parameters, there are I four regions that may occur at date - distinguished by the four combinations of binding constraints, (4) and (3). in state 1 We focus on the cases where the international financial constraint (4) binds, and study cases in which the domestic constraint (3) state is or may not bind. If (4) binds, the exchange rate in the / depreciated. Lemma Lemma may 2 (Exchange Rates) If (4) binds, then c D of an intact firm at date 2 1, we can conclude than zero, and e = 1. that e l In the high > 1. If (4) does not bind, state, since there are 10 no D c is zero. of an intact firm liquidity shocks, e — is 1. From greater With on just this constraint, firms will anticipate that taking baht payments in the I in this state. We will where they state. Since this is precisely the state to finance the production shock, they will shy show that away from contracting will need resources make high payments to in this case, firms appropriately value the insurance offered by contracting domestic currency debt. This changes when are credit constrained. dollar debt implies higher We shall show (3) binds, in this case that there is and domestic firms an externality whereby firms undervalue insurance against the low state. Currency Denomination 2.4 as an Aggregate Contingency Let us pause at this juncture and contrast dollar debt and baht debt within our framework. An uncontingent repayments of f h dollar debt contract will specify since the dollar interest = f l = fdollar- Therefore one, is = ^ fdollar + bo,dollar (1 ~ ^)fdollar = fdollar- Next consider an uncontingent baht debt contract with repayments given as f h The fbaht- state, where at date repayments are dollar equivalent e l > 1. the high state, and fbaht in Thus, the amount of dollar equivalent that the firm ^f^ is = in the f l = low able to borrow is, J baht j- + , i , (1 - s ~ ^(e - TT) fbaht = Jbaht = = f Then \ c t I i\ 1) J baht e' Let us normalize by choosing, fdollar fbaht — it is r- & easy to see that, bo dollar ^v.aoaar - n{e " v>- or that, for a fixed face value of debt, the firm is able to borrow less dollars by issuing baht JQ.baht — t l)-r, *j el i debt than dollar debt. The reason behind vis this is that the firm takes exchange rate risk in baht debt, when the firm borrows in dollars one on / baht. Equivalently the firm it is it simultaneously sells its debt issue. Vis-a- a put option struck at able to obtain a lower interest rate on its debt when takes on dollar debt than baht debt, but in return takes on the risk that the exchange rate may depreciate at date With these 3 3.1 1. preliminaries behind, we can turn to our main substantive results. Underinsurance: Excessive dollar Liabilities Competitive Equilibrium versus Planner's Choice Definition 2 // f h and f are debt repayment choices decentralized equilibrium, and F h and F l (in units of dollars) in a competitive are debt repayment choices of a central planner, 11 then we shall say that the The economy has excessive dollar fh Fh fi Fi liabilities if, behind this definition stems from the previous section. logic When normalized and baht so that all repayments are in dollar terms, the only difference between dollar contracted debt repayment in the size of the is repayment in the I Taking on dollar debt means, per unit of debt, higher repayments now Let us (3) and write (4) bind. A and (P2). down the date I state. state. 15 decentralized problem in an environment where both h choice of (k,f ,f debt) in the high state l result in date 2 resources (net of ) from (PI) any contracted of, w- f h Rk + On in the First, let us rewrite (P3), substituting in the value function date h state versus the the other hand in the low state, if ( the firm . distressed, is it has date 2 resources of, w -f)A + ^-A e' This is because (w — project return of A. /') is directly The rk the proceeds invested at A. pledged to foreigners, and the proceeds invested at the of domestic collateral If the firm is [w Thus the date (PA) program maxkif K if l l )e sold at the exchange rate of + w-f h (l~n)(Rk + i f h Rk. + nl{(R + r^)k + (A + ) = l irf + who maximizes an Suppose the central planner makes a date up with resources V h = RK + We allow firms (/ >/ )- To map 1 )) choice of (K,Bo,F ,F Then, at date ), where 1, in l W - Fh and have them choose the pattern of payments language of uncontingent baht and dollar debt contracts, note that spanning results apply. The pattern of repayments of (f We the . there are two states of the world and that these two (uncontingent) liabilities are linearly independent. debt and some dollar debt. sum of, to take on contingent dollar debt contracts this directly into the equally weighted and international borrowing capital letters denote the central planner's aggregate quantities. 15 )(w-f (l-ir)f h of the utilities of the domestic agents, subject to the domestic firms will end l l Consider next the choices of a central planner all e J <w c(k) high state, and is, s.t. constraints. e', date 2 resources are, intact, -f is say there is h , /') can be implemented by taking on some baht excessive dollar debt lower fraction of dollar debt than private agents. 12 Thus when a central planner would choose a In the I state, a distressed firm receives, (W - F )A + l But we since ^A. are interested in an expression that is free of prices, simply substitute the market clearing condition, , tK - e W -F Then, into the above. VB = 2{W - F l Similarly, l )A l an intact firm receives, [W—F )e + RK. l l And substituting in the market clearing condition, = Vl Therefore the constrained max KFhF (P5) debt choices in this economy efficient F h ,F s.t. h are, 16 )+n\({R + r)K + 2A{W - F l <W l = c(K) The W-F {l-Tv){RK + i + R)K (r ttF 1 + - w)F h (1 objective in (P5) represents the constrained efficient solution as opposed to the best solution. Our central planner the economy raises ivF Lemma + 1 — (1 is still ir)F h subject to the international debt constraint - which Fh = 3 In both (P4) and (P5), first f is h limited by i.e. W. = W. Proof: see appendix. The there is fairly obvious. In the h state, since there no reason to leave an slack is borrow date result as much in W as possible against is the debt repayment. in this state no chance of a liquidity shock, Optimality requires firms to and use the proceeds to increase K at 0. Lemma 4 If A > e l , then F < l l f , or debt repayments are set too high in the I state in 1 dollars of the decentralized equilibrium. 16 This expression can be arrived at directly by noting that if (4) the economy must go toward saving capital units. Since there are aggregate, and each generates a return of A, reduces output to rK for one-half the firms, we have the l all the date dollars that the latter half of this expression. we have the former part 13 binds, then W—F of the expression. economy has in Given a shock that Proof: see appendix. The way easiest corresponding to the to understand this result I to contrast the terms in the objectives is and (P5). Suppose that / and state in both (P4) Now, imagine an experiment of increasing f and F by $1. In both cases for more borrowing at date 0, and will raise an additional tx dollars. This l used to increase capital investment by pr^y investing Now cases. In both cases, there consider the cost. distressed capital units in the I were equal. this will allow in turn can be Therefore the benefit to borrowing a dollar and . same amount to increase capital by the it is F l state at date 1 - in is (w i.e. both centralized and decentralized one dollar —f l ) falls funds for saving less of by one The dollar. firms in (P4) take this cost to be, while the central planner recognizes that this cost truly -2A = is, A. 2 The central planner's cost of taking on the extra dollar of debt exceeds the firm's perceived cost by, As long as A> e' it must be that firms set /' > F l . Definition 3 Define the index of domestic underdevelopment as the difference be- tween the marginal profit of saving a distressed production unit and the exchange rate of e l . Sd Then note the Lemma e h = then 1. it l , following: // (3) also binds, then must be that Sd greater than zero environment = e 5 (Domestic Underdevelopment) Suppose that (4) binds in the The reason we sd =A- in = it must > 0. l > Alternatively, if (3) does not bind, 0. refer to Sd as the when be that Sd state so that e I index of domestic underdevelopment is that the credit constraint between domestic firms binds which there are no domestic credit constraint is it is only (i.e (3)). An analogous to one in which 0- Proof: Consider the the firm finds it FOC profitable to for the borrow Assuming maximum borrowing, e l = exchange transaction by a distressed as much rk . fl A> e l , as possible against its baht revenues of rk. Now 14 firm. If there are two cases. If A> _f t , this is an equilibrium and both case maximum when, at is to borrow as capital is less much and (3) borrowing, A < > l that s d we have > 0. Here, a firm will not find . _J^ fl than or equal to the exchange A— e = l 0, rate. we have that As a s d Firms (Excessive dollar liabilities.) 1 e The other profitable it as possible against rk since the return from saving a distressed unit of bind. Additionally, since Proposition A— binds and since (3) = result, < 6f and ^V, (3) does not 0. economy in an in which Sd > 0, hence (4) are binding, contract excessive dollar liabilities. 3.2 The Externality When a domestic firm makes its financing and production choices, it looks not only at the revenue generated by capital investment but also at the liquidity services that this capital provides at date That 1. investment decisions completed. the creation of a liquid asset is, when a sudden need From the aggregate liquid assets is for fresh resources an important aspect of real may arise before projects are point of view of the central planner, only internationally have such value, since date 1 may be production needs satisfied only with dollars. Comparing (P4) and (P5) at /' = F l reveals that, relative to the central planner, individual agents add a term to their objective function which microeconomic liquidity service of domestic and international 4((A-e<)^-(A- e When the economy in the the margin If distressed, at its is state, a firm I it uses its may be coincide with that of the central planner, there s d > full directly related to the assets: (u,-/<)). (6) distressed or intact with equal probability. domestic asset international liquidity to distressed firms. the latter must yield the <) is (i.e. rk) as collateral. In order for the microeconomic problem to must be no rent in the marginal product of reinvestment. former activity, while It is apparent that when neither condition holds; the collateral value of domestic assets the return on supplying valuable international liquidity Both problems have the same root. If not, it sells When is is overvalued while depressed. domestic financial markets are underde- veloped, distressed firms can only pledge a fraction of their marginal product of date investment of A. news for assets, The constrained nature of their demand for international liquidity is 1 good a distressed firm in need for such liquidity and in possession of pledgeable domestic because it faces less competition for international liquidity, retain a fraction of the surplus from fresh investment. attempting to profit from its It is and bad news is for thereby able to an intact firm international liquidity, because this firm receives less than the 15 social surplus spread of A 4 Sd- due to investment. Both margins are distorted exactly by the domestic credit U Closer Look at the Connection between Financial Un- derdevelopment and Underinsurance We have argued that excessive dollarization of market underdevelopment. When between domestic agents, agents on domestic currency steps. First, we liabilities stems from domestic financial there are credit constraints affecting borrowing/lending will systematically liabilities affords. undervalue the insurance that taking This section develops this argument further from the direct effect of credit constraints Second, we show that when domestic financial constraints vanish is two relax the credit constraints of a few domestic agents in order to separate the effect of distorted asset prices best in for all We conclude that it is choices. domestics the second attained, despite the fact that the aggregate shocks have a real on the economy. on agents' and negative effect domestic financial constraints, rather than negative shocks, which are behind our underinsurance results. Constrained and Unconstrained Firms 4.1 Suppose that a fraction A of the domestic firms face no domestic credit constraints. Thus, for these firms the date debt constraint 1 6f < For 6 > 0, it is is, c-yy ( 7) clear that, (l-0)r + 6R, k> 1 rk — or that these firms are less credit constrained than those of (3). Lemma 17 6 (7) will never bind for unconstrained firms. For the distortion in the relative valuation of the liquidity service of both assets to case that r > 0. Otherwise, the gives liquidity service, and all initial capital On the other hand, the case where r = is is done via changes constraint arising from the limited pledgeability of date friction would no consequence as there date 1 is 1 and domestic 16 if r = we Thus, usually think that capital but there is a domestic financial any down payment available at date liquidity at date no action by a domestic that would and therefore be altered by the distortion. in financing decisions. reinvestment, there would be no externality. affect the leverage coefficient of distort the relative valuations of international must be the not a sufficient condition for the externality to non-generic in the sense that investment creates a liquid asset as a byproduct. Nevertheless, While such a it investment in k does not yield any domestic collateral that liquidity provision can only be to be precise, the presence of domestic financial constraints arise. arise, affect the 1, this distortion amount 1, and would have of domestic liquidity at , Proof: Suppose that these firms chose to save 9 units of capital by issuing debt that Then the maximum amount garners 6k dollars. of dollars they can raise el is, el Since, R=r+ we can conclude A>A>e l that, (l-ey + oR k> 6k. el Thus given any 6 these firms will always be able to obtain funds required to restructure all of their capital units. program Next consider the date Rk + w — f . In the I state, if the firm V\ while if l l s is at the intact is it = Rk + (w- Vh = before in the h state, receives, l l )e f distressed, V =rk + (A- e This As for these firms. because the firm exchange rate of is e l - f)e = Rk l A and generating , all l e k + (w - f)e of its capital units baht at date 2. l by borrowing k dollars Combining, yields the date firm: (l-n)(Rk + max kJ kji s-t. f h w-f h + n({R-^)k + ) e l (w-f 1 )^ J <w c(k) Lemma (w always able to save program of an unconstrained (P6) + )k l = nf + l (1 - ir)f h 7 Let / w represent optimal choices in (P6). Then, for A> e l > r ^ f l /' Proof: see appendix. Thus unconstrained firms are more The reason for this is that, willing to take on on the margin, firms that costly are the ones that prefer to take on baht debt. dollar debt find balance sheet firms. mismatches more However, unconstrained firms are precisely the ones that can afford balance sheet mismatches. 17 than constrained As a result, they undervalue the insurance aspect of baht debt even more strongly, as they prices brought about now Let us by the constrained nature of when A = look at the limit, liquidity still face the distorted asset demand. 18 ' 19 so that no domestic firm 1, is (domestically) credit constrained. In this case, domestic firms follow socially optimal debt choices. Lemma 8 = X If one? (4) binds, then e 1 Proof: Consider investment diction that A> e Then the l . Now save production units. domestic firm and save all is demand and at date < 1. from Lemma we note 6 = 1). firm. Suppose borrow as much as in contrait can and that (7) can never bind, or that the (4) binds, J implies that, This violates market clearing. There The 1. Thus, bf>w-f 18 = never credit constrained. This implies that distressed firms will issue debt Which must be that h by a distressed 1 2 for 6 e distressed firm will choose to of their capital units (6 However, since =A l A= e is excess l . demand for dollars at date 1. As a result it l . restriction that r < e' is fairly weak in the sense that it is sufficient but not necessary, and there is a wide range of parameters for which the result holds. 19 When there is heterogeneity among domestics, the fact that the unconstrained firms take on excessive dollar liabilities can be partially solved via private contracts. 0: pay unconstrained firms in Imagine the following insurance contract at date constrained firms receive dollars from unconstrained firms in the any case). 0. state, and in return constrained firms the h state and in baht (out of the rk of baht collateral, which was unmortgaged This contract would be possible observed at date The value in I if the type of firm (constrained versus unconstrained) This seems a reasonable assumption of a dollar in the I (i.e. state to a constrained firm credit worthy versus less credit worthy is firms). is, ^>. while the cost of mortgaging one unit of the baht collateral in the h state is only one. Thus, constrained firms are willing to pay a high price for this insurance contract. Unconstrained firms find that reducing /' in order to take the other side of the insurance contract costs only A of jp by The selling this insurance. firms' valuation of dollars in the / I state while they stand to reap as high as a premium price implicit in the insurance contract increases the unconstrained state to constrained firms in line with each other. value of dollars in the e', *e from e . However, since and the private value of dollars dollarized. 18 This brings debt choices of unconstrained and A> e', it still leaves and the economy room between the is still left social excessively liability Now e l if e l = A, then Sd = (1 s.t. This program is f h - +w- ir)(Rk f h + ) ir (^k + A(w - /')) l Hence, if A = the economy makes efficient 1, its liabilities. Discussion 4.2 The economy with A = has aggregate shocks and exchange rate fluctuations, yet there 1 = no underinsurance. The only difference between the case with A > in the I state when A = 0. It is > domestics undervalue insurance against the specification of shocks state, and Sd = in the / Hence we have argued that 0. state. At an abstract theoretically possible to construct a it is I state. In this case reality I it is by high domestic spreads, and only indirectly The general principle behind our result is for that model in in this case with a which s^ different- > in the result that domestics state of the world. perfectly To is that there is us, this in underinsurance worth pointing out that underinsurance directly demand 1 is - domestic spreads tend to be higher periods - thus the natural prediction of the model against these states. Nevertheless, = The model we have presented aligns the incidence of a high domestic spread with the seems the empirically plausible model of level, we would have the odd don't sufficiently value insurance against the h state. crisis and A important to recognize that the underinsurance result only arises in states of the world where Sd h (P6) modified to let is J <w identical to that of (P5). choices in the denomination of Sd a firm at date for =A. max kJh ji is The program 0. is caused by bad shocks. that credit constraints leads to constrained funds - those in need of funds are not credible in transferring surplus created by these funds to the providers of these funds. Suppliers of funds, find that the business of lending to firms with bad collateral is why dynamic context, not a particularly profitable business and transfer their resources elsewhere. In the next section, also helps explains in a we argue that this logic the entry of "specialist" foreign lending into domestic markets (i.e. retarded. This also suggests a supply channel for why credit line facilities, foreign banks) is foreign lenders are reluctant to lend in domestic currency. 5 Limited Foreign Credit Lines: Further Costs of Domestic Financial Underdevelopment Foreigners, thus far, are passive lenders. They make no profits and simply lend at the international interest rate of one. Needless to say, such characterization of foreigners (and 19 — many domestic savers) is hardly of financial underdevelopment realistic. This section extends the model to study the effects we introduce an active on foreign lending decisions. For this, margin whereby foreign lenders may pay a fixed cost and choose to specialize the domestic market. it The takes. in lending to We characterize both the quantity of foreign lending and the form that quantity results stem directly from section If 4. domestic financial markets are underdeveloped there will be limited entry of foreigners into the domestic market, as lenders find that the business of lending to firms with poor collateral one. we show Conditional on entry, extend look very much like increase the interest rates not a profitable that the optimal contracts that these foreign lenders contingent credit we introduce complementarities lines. Finally, and demonstrate how in foreign lending decisions is financial underdevelopment may also and insurance premia charged by foreign lenders. Foreign Specialists 5.1 Let us return to the model of section 4.1, F 1. endowment measure of foreign lenders, each with a date can choose to pay a fixed cost of < A < with (dollars) at date Suppose that there in dollars of w^ . is a unit Each lender domestic in order to "specialize" in lending. If they pay the fixed cost, they can value baht goods as do domestic agents: U = cD + cB while if they do not: U = cD . Paying the fixed cost allows specialists to count the rk of baht output as collateral any lending. With this modification, we capture the idea that for specializing in the domestic market enables the investor to receive higher returns on lending to domestics.20 A specialist has w? But he can improve dollars that he will invest in loans backed return (see below) its with foreign non-specialists. international liquidity date vjf dollars into when — 7T he if first by domestic baht enters into a contingent agreement In particular, he will want to have the it is most valuable (the dollars at date 1 in the I I state). state He does — and by transacting with other foreign non-specialist lenders. That a contingent claim that pays - dollars in the I collateral. maximum so possible by converting dollars in the his h state the specialist invests in is, state. Since foreign non-specialists are risk neutral, they would willingly supply such a claim. 20 Concretely, one might imagine these baht goods are acceptable after date 2, in this way and the foreign retrieve his specialist payment if there are many more periods can reinvest these baht goods in production of the export sector and payments over time. In knowledge of business practices that enables this case, we should this specialist to 20 interpret remain in the F as the cost of acquiring the domestic market. e The rate of e — l dollars can then > be invested against baht collateral at date 1 at the exchange earning the specialist, 1, — baht at date 2 in the I Thus entry state. gives expected utility of, Ue = w f e l F, while non entry gives utility of U ne = w f Let us denote by market. The < a< 1 . the fraction of foreigners that choose to enter the domestic determines that a free entry constraint w f(e - 1) = l Note that Sd if is be limited. will high, then e This is l — 1 will be low is such that: F. given participation) and foreign entry (for the effect of financial development on specialist entry. Before characterizing the equilibrium with entry, let us define the lending contract. Contingent Lending Contract 5.2 Definition 4 A (Specialist Lending Contract) and a domestic firm contract between a foreign specialist payments from specialist to will specify, (d$, d§,d§), as domestic firm. The resource constraint for the lender will require that, + (l-vr)4 + 7r4 = ^. d° That is, the expected present value of these (8) payments must equal w^ (recall that the specialist has already signed a contingent contract with foreign non- specialist) specify baht repayments of(t h ,t l ) from firm n 4-h t if the firm is h firm The is last The contract will also to specialist lender, satisfying, J < Rk l ,t unconstrained, and t if the . A <rk l constrained. two constraints are simply that promised repayments of collateralized. Since the specialist can always earn e — l 1 on his (t h ,t l ) must be dollars, the contract fully must satisfy, (1 - n)t h + wb l = l e (d°s 21 + (1 - 7r)dJ + tt<4) (9) Consider next the problem a constrained firm. In addition to making a borrowing for decision from foreign non-specialists, it can also choose to borrow from foreign (1 - n)(Rk + w + specialists. Thus, (P7) rnax fei/ fc i/i)do id id , £ >tV Tri ((i? si- f t h h + i' (r - - fh - lemma following Lemma h ) + + (A + i')f )fc l - < e')(u; + 4 - /')) rk tt)*' 1 + Tri' = e'(d° c(A:)=^/'-+(-l-7r)/ The t ,f<w ,t (1 - d£ ft + (1 +4 tt)4 + tt4) characterizes the solution to this problem. 9 (Contingent Credit Lines) The optimal specialist lending dollars at date and at date 1 in the receive baht repayments in the h state, have the specialist lending contract will I In return, the specialist will state. and potentially some baht repayment in the This can be interpreted as a credit line from which the firm draws down d$ at date d$ in the I I state, a fee We and pays a state, oft fee oft h in the high state, and if state. I 0, drawn down the line is and in the 1 . relegate the proof of this lemma to the appendix, but the essence of can be it immediately seen from the structure of the constrained firm's problem: • For the same reason that f h = w, the firm will choose to set dj having excess dollars in the h state • > Moreover, since production at date 0. Firms don't value having any slack in the Borrowing against one unit of domestic date in the of 4- However, waiting I till is is h 1 a profitable enterprise, dj state. t = As a is, since 1 in / I l t > borrow state generates dollars at borrow against unit of domestic the 0. rk. collateral in the to > result they will fully state of — . t collateral In the objective function, weighted by n, and hence we can conclude that the firm any choice of is indifferent on 0. Equilibrium, Entry, and Financial Development Let us turn next to equilibrium and ask date date state generates dollars at date the latter 5.3 That not valued, the firm would prefer to have none. against their domestic collateral of rk and set • 0. Likewise since dollars are valued in the low state to cope with the production shock, dj • is = how many 0. 22 specialists enter the domestic market at At date in the 1 I of domestic collateral market clearing has distressed constrained firms state, and distressed unconstrained firms = l then t 0, l otherwise t > 0. In fact there are two cases. l are indifferent on the choice of t 21 . For simplicity, dollars: , _ (1 - dj a that valid) is X)rk w- f 1 Rk — t aw^e < . — l t — X)rk + XRk, (1 = l t (there Since intact firms have w — f l is a + d'$ + XRk +4 specialists choose to enter the market, =a a* . 1 If, rk As we noted, firms l us take the case where let wide range of parameters where this assumption Now suppose = 0, we have, selling selling then rewriting (8) with . TV Substituting this into the above expression yields: l e = ((l-X)r + XR) - w— V+a All things being equal, e l is decreasing in a. In fact, k, f « wf J 7T and d^ changes since these quantities are determined in equilibrium. However, ^< that the in / 22 0. As more foreign state rises specialists choose to enter the market, the worth noting static market has more unconstrained firms so that 1 and the return on lending demand l easy to show supply of dollars in > for That 0. more developed, there it is as the domestic is, and hence the entry decision. The more domestic is e l rises. Figure solid line represents funds as a function of the exchange rate. The curve for foreign specialist demand represents that^- is dollars to these firms rises below represents this as an equilibrium the it is e and hence the exchange rate appreciates. The next comparative collateral change when will also an economy which is more financial developed (i.e. A is d! higher). Since foreigners require that, l e + 1 = — w jr J 21 The logic behind this constraint the h state. At of aw' " maximum, this is (1 that firms always prefer to pay foreign specialists with collateral in is — X)rk + XRk. The that enter the domestic market. The algebra is as follows. Differentiating de •"'-*$ * /'+Q„. Foreigners require a return of e on the aggregate dollars constraint follows. and rewriting, l 9a x \{1-X)r + \R Since, #r , In- , fl-f < 0, we have the sign. 23 in order to enter, in is exactly e = l 1 + — -±j w the region where a < a 1, 1. f a Figure Lemma a Entry by Foreign Specialists < 1 and l t = specialist entry into the 0, a is increasing in A. Financial develop- domestic lending market. Supply Multiplier and Financial Development Let us pause for a is 1: 10 In the region where a ment increases foreign 5.4 such that the equilibrium exchange rate is moment and take stock. Our explanation that domestic firms look at the interest rate charged this to the interest rate when borrowing charged in dollars, fully recognize that there they opt for dollar borrowing, and yet conclude that the central planner does) in dollars. The explanation insurance against bad aggregate outcomes. But another side of the story. Lenders vis-a-vis lending in dollars, the former option. for excessive dollar liabilities — i.e. it it is is in baht, compare exchange rate risk preferable to borrow (more than rests purely on domestic demand would appear that mostly foreigners in reality there is it is for also — may evaluate lending in baht and conclude that they require an extremely high premium Below we show that if for possible to connect such supply response to the very same deficiencies that led to depressed demand. Of course, one to what? In our tricky question that needs to be addressed model with passive eigners were risk neutral fair prices. is: too high a premium relative foreigners, this question could not and competitive and provided insurance be addressed. For- elastically at actuarially Introducing foreign specialist lenders gives a point of comparison whereby this question can be addressed. 24 We have already seen that foreign specialists provide an insurance role by entering market and accepting baht into the domestic collateral against their loans. Indeed, the contingent credit lines that they extend are fashioned to provide insurance against the state (especially when l t = Limited insurance supply would 0). mean that foreign specialists require too high a return on their provision of contingent credit lines. For example, and receive back dollars at date foreign specialist were to lend dollars at date I charge at the international interest rate of one on this loan. However, if 1, if a he would the same specialist received baht collateral against his loan, then (9) tells he would require a return of e' on his loan. In the case of contingent credit lines with (l-n)t h = Thus, we shall interpret e l — 1 as the e l = r (4 + premium 0, the foreign specialist requires that, ndl). baht collateral for lending against for a foreign specialist. Moving beyond emerging market for limited insurance example, is it debt, there seems to be some theory and evidence supply in other asset markets leading to high liquidity premia. For often pointed out that the catastrophe insurance market is severely capital constrained leading to very high insurance premiums and insurance cycles (see Froot and Some O'Connell (1997)). during the fall in the US credit LTCM, grew constrained and became liquidity users as they exited from 23 Given our focus on domestic underdevelopment, we shall restrict attention to studying a channel whereby domestic under velopment causes too high a lending premium. domestic market is underdeveloped, foreign specialist entry lending in the domestic market unprofitable. may supply affect is Now is phenomena one plausible avenue through which a thin market externality. That is When is, the also limited, as foreigners find this suppose foreign lenders prefer to lend in a market in which there are already other foreign lenders. for this market of '98 argue that hedge funds which are normally liquidity suppliers to markets, such as markets. meltdown of the explanations for the provided in Allen and Gale (1994). We A microeconomic model defer to their results, and take a reduced form approach to this issue by positing complementarity in foreign entry decisions. 24 Suppose that F depends on the amount of F(a) *""* = <- if 23 a > a. See Xiong (1999) for an analysis of wealth constrained hedge funds, or Allen and Gale (1994) for a model with similar results, but in a market segmentation, stock market model. See Holmstrom and Tirole (1998) and Krishnamurthy (2000) 24 entry, so that, risk that they may have supply in a macroeconomic context. model must pay a fixed cost to enter the market, and for limited insurance Investors in the Allen and Gale (1994) to liquidate and exit the face the market. In this instance they prefer that there are other investors in the market as this increases liquidation values and reduces ex-ante variance of asset prices. There is thus a complementarity between the entry decisions of investors. 25 Figure 2 illustrates the effect of complementarity. e 1 ' L V s \d \ ' d \ \^ \ F + + wf V \ F — w \ f a Figure At the low \ a the market sufficiently \ Complementary Entry Decisions 2: level of financial is \ a development corresponding to foreign specialists anticipate little entry by others if \ developed d, there is the possibility that and stay out of the market. However, this possibility disappears, as all specialists (d'), expect plenty of entry and therefore enter themselves. Contrasting the equilibrium liquidity premia, we see that as A rises and equilibrium shifts to d', the premium charged by foreign specialists also falls. Final 6 We Remarks began the paper highlighting an apparent dilemma: Excessive dollarization of private liabilities has played a central role in most recent emerging market this is a private decision, why crises. But given that don't firms anticipate the dangers and avoid dollar debt? Our answer began with a motive to hoard international liquidity: firms experience oc- casional needs of external resources and the country has only limited access to international financial markets - moreover, liquidation or limited reinvestment in unfinished projects is very costly. Purchasing international insurance against shocks that increase the aggre- gate ratio of needs to availability of external resources this liquidity, implements The and issuing debt denominated in is an efficient mechanism to hoard domestic rather than in foreign currency this insurance. reason for underinsurance lies in domestic collateral of distressed firms. another financial market imperfection: the limited When 26 these constraints are widespread, there is restricted competition for available international liquidity, since insufficient constrained by is domestic collateral rather than by marginal product. In equilibrium, this leads to a private overvaluation of domestic collateral other domestic collateral A —since syndrome is specialists shrink. financial it is competition with it faces a of liquidity and demand for external the excessive dollar-indebtedness. underdevelopment shrinks the size of the effective international takes only a step to have also the supply of such liquidity by foreign Thus supply and demand In Caballero and Krishnamurthy (2000) tervention to little the insufficient insurance, and a particular dimension of underinsurance liquidity market, is demand and produce the wrong kind particular dimension of this Once domestic there — and undervaluation of international liquidity —since constrained demand. As a result, firms collateral. demand manage the factors reinforce each other. we show that this environment motivates international liquidity of a country. However, attempting to do so within the constraints of dangerous and policy intervention may illiquid also show that domestic financial markets can be Many backfire. we in- of the issues and limitations we discussed there would apply here as well: in particular, to the question of whether a central planner should contract contingent external credit lines for the private sector. Another issue that arises is whether permanently fixing the nominal exchange rate solves the problem by eliminating the need for insurance. At some level the answer to this question is obviously no. rate, Our discussion referred to the real rather than the nominal exchange and the international domestically mandated one. liquidity constraint On is a real resource constraint rather than a the other hand, our current framework is not well suited for addressing such question since our real exchange rate represents a convolution of observed exchange rates and domestic asset prices (interest rates). Different models about goods markets and the extent and nature of segmentation of the different domestic assets markets yield different issue is mappings from our framework to nominal exchange the next step in our agenda. 27 rates. Addressing this A Appendix A.l Lemma Proof of Take the Lagrangian L* for (P5), W-F = (l-n){RK + 3 h ) 2A(W-F n^((R + r)K + + - wF + l X(c(K) (1 - n)F h ) 1 ))- - ^i h (F h -W)- W (F ( - W) First, ^ dL* Likewise, if fih Where we = 0, it = (l- 7r )R + 7T r c would be that, substituted in A from above and noted that the project to arrive at the inequality. Since at date R+ f^-\ >(K) = 4^ > 0, it is must be that sufficiently profitable F h = W. The same logic applies to the case of (P4). A. 2 Lemma Proof of Consider the FOC for F l in (P5), dL* OF 4 ir -—- + n\= 1 Given our technical assumptions, this /' in (P4). This Suppose that ^fc) ((1 F = l A. 3 A> e l . Proof of - *) R+ ensure it n^^ W>F This violates the > 0. - Ac' (K Now consider the FOC for K= , FOC, so it fc ) + A " *K*\ + ( df l ))) k so that, dL* v must be that f- F > if 3, we will have that f h = f h = Sd = A— w. Take the from (P6): d6 c '( fc e > 0. 7 lemma For the same logic as in " Ac '( n dL* Lemma *^ must be that l f then 3C as long as will - n)R + is, = f£ ((1 n l-ir)R + Tr(R--)-eJ'/ c(k) l c'{k) 28 . FOC for /' Likewise, from (P4), dC W Suppose that = /' m = then /', (n it must + d fi-df Which can be A > For e We /'>/'. be that k also {R = — A+e r ' *fn\ c{k) ) and, k, r)+ C c'(k){2 C{h 2 e' >)- rewritten as, > l R+ R+ —2 $ {^-^\i>± * both terms r, in the parentheses are positive. could also rewrite this d f~df In this case, it must be that as, + C(fc) c>(k) This produces the weaker restriction that 2 A> e l e) j- e^ l or Sd > 0, as well as, c'(k)-- (r-e )>0, l l e' which, for example, is is easily satisfied for small tt, or if the probability of the negative shock small. A. 4 Proof of We shall L = do Lemma this proof in steps. First, consider the 4- f h -t {l-n){Rk + w + h -w)-p {f -w) h - h -ip l (f -fj. h 9 (t -4>((1 - rk) - h - n)t h Lagrangian )+n^({R + r^)k-t l for (P7), {l l l l ip {t nt l - - rk) l e (d° $ -X(c(k)-nf -(l-n)f h L + - ir)4 + *ds )) l (1 -4) Then, R fir 4- r— ^r = (l-n)R + ir^^--\c>(k) = Hence, 1 / JJ 29 + r4 + ^) + {A + l e ){w + l ds -f 1 )) h Let us start with f ^ where the f h < w, then last inequality follows f h Consider t h next. = /i h = 0. Suppose ^|(l- = -(l-^) + A(l- 7r) = profitable. Hence, to If . 7r)i? so, + 7r^±^-c'(^>0 from the assumption that date investment is sufficiently w. If t h < rk then tp h = First note that, 0. from the FOC with respect d<jj, Then, # = - -)-*|1 Hence we can conclude that ip h | =?i(t -)B+ '£Ti > -H 1 and t h = rk. Take d$ next. Assuming the constraint does not bind, —= Therefore it must be Finally consider dj t'. (!-„) + = e '(i - tt) = (1 - T) - A(l - tt) < 0. First note that H = £(A + e') from the - vr^e' FOC for d$, = |(A + e <) - ttA = 0. Therefore, A + e< . Now, ~(A + e') <9i' Hence the firm is 2 V 4>ix = - J(A + 2 indifferent over its choice of r. 30 V e l ) + -ttt = 0. 0. >0 - References [1] Aghion, P., P. Bacchetta and A. Banerjee, "Currency Crises and Monetary Policy in an Economy with Credit Constraints," mimeo MIT, March 2000. [2] and D. Gale, "Limited Market Participation and Allen, F. American Economic Review [3] September 1994, pp. 933-955. 84-4> and D. Gale, "Optimal Currency Allen, F. Volatility of Asset Prices," Crises," mimeo, Wharton School, March 2000. [4] Burnside, C, M. Eichenbaum, and Rebelo, "Hedging and Financial Fragility in S. Fixed Exchange Rate Regimes," mimeo Northwestern University, 1999. [5] Caballero, R.J. and A. Krishnamurthy, "Emerging Market Crises: Perspective," [6] MIT Caballero, R.J. and A. Krishnamurthy, "International Liquidity NBER WP Calvo, G.A., Money, Exchange Rates, and Output; 1996, Chapter [8] Management: #7740, MIT May Steril- 2000. Press, Cambridge, MA, 12. Calvo, G.A., "Capital Market And The Exchange mimeo Dollarization Debate in Latin America." [9] Asset Markets mimeo, November 1999. ization Policy with Illiquid Financial Markets," [7] An Rate: With Special Reference to the University of Maryland, April 2000 Calvo, G.A. and P. Guidotti, "Indexation and Maturity of Government Bonds: An Exploratory Model," in R. Dornbusch and M. Draghi (editors) Capital Markets and Debt Management, 1990, [10] York: Cambridge University Press: 52-82. Chang, R. and A. Velasco, "Liquidity Crises in NBER MIT [11] New in Emerging Markets: Theory and Macroeconomics Annual 1999, edited by B.S. Bernanke and J.J. Policy," Rotemberg, Press, Cambridge, Massachusetts, pp. 11-57. Dooley, M., "A Model of Crises in Emerging Markets," NBER WP# 6300, December 1997. [12] Froot, K., and P. O'Connell, "On the Pricing Application to Catastrophe Insurance," [13] Froot, K., D. Scharfstein and J. of Intermediated Risks: NBER Working Stein, "Risk Theory and Paper #6011, 1997. Management: Coordinating Corporate Investment and Financing Policies," Journal of Finance, 48(5), 12/93, pp. 1629-58. [14] Hausmann, R., U. Panizza, IADB mimeo, November and E. Stein, 1999. 31 "Why do countries float the way the float?," [15] Holmstrom, B. and Political [16] Economy J. Tirole, "Private and Public Supply of Liquidity," Journal of 106-1, February 1998, pp. 1-40. Krishnamurthy, A., "Collateral Constraints and the Amplification Mechanism," mimeo Northwestern University, April 2000. [17] Krugman, MIT, [18] P., "Balance Sheets, the Transfer Problem, and Financial Crises," mimeo 1999. Xiong, W., "Imperfect Arbitrage with Wealth Effects," vember 1999. 32 52 3G 038 mimeo Duke University, No- Date Due Lib-26-67 MIT LIBRARIES 3 9080 01918 3620 idna saiuvuan nv\i P 1 — -> ==- OREGON a— ! J -3 "jWfc^aPWlL.^El: RULE CO. ! -^ .:; •1 3; . ...; " '& 7 ! - : , • . . . "=] 3 s ... ,. ''• '" '" $ 'i = , = -. 1 , s S =T 5 U.S.A. I = 2 I "^= J E & i . : : . : :ri*ii\ l i 3J . s s - * > i > : . i : 1- 4 '' - ! , • ' J 5 s 1 ' >v .. V"..-.. ' HS, .' i : 3f ..... ; 1 ..1 • - ... ' i . ~~ ,. • . fc 5 ii ? : i ' - HUH »V3 | . | . A i ii . :: ' " ' . • ... . M'lrpl II l|l|T|l|l|l|i|ir|M' 8 : . : - 7 , ; 8 9 10 ; . .. _ -..- . ... . llmlllllllllllillllllllltllf Illllllllllilllllllllllllllllllllllllll llllllllllllllllll \\\\\\\\\ 70 . "... '" "' '. -f 1 I ; . ' '.. ; . . .. t • • . ' . : t : : ; i .' . , : , ,. 1; s ; : -, ; .; . ! :-. . s . .