Information asymmetry and foreign currency borrowing by small firms.

Brown, Martin ; Ongena, Steven ; Yesin, Pinar 等

INTRODUCTION

A large proportion of corporations in many countries have traditionally been borrowing in a foreign currency. (1) More recently and before the financial crisis also, many retail clients, that is, households and small firms, in transition countries have taken out foreign currency loans. In countries such as Latvia, Lithuania, Hungary, and Bulgaria, for example, retail clients now hold a similar or larger share of their loans in foreign currency than do corporations (European Central Bank, 2010). These retail loans in foreign currency are popularly believed to be 'small men's carry trades', that is, loans in which households and entrepreneurs seek lower interest rates and take unhedged exchange rate risk upon themselves (see Sorsa et al., 2007 and Beer et al., 2010). (2)

Empirical work, due to a lack of micro data, has yet to investigate in detail the drivers for the rapid expansion of foreign currency mortgages and car loans to households in transition countries in the decade before the financial crisis. In Brown et al. (2011), we document, however, that foreign currency borrowing by small firms in transition countries is strongly related to their foreign currency revenues and age, an oft-used proxy for the availability of public information about the firm, than it is to between-country interest rate differentials. Degryse et al. (2012) show that foreign banks that enter via greenfield investment, and hence face an acute information asymmetry about potential borrowers, lend more in foreign currency than do foreign banks that take over existing local banks (that have established relationships with local firms). (3) Brown and De Haas (2012) confirm that foreign-owned banks lend more in foreign currency to corporate clients but not to households than do domestic banks. This evidence suggests that information asymmetries may actually limit the provision of foreign currency loans to the more financially opaque retail clients.

Hence, an information asymmetry between banks and firms may be a key determinant of the demand and supply of foreign currency loans. Our paper fills a gap in the theoretical literature by introducing information asymmetry in a framework that also features a trade-off between the cost and risk of firm debt. Assuming an interest rate differential in favor of foreign currency funds, we compare the currency structure of borrowing by small firms under two information conditions. Under perfect information, banks can verify the currency in which the firms contract their sales. Under imperfect information, the banks cannot verify the currency structure of firms' revenues.

Our model confirms that under perfect information all foreign currency earners and all local currency earners with low distress costs will choose foreign currency loans. By contrast, local currency earning firms with high distress costs will prefer local currency loans. Under imperfect information, more local earners will borrow in foreign currency, as these firms do not bear the full cost of the corresponding default risk. The intuition behind this result is the following: as the banks cannot distinguish the risky from the safe FX borrowers, they will charge a single interest rate to both types. This interest rate will be lower than the interest rate that prevails when the banks can identify the risky firms. Thus, it will be profitable for the risky firms to pool with safe firms and pay a lower risk premium for a foreign currency loan.

Consequently, our model identifies the information asymmetry between lending banks and borrowing firms as a so-far overlooked potential driver of dollarization or euroization in credit markets. We establish the conditions under which all firms will be borrowing in foreign currency (full pooling equilibria), as well as when no foreign currency loans will be offered by banks to firms that cannot prove they have either high or foreign currency income (market failure).

The key predictions of our model are consistent with suggestive evidence in Brown et al. (2011) in which we document that foreign currency borrowing by small firms in transition countries significantly decreases in their age (though not in two alternative yet crude proxies for the existing bank-firm information asymmetry, ie, whether the firm was audited or received income through a bank account). Our result is also consistent with the recent evidence in Degryse et al. (2012) who find that foreign banks that enter via greenfield investment, and that may face more information asymmetry than those foreign banks that enter via domestic takeovers, lend more in foreign currency.

The rest of the paper is organized as follows. The literature is reviewed in the next section. Then our model assumptions are introduced, followed by the analysis of the model with perfect information. In the last-but-one section, imperfect information is introduced. The last section concludes.

LITERATURE REVIEW

Though a number of recent theoretical papers have started to model the choice of loan currency in a way that may also be relevant for small firms (Allayannis et al., 2003), (4) a theoretical framework to understand foreign currency borrowing in retail credit markets where informational asymmetries are acute is still lacking (see also the review in Nagy et al., 2011). Yet, information asymmetries between banks and firms underpin our modern understanding of financial intermediation (Freixas and Rochet, 2008), and the asymmetries may be aggravated in transition and developing countries where, due to the weak corporate legal system, it is hard for banks to assess the credibility of available firm-level financial information (Pistor et al., 2000; Brown et ai., 2009a).

The costs of information acquisition by banks can be particularly high when dealing with small firms (eg, Berger and Udell, 1995; Berger and Udell, 2002; Degryse et al., 2011), which are less likely to have audited financial accounts. Depending on bank type, size or ownership and the degree of competition in the banking sector, banks may have difficulties or lack incentives to collect detailed information about current and especially expected future sales revenues (Berger and Udell, 2006). (5) In particular, the currency denomination of firms' sales contracts is often negotiated and a closely guarded secret. (6) Consequently, soft information about sales revenues and its currency denomination may be the only type of information that is available, but foreign banks that are widely present in transition and developing countries may struggle to collect and use it (Stein, 2002; Detragiache et al.; 2008).

Existing models demonstrate that firms' choice of loan currency is affected by the structure of firm revenues, interest rate differentials between local and foreign currency funds, and the distress costs of firms facing potential default. (7) Our theoretical model augments extant work by featuring not only the trade-off between the risk and the cost of debt, but also a relevant information asymmetry between banks and firms that can have either domestic or foreign currency earnings. This information asymmetry issue may be particularly relevant regarding small and young firms in transition and developing countries.

In Jeanne (2000), financiers also face an information asymmetry, but one concerning the effort level of the exporting entrepreneurs. Exporters borrow locally in domestic or foreign currency. But borrowing in a foreign currency serves as a commitment device: the entrepreneurs have a stronger incentive for effort if they have foreign currency debt because failure to achieve high returns is automatically sanctioned by termination. In Cowan (2006), firms with more foreign income and firms in countries with a higher interest differential (where foreign currency funds are cheaper) will have more foreign debt. Firms that are more financially constrained, that is, firms that experience a higher risk premium when borrowing from a bank, are more likely to match the denomination of debt to their income streams. These firms would have to borrow at higher costs if they become financially distressed due to the accumulated currency mismatches. If a bank knows a firm is mismatched, it may pass on the corresponding expected default costs.

In contrast to Jeanne (2000) where firms only have foreign revenues, in our model firms have domestic or foreign currency earnings. In Jeanne (2000), entrepreneurial effort is unobservable to the financiers; in our model, the currency and level of sales revenues cannot be observed by the bank.

MODEL ASSUMPTIONS

Define [e.sub.t], the exchange rate at time t, to equal the amount of local currency per unit of foreign currency, normalized at t = 0 to [e.sub.0] = 1. At t = 1, the local currency either appreciates to [e.sub.A] < 1, with probability p, or it depreciates to [e.sub.D] > 1, with probability 1-p.

We assume that

[pe.sub.A] + (1 - p)[e.sub.D] = 1, (1)

so that the expected exchange rate at t = 1 equals [e.sup.*.sub.1] = 1 and the expected depreciation of the local currency is [DELTA]e = [e.sup.*.sub.1]-[e.sub.0]/[e.sub.0] = 0. (8)

There is a continuum of firms and each firm needs to invest I = 1 in local currency at t = 0 to receive any revenues at t = 1. Firms differ in their revenue currency. There are two types of firms, foreign (Y) and local (L) currency earners. Foreign currency earners have revenue [R.sup.F] in foreign currency, which equals the expected revenue in local currency as the expected exchange rate equals 1 ([e.sup.*.sub.1] = 1, hence [R.sup.F][e.sup.*.sub.1] = [R.sup.F]). Local currency earners have earnings RL in local currency. We assume that [R.sup.L] < [R.sup.F] without loss of generality. We abstract from the possibility that local and/or foreign currency earners may differ in their debt-to-income levels because assuming more types of firms does not alter the main insights and results of our model. (9) Furthermore, we abstract from exchange rate pass-through considerations without loss of generality. (10)

Let both firm types be physically located in the domestic country. Their owners will spend their profits locally, so firms care about their expected payoff in local currency. Firms maximize their expected income and have no other wealth (and are thus limited liable).

There are at least two identical banks that offer loans in both local and foreign currency and that are engaged in Bertrand competition setting prices simultaneously. When they can identify firm type, they charge a net interest rate [r.sup.j.sub.k] on a loan in foreign or local currency k, k [member of] {f; l}, to a firm of type j [member of] {F, L}. (11) Banks have no capacity limits on foreign or local currency funds. We normalize the cost of foreign currency funds to [i.sub.f] = 0 and set the unit cost of local currency funds to [i.sub.l]. We assume that the uncovered interest rate parity (UIP) is not fulfilled, (12) and that there is an interest rate advantage to foreign currency funding for the bank, that is, [i.sub.l] > [i.sub.f] + [DELTA]e = 0. Extensive empirical research, using a variety of methods, finds that the UIP rarely holds. Furthermore, the literature finds that the deviation from the UIP in emerging markets is systernatic in nature and that a significant part of the excess return can be attributed to a risk premium. (13)

For simplicity, we assume that interest payments are made upfront at t = 0, and the loan repayment is made at t = 1. (14) Firms' earnings are verifiable ex post so that payments are enforceable if a firm has sufficient earnings.

We assume that the exchange rate volatility is such that local currency earners will default if they take a loan in foreign currency and the local currency depreciates, that is, [R.sup.L] < [e.sub.D]. We also assume that foreign currency earners have revenues that will enable them to fully repay a local currency loan even if the local currency appreciates, that is, [R.sup.F] > 1/[e.sub.A].

If firms default on a loan, they face costs of financial distress. For example, defaulters can henceforth find external financing only at penalty costs. In this case, the distress costs C may be proportional to or convex in the default amount (though still homogenous across firms). Alternatively, these costs may involve the private value to its owner of a firm that is lost in bankruptcy, eg, in the case of small and family-owned firms (Froot et al. 1993). (15) In this case, C will be independent of the default amount, but will be heterogeneous among firms. (16)

As the focus of our analysis is the information asymmetry between banks and small firms, we assume that distress costs in local currency units differ across firms. Among each type of firm j [member of] {F; L}, there is a share [phi] with low costs [C.bar] and a share 1-[phi] with high distress costs [bar.C]. (17)

While we assume that firms maximize expected income, their payoff is not linear in expected income when we assume distress costs. The assumption of distress costs implies that firms care about income variance, as would be the case if we assumed firms were risk-averse.

Given the above assumptions, the expected payoff [v.sup.i,j.sub.k] in local currency to a firm of type j [member of] {F; L} with a distress cost [C.sub.i] [member of] {[C.bar]; [bar.C]} taking a loan of type k [member of] {l; f} equals:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (2)

PERFECT INFORMATION CASE

When banks are perfectly informed about the type of each firm, each bank sets four interest rates. For each of the two firm types, j [member of] {F; L}, they set two interest rates, depending on whether a foreign or local currency loan is offered.

The expected profits of banks in local currency from each loan type are:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (3)

Note that the term in equation 3 already uses two assumptions: (1) interest rate payments are made up front, thus the bank always receives them even in the case of a default, and (2) if an L firm, which took a foreign currency loan, defaults due to a depreciation of the local currency then the bank receives all of its [R.sup.L].

Assuming perfect price competition in the banking sector, the expected profit on each loan type will be 0. Given our assumptions that [i.sub.f] = 0, and that (1) holds, this leads to the following equilibrium interest rates:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (4)

elegant, yet does not alter the main intuition that imperfect information leads to more foreign currency borrowing.

Note that the term ([e.sub.D] - [R.sup.L]) in equation 4 is the amount of the loan that a defaulting L firm does not pay back. Furthermore, the term (1 - p) is the probability of its default.

Proposition 1: Under perfect information, all F firrns take foreign currency loans. The equilibrium share of L firms that choose foreign currency loans is given as:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (5)

Proof. In the Appendix.

Proposition 1 shows that, under perfect information, foreign currency earners (F types) always choose foreign currency loans. They do so because there is an interest rate advantage to foreign currency loans and they do not run the risk of incurring distress costs when taking such a loan. On the other hand, local currency eaming firms (L types) face a trade-off: if they choose a foreign currency loan, they benefit from an interest rate advantage, but they may incur distress costs if the local currency depreciates. As a consequence, if the interest rate differential is low compared with the minimum distress costs of firms, that is, when [i.sub.l] < (1 -p)[C.bar], we have a 'separating' equilibrium in which all L types, that is, firms with local currency revenue, take local currency loans. On the other hand, if the interest rate differential is high, that is, when [i.sub.l] > (1 - p)[bar.C], we have a pooling equilibrium in which all firms take foreign currency loans. For intermediate values of interest rate differentials, we have a 'partial pooling' equilibrium in which L firms with low distress costs take foreign currency loans, and L firms with high distress costs take local currency loans.

IMPERFECT INFORMATION CASE

We now introduce an information asymmetry between banks and firms about the revenues of the firms. Assume that banks cannot verify the currency denomination of a firm, that is, banks cannot distinguish between the two types of firms: F and L firms.

Banks, however, know that among each type of firm j [member of] {F;L} there is a share [phi] with low costs [C.bar] costs and a share 1-[phi] with high distress costs [bar.C]. Furthermore, banks know that a proportion [lambda][member of][0,1] of the total firm population is L firms, and that the remaining proportion 1 - [lambda] is F firms. Banks can no longer condition their interest rates on firm types, and thus only offer two rates: [r.sub.l] for local currency loans and [r.sub.f] for foreign currency loans. (18)

In this case, the expected profit of banks in local currency from the two loan types is:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (6)

where [delta][member of][0, 1] is the equilibrium share of L firms taking foreign currency loans. Given the assumption of equation 1, interest rates in equilibrium with 0 expected profit must equal:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (7)

The interest rate charged on foreign currency loans covers the expected losses due to default on such loans. Under imperfect information, this depends on the share of L firms taking such loans relative to F firms. This characterization of the interest rates implicitly uses a Nash equilibrium concept and is based on O profit conditions for the banks given the share of L firms taking a foreign currency loan.

Note that the expression 0 [less than or equal to] [delta][lambda]/[delta][lambda] + (1-[lambda]) [less than or equal to] 1 when 0 [less than or equal to] [delta] [less than or equal to] 1. Therefore, the interest rate that banks charge on foreign currency loans under imperfect information lies between the rate they charge for such loans under perfect information to F firms, that is, 0, and the rate they charge under perfect information to L firms, that is, [r.sup.L.sub.f] = (l-p) ([e.sub.D] - [R.sup.L]). In other words, [r.sub.f] [member of] [0, [r.sup.L.sub.f]].

A local currency earning firm will choose to take a foreign currency loan if its expected payoff is higher when it takes a foreign currency loan than when it takes a local currency loan, that is, when [v.sup.L.sub.f] ([r.sub.f], [C.sub.i]) [greater than or equal to] [v.sup.L.sub.l]([r.sub.l], [C.sub.i]).

Substituting that [r.sub.t] = [i.sub.l] in (2) and making use of equation 1, we see that a local currency earning firm will choose a foreign currency loan only if

(1 - p)([C.sub.i] + [R.sup.L] - [e.sub.D]) [less than or equal to] [i.sub.l] - [r.sub.f]. (8)

In other words, a local currency earning firm will choose a foreign currency loan if its expected cost of default on the foreign currency loan is less than the interest rate advantage he will get over a local currency loan.

We assume from now on that:

[C.bar] [greater than or equal to] [e.sub.D] - [R.sup.L] > 0. (9)

Assumption (9) ensures that, unless there is a positive interest rate differential to the advantage of foreign currency funds, all L firms will choose local currency loans. This assumption prevents some L firms opting for foreign currency loans due to their limited liability even in the absence of an interest rate differential. Note that assumption (9) states that the minimum distress costs is at least as large as the part of the foreign currency loan that a defaulting L firm does not pay back in the case of a default.

From (8), we see that under imperfect information the share of L firms, which will take foreign currency loans, will be:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (10)

Note that the conditions under which firms choose to borrow in local or foreign currency involve the expected cost of default on a foreign currency loan. The expected cost of default would be the distress cost, C, minus the part of the foreign currency loan not paid back ([e.sub.D] - [R.sup.L]), in other words, a 'gain' for the firm as it does not pay back its full loan.

The following propositions show when different types of equilibria emerge under imperfect information given the values of the parameters in our model.

Proposition 2 (Separating Equilibrium): Ir [i.sub.l] <(1-p)([C.bar] + [R.sup.L] - [e.sub.D]), then a separating equilibrium will emerge.

Proof. In the Appendix.

Above, we see that none of the local currency earners will choose to borrow in foreign currency if the interest rate disadvantage on local currency loans is smaller than the minimum expected cost of default when a foreign currency loan is taken.

Proposition 2 also shows that a separating equilibrium under imperfect information exists only for lower interest rate differential between local and foreign currency loans compared with the minimum interest rate differential needed for a separating equilibrium under perfect information. In order to see that, compare the first lines of (5) and (10) and utilize assumption (9).

Proposition 3 (Partial Pooling Equilibrium): If [i.sub.l] [greater than or equal to] (1-p) [bar.C]-(1 - [lambda])/[lambda][phi]+(1-[lambda])(1-p)([e.sub.D] - [R.sup.L]) and [i.sub.l] [less than or equal to] (1 - [lambda])[bar.C] - (1 - [lambda])/[lambda][phi] + (1 - [lambda])(1 - p)([e.sub.D] - [R.sup.L]) a partial pooling equilibrium exists in which only L firms with low distress costs [C.bar] take foreign currency loans while L firms with high distress costs [bar.C] take local currency loans.

Proof. In the Appendix.

Proposition 3 similarly shows that a partial pooling equilibrium is feasible under imperfect information starting at a lower interest rate differential between local and foreign currency loans than it is feasible under perfect information.

Proposition 4 (Full Pooling Equilibrium): If [i.sub.l] [less than or equal to] (1-p)[bar.C]-(1-p)(1-[lambda]) ([e.sub.D] - [R.sup.L]) a full pooling equilibrium exists in which all L firms take foreign currency loans.

Proof. In the Appendix.

Last but not the least, proposition 4 also shows that a fully pooling equilibrium under imperfect information is more feasible for a lower interest rate differential between local and foreign currency loans than it is under perfect information.

Note that in the partial-pooling and full-pooling equilibria described above we have assumed that all F firms chose foreign currency loans, which will be the case as long as [r.sub.f] ([delta]) [less than or equal to] [i.sub.l]. Assumption (9) ensures that in any equilibrium where [delta] > 0 we have [r.sub.f]([delta]) [less than or equal to] [i.sub.l].

To summarize our findings, under perfect information there always exists either a separating, partial-pooling, or full-pooling equilibrium. Under imperfect information, two main things change. First, partial-pooling or full-pooling equilibria exist at a larger range of interest rate differentials than under perfect information. This is due to the fact that foreign currency loans to L firms are not fully pricing the credit risk of these loans due to expected exchange rate depreciations. And second, the market for foreign currency loans may in fact collapse. Proposition 5 summarizes the range of parameters for which an equilibrium with lending in foreign currency does not exist.

Proposition 5 (Market Failure Imperfect Information): Under imperfect information, there is no equilibrium in which foreign currency loans are extended if one of the following two conditions is met:

(1 - p)[C.bar] - (1 - p) ([e.sub.D] - [R.sup.L]) < [i.sub.l] < (1 - p) [C.bar] - [[(1-[lambda])]/[[lambda][phi]+(1-[lambda])]] (1-p)([e.sub.D] - [R.sup.L])] (11)

or

(1 - 2)[C.bar] - [[(1-[lambda])]/[[[lambda][phi]+(1-[lambda])]] (1-p)([e.sub.D] - [R.sup.L]) < [i.sub.l] < (1 - p) [C.bar] - (1 - [lambda])(1 - p)([e.sub.D] - [R.sup.L]) (12)

Proof. In the Appendix.

Proposition 5 shows that there are two constellations under which the market for foreign currency loans may collapse with imperfect information. The first constellation is a range of interest rate differentials (11) at which L firms with low distress consider switching from local currency to foreign currency loans if banks charge 0 interest rates on foreign currency loans. However, if a set of [phi][lambda] of L firms with low distress costs would switch to foreign currency loans, the zero-profit interest rate on these loans would rise to (1-[lambda])/[phi][lambda] + (1-2)(1-p)([e.sub.D] - [R.sup.L]). At that interest rate, for foreign currency loans, all L firms will prefer to take local currency loans, and thus there is no equilibrium in which foreign currency loans are offered.

A similar effect leads to market collapse at interest rates in the range (12), to the extent that these firms would be deterred from taking foreign currency loans, A set of (1-[phi])[lambda] of L firms with high distress costs consider switching from local currency to foreign currency loans.

In both of these ranges of interest rate differentials the only credit market equilibrium is characterized by all firms taking local currency loans.

It should be noted that throughout the model we assume that there are no alternative financing options for foreign currency loans. Thus, foreign currency earners do not have any option other than subsidizing foreign currency loans for local currency earners. However, the more local currency earners borrow in foreign currency, the more expensive will a foreign currency loan be. In that case, a foreign currency earner can possibly opt out and find an alternative financing source, such as trade credit or capital market financing.

CONCLUSION AND TESTABLE HYPOTHESES

Motivated by policy concerns about the credit risks resulting from unhedged foreign currency loans, we investigate how an information asymmetry between banks and firms in a theoretical framework, that also features the trade-off between the cost and the risk of debt, may determine the currency denomination of bank loans to firms. Banks in our model may not know the currency in which firms have contracted their sales.

Our model shows that under imperfect information concerning the currency of firm revenues more local currency earners choose foreign currency loans than when banks have perfect information about firms' revenues. The intuition behind this result is that when banks cannot distinguish the risky from the safe FX borrowers they will charge a single interest rate to both types. This interest rate will be lower than the interest rate for risky borrows (ie, local currency earners) that prevails when the banks can identify these firms.

In line with previous models (eg, Cowan, 2006), our model predicts that the currency structure of firm revenues as well as firm-level distress costs will affect loan currency choice. At the macroeconomic level, our model further confirms that the choice of a foreign currency loan will be positively related to the interest rate advantage on foreign currency funds and negatively related to exchange rate volatility.

The novel prediction of our model is that the choice of a foreign currency loan by local currency earners may be positively related to the opaqueness of the firm's revenue structure. More local currency earners choose foreign currency loans under imperfect information than under perfect information. The impact of information opaqueness is stronger for firms with higher shares of revenue in local currency (our model suggests that imperfect information does not alter the currency choice for firms with foreign currency earnings only). Our model suggests that characteristics of the banking sector or of the legal environment that exacerbate information asymmetries between banks and firms may foster unhedged foreign currency borrowing. Finally, our model predicts that under information asymmetry the foreign currency loan market can collapse if too many local currency earners prefer to take foreign currency loans making them too expensive. Whether the market will collapse or not depends on the interest rate differential and the distribution of distress costs in the economy.

How do the novel predictions of our model square with recent firm-level and bank-level evidence? In Brown et al. (2011), we examine the determinants of foreign currency borrowing by small firms, using information on the most recent loan extended to 3,101 firms in 25 transition countries between 2002 and 2005. There we show that firms' revenue structure rather than the interest rate differential between local currency and foreign currency funding is the key determinant of loan currency choice. That evidence confirms the fact that firm characteristics as opposed to monetary conditions seem to be the key drivers of foreign currency borrowing by small firms in Emerging Europe. With respect to the importance of information asymmetries as a driver of foreign currency loans, we find mixed evidence: Borrowing in foreign currency is unrelated to firm-level auditing procedures. At the same time, borrowing in foreign currency is negatively related to firm age suggesting that more opaque (younger) firms are more likely to borrow in foreign currency.

Degryse et al. (2012) as well as Brown and De Haas (2012) use bank-level data to study how the ownership structure of banks affects the currency denomination of their loan portfolio. Examining regulatory data for Polish banks during the period 1996-2006, Degryse et al. (2012) find that foreign banks that enter via greenfield investment, and hence face an acute information asymmetry about potential borrowers, are more likely to lend to larger, financially transparent firms than domestic-owned banks or foreign banks that take over existing local banks. They also find that greenfield banks are more likely to lend in foreign currency than domestic or takeover banks. These results suggest that information asymmetries play an important role in lending to small businesses in the region and may actually limit the provision of foreign currency loans to this segment of financially opaque borrowers. Brown and De Haas (2012) examine survey data covering 193 banks in 20 countries in 2001 and 2004. Their results confirm that foreign-owned banks lend more in foreign currency to corporate clients (but not to households) than domestic banks. Again, this evidence suggests that information asymmetries (that are arguably stronger for foreign-owned banks) may limit the provision of foreign currency loans to the more financially opaque retail clients.

APPENDIX

Proposition 1: Under perfect information, all F firms take foreign currency loans. The equilibrium share of L firms that choose foreign currency loans is given as:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (A.1)

Proof. Recall that the equilibrium interest rates on loans under perfect information can be written as:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (A.2)

Recall also that the expected payoff of firms can be written as:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (A.3)

and that the expected depreciation of the local currency is assumed to be 0:

[pe.sub.A] + (1 - p)[e.sub.D] = 1. (A.4)

Inserting the equilibrium interest rates from (A.2) into the expected payoff of firms (A.3), and using the equation (A.4), we obtain the following two results:

1. Foreign currency earners (F types) will always choose foreign currency loans, because their expected payoff will be higher when they take a foreign currency loan than when they take a local currency loan. Thus, all F firms will take foreign currency loans.

2. An L firm will choose a local currency loan when

(1 - p)[C.sub.i] [greater than or equal to] [i.sub.l]. (A.5)

The condition (A.5) tells us when it will be preferable for a local currency earner to borrow in foreign currency based on the values of the distress costs, the probability of depreciation, and the interest rate gap. In other words, a local currency earner will choose to take a local currency loan if its expected cost of default on a foreign currency loan is larger than the interest rate on local currency loans.

Recall that we assumed only two values for [C.sub.i] [member of] {[C.bar];[bar.C]}. Thus if [bar.C]< [i.sub.l]/(1-p), then all local currency earners will choose a foreign currency loan; and if [C.bar] > [i.sub.l]/(1 - p), then no local currency earner will choose a foreign currency loan. For intermediate values of [i.sub.l], only local currency earners with low distress costs, [C.bar], will choose a local currency loan. Thus, the equilibrium share of L firms that choose foreign currency loans can be written as:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (A.6)

Note that (A.6) is equivalent to (A.1).

Proposition 2 (Separating Equilibrium): If [i.sub.l] < (1 - p)([C.bar] + [R.sup.L] - [e.sub.D]), then a separating equilibrium will emerge.

Proof. Ina separating equilibrium, all local currency earners will choose a local currency loan by definition. Thus, we have the share of L firms taking a foreign currency loan equal to 0, that is, [delta] = 0. Recall that the equilibrium interest rate for foreign currency loans can be written as:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (A.7)

Therefore, [r.sub.f] = 0 when [delta] = 0.

Also recall that

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]. (A.8)

From (A.8) it follows that a separating equilibrium exists, if [i.sub.l] < (1 - p)([C.bar] - [e.sub.D] + [R.sup.L]).

Proposition 3 (Partial Pooling Equilibrium): If [i.sub.l] [less than or equal to] (1 - p) [C.bar] - (1 - [lambda])/[lambda][phi] + (1 - [lambda])(1 - p)([e.sub.D] - [R.sup.L]) and [i.sub.l] [less than or equal to] (1 -p)[bar.C] (1 - [lambda])/[lambda][phi] + (1 - [lambda]) (1 - p)([e.sub.D] - [R.sup.L) a partial pooling equilibrium exists in which only L firms with low distress costs [C.bar] take foreign currency loans while L firms with high distress costs [bar.C] take local currency loans.

Proof. In a partial pooling equilibrium, some local currency earners will choose a local currency loan, whereas others will choose a foreign currency loan. Recall that distress costs can take only two values, [bar.C] and [bar.C]. Therefore, in a partial pooling equilibrium, the share of L firms that take a foreign currency loan should equal to the share of L firms that have low distress costs. Hence, [delta] = [phi].

Recall that the equilibrium interest rate for foreign currency loans can be written as in (A.7). Therefore,

[r.sub.f] = [[[phi][lambda]/[[phi][lambda] + (1 - [lambda])]] (1 - p)([e.sub.D] - [R.sup.L]). (A.9)

Substituting (A.9) into (A.8), it follows that only L firms with low distress costs will chose a foreign currency loan if:

[i.sub.l] [greater than or equal to] (1 - p)[C.bar] - [(1-[lambda])/[[phi][lambda] + (1 - [lambda])]] (1 - p)([e.sub.D] - [R.sup.L])

and

[i.sub.l] < (1 - p)[bar.C] - [(1 - [lambda])/[[phi][lambda] + (1 - [lambda])]] (1 - p) ([e.sub.D] - [R.sup.L])

Proposition 4 (Full Pooling Equilibrium): If [i.sub.l] [less than or equal to] (1 - p)[bar.C] - (1 - p)(1 - [lambda]) ([e.sub.D] - [R.sup.L]) a full pooling equilibrium exists in which all L firms take foreign currency loans.

Proof. Ina full-pooling equilibrium, all local currency earning firms take foreign currency loans, that is, [delta] = 1.

In this case, the expression (A.7) yields that the equilibrium interest rate for foreign currency loans is

[r.sub.f] = [lambda](1 - p)([e.sub.D] - [R.sup.L]). (A.10)

Substituting (A.10) into (A.8), it follows that a full-pooling equilibrium exists if [i.sub.l] [less than or equal to] (1 - p)[bar.C] - (1 - [lambda])(1 - p)([e.sub.D] - [R.sup.L]).

Proposition 5 (Market Failure Imperfect Information): Under imperfect information, there is no equilibrium in which foreign currency loans are extended if one of the following two conditions is met:

(1 - p)[C.bar] - (1 - p)([e.sub.D] - [R.sup.L]) < [i.sub.l] < (1 - p) [C.bar] - [(1 - [lambda])/[[lambda][phi]+(1 - [lambda])]] (1 - p)([e.sub.D] - [R.sup.L]). (A.11)

or

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (A.12)

Proof. Follows directly from propositions 2, 3, and 4.

Acknowledgements

We thank an anonymous referee for Comparative Economic Studies and an anonymous referee for the Swiss National Bank Working Paper Series, Raphael Auer, Sohnke Bartram, Henrique Basso, Katalin Bodnar, Josef Brada (the editor), Geraldo Cerqueiro, Andreas Fischer, Davide Furceri, Luigi Guiso, Werner Hermann, Herman Kamil, Anton Korinek, Marcel Peter, Alexander Popov, Maria Rueda Mauer, Philip Saure, Linus Siming, Clas Wihlborg and seminar participants at the University of Amsterdam, the Federal Reserve Bank of Boston, the University of Zurich, the Swiss National Bank, as well as participants at the European Finance Association Meeting (Bergen), the SNB-CEPR conference on 'Foreign Currency Risk Taking by Financial Institutions, Firms and Households' (Zurich), the Financial Intermediation Research Society Meeting (Prague), the CEPR/Studienzentrum Gerzensee European Summer Symposium in Financial Markets (Gerzensee), the European Economic Association Meetings (Milano), the CREDIT Conference (Venice), ESCE Meetings (Paris), the NBP-SNB Joint Seminar on 'Challenges for Central Banks during the Current Global Crisis' and the Tor Vergata Banking and Finance Conference (Rome) for useful comments and discussions. Ongena gratefully acknowledges the hospitality of the Swiss National Bank. Any views expressed are those of the authors and do not necessarily reflect those of the Swiss National Bank.

REFERENCES

Allayannis, G, Brown, GW and Klapper, LF. 2003: Capital structure and financial risk: Evidence from foreign debt use in East Asia. Journal of Finance 58(6): 2667-2709.

Allayannis, G and Ofek, E. 2001: Exchange rate exposure, hedging, and the use of foreign currency derivatives. Journal of International Money and Finance 20(2): 273-296.

Alper, E, Ardic, OP and Fendoglu, S. 2009: The economics of the uncovered interest parity condition for emerging markets. Journal of Economic Surveys 25(1): 115-138.

Andrade, G and Kaplan, SN. 1998: How costly is financial (not economic) distress? Evidence from highly leveraged transactions that became distressed. Journal of Finance 53 (5): 1445-1493.

Basso, HS, Calvo-Gonzalez, O and Jurgilas, M. 2011: Financial dollarization: The role of foreign-owned banks and interest rates. Journal of Banking and Finance 35(4): 794-806.

Beer, C, Ongena, S and Peter, M. 2010: Borrowing in foreign currency: Austrian households as carry traders. Journal of Banking and Finance 34(9): 2198-2211.

Benavente, JM, Johnson, CA and Morande, FG. 2003: Debt composition and balance sheet effects of exchange rate depreciations: A firm-level analysis for Chile. Emerging Markets Review 4 (4): 397-416.

Berger, AN and Udell, GE 1995: Relationship lending and lines of credit in small firm finance. Journal of Business 68(3): 351-381.

Berger, AN and Udell, GF. 2002: Small business credit availability and relationship lending: The importance of bank organisational structure. Economic Journal 112(477): 32-53.

Berger, AN and Udell, GF. 2006: A more complete conceptual framework for SME finance. Journal of Banking and Finance 30(11): 2945-2966.

Bodnar, GM, Hayt, GS and Marston, RC. 1998: Wharton survey of financial risk management by us nonfinancial firms. Financial Management 27(4): 70-91.

Booth, LV, Aivazian, A, Demirguc-Kunt, A and Maksimovic, V. 2001: Capital structures in developing countries. Journal of Finance 56 (1): 87-130.

Borsum, OG and Odegaard, BA. 2005: Currency hedging in Norwegian non-financial firms. Norges Bank Economic Bulletin 9(3): 133-144.

Briggs, P. 2004: Currency hedging by exporters and importers. Reserve Bank of New Zealand Bulletin 67(4): 17-27.

Brown, GW. 2001: Managing foreign exchange risk with derivatives. Journal of Financial Economics 60(2): 401-448.

Brown, M and De Haas, R. 2012: Foreign banks and foreign currency lending in emerging Europe. Economic Policy 27(69): 57-98.

Brown, M, Jappelli, T and Pagano, M. 2009a: Information sharing and credit: Firm-level evidence from transition countries. Journal of Financial Intermediation 18 (2): 151-172.

Brown, M, Kirschenmann, K and Ongena, S. 2012: Bank funding, securitization and loan terms: Evidence from foreign currency lending. Center for Wealth & Risk Working Paper No. 170, University of St. Gallen: St. Gallen.

Brown, M, Ongena, S and Yesin, P. 2009b: Foreign currency borrowing by small firms. Swiss National Bank Working Paper No. 2, Swiss National Bank: Zurich.

Brown, M, Ongena, S and Yesin, P. 2011: Foreign currency borrowing by small firms in the transition economies. Journal of Financial Intermediation 20(3): 285-302.

Calvo, GA. 2001: Capital markets and the exchange rate with special reference to the dollarization debate in Latin America. Journal of Money, Credit, and Banking 33(2): 312-334.

Cetorelli, N and Goldberg, LS. 2011: Global banks and international shock transmission: Evidence from the crisis, International Monetary Fund Economic Review 59(1): 41-76.

Claessens, S, Djankov, SD and Lang, LHP. 2000: The separation of ownership and control in East Asian corporations. Journal of Financial Economics 58(1): 81-112.

Clark, E and Judge, A. 2008: The determinants of foreign currency hedging: Does foreign currency debt induce a bias? European Financial Management 14(3): 445-469.

Cowan, K. 2006: Firm level determinants of dollar debt? Mimeo, Central Bank of Chile: Santiago.

Cowan, K, Hansen, E and Herrera, LO. 2005: Currency mismatches in non-financial firms in Chile. Journal Economia Chilena-Chilean Economy 8(2): 57-82.

Degryse, H, Havrylchyk, O, Jurzyk, E and Kozak, S. 2012: Foreign bank entry, credit allocation and lending rates in emerging markets: Empirical evidence from Poland. Journal of Banking and Finance 36(11): 2949-2959.

Degryse, H, Masschelein, N and Mitchell, J. 2011: Staying, dropping, or switching: The impacts of bank mergers on small firms. Review of Financial Studies 24(4): 1102-1140.

Detragiache, E, Tressel, Tand Gupta, P. 2008: Foreign banks in poor countries: Theory and evidence. Journal of Finance 63(5): 2123-2160.

Diamond, DW. 1984: Financial intermediation and delegated monitoring. Review of Economic-Studies 51(3): 395-414.

Diamond, DW. 1991: Monitoring and reputation: The choice between bank loans and privately placed debt. Journal of Political Economy 99(4): 689-721.

Dollar, D and Hallward-Driemeier, M. 2000: Crisis, adjustment, and reform in Thailand's industrial firms. World Bank Research Observer 15(1): 1-22.

Egli, D, Ongena, S and Smith, DC. 2006: On the sequencing of projects, reputation building, and relationship finance. Finance Research Letters 3 (1): 23-39.

Engel, C. 1996: The forward discount anomaly and the risk premium: A survey of recent evidence. Journal of Empirical Finance 3 (2): 123-192.

Engel, C. 2006: Equivalence results for optimal pass-through, optimal indexing to exchange rates, and optimal choice of currency for export pricing. Journal of the European Economic Association 4(6): 1249-1260.

Esho, N, Sharpe, IG and Webster, KH. 2007: Hedging and choice of currency denomination in international syndicated loan markets. Pacific-Basin Finance Journal 15 (2): 195-212.

European Central Bank. 2006: EU banking sector stability. European Central Bank: Frankfurt.

European Central Bank. 2010: European stability review. European Central Bank: Frankfurt.

Francis, BB, Hasan, I and Hunter, DM. 2002: Emerging market liberalization and the impact on uncovered interest rate parity. Journal of International Money and Finance 21 (6): 931-956.

Frankel, JA. 2004: Experience of and lessons from exchange rate regimes in emerging economies. In: Asian development bank (ed). Monetary and Financial Integration in East Asia: The Way Ahead. Palgrave Macmillan Press: New York.

Freixas, X and Rochet, JC. 2008: Microeconomics of banking. MIT Press: Cambridge, MA.

Friberg, R and Wilander, F. 2008: The currency denomination of exports--A questionnaire study. Journal of International Economics 75 (1): 54-69.

Froot, KA, Scharfstein, DS and Stein, JC. 1993: Risk management: Coordinating corporate investment and financing policies. Journal of Finance 48 (5): 1629-1658.

Froot, KA and Thaler, RH. 1990: Anomalies: Foreign exchange. Journal of Economic Perspectives 4(3): 179-192.

Galindo, A, Panizza, U and Schiantarelli, F. 2003: Debt composition and balance sheet effects of currency depreciation: A summary of the micro evidence. Emerging Markets Review 4(4): 330-339.

Gelos, GR. 2003: Foreign currency debt in emerging markets: Firm-level evidence from Mexico. Economics Letters 78(3): 323-327.

Goldberg, PK and Knetter, MM. 1997: Goods prices and exchange rates: What have we learned? Journal of Economic Literature 35(3): 1243-1272.

Gorodnichenko, Y and Schnitzer, M. 2010: Financial constraints and innovation: Why poor countries don't catch up? NBER Working Paper No. 15792, National Bureau of Economic Research: Cambridge, MA.

Goswami, G and Shrikhande, MM. 2001: Economic exposure and debt financing choice. Journal of Multinational Financial Management 11(1): 39-58.

Graham, JR and Harvey, CR. 2001: The theory and practice of corporate finance: Evidence from the field. Journal of Financial Economics 60(2): 187-243.

Hauswald, R and Marquez, R. 2006: Competition and strategic information acquisition in credit markets. Review of Financial Studies 19 (3): 967-1000.

Hodrick, R.J. 1987: The empirical evidence on the efficiency of forward and futures foreign exchange markets. Harwood Academic Publishers: Chur.

Jeanne, O. 1999: Foreign currency debt and signaling. Mimeo, International Monetary Fund: Washington DC.

Jeanne, O. 2000: Foreign currency debt and the global financial architecture. European Economic Review 44(4): 719-727.

Kamil, H. 2009: How do exchange rate regimes affect firms' incentives to hedge currency risk in emerging markets? Mimeo, International Monetary Fund: Washington DC.

Kamil, H and Sutton, B. 2008: Corporate vulnerability: Have firms reduced their exposure to currency risk. In: international monetary fund (ed). Regional Economic Oatlook: Western Hemisphere. International Monetary Fund: Washington DC.

Kedia, S and Mozumdar, A. 2003: Foreign currency-denominated debt: An empirical examination. Journal of Business 76(4): 521-546.

Keloharju, M and Niskanen, M. 2001: Why do firms raise foreign currency denominated debt? European Financial Management 7(4): 481-496.

Lewis, K. 1995: Chapter 37: Puzzles in international financial markets, In: Grossman, G and Rogoff K (eds). Handbook of International Economics, Vol. III. Elsevier: North-Holland.

Luca, A and Petrova, I. 2008: What drives credit dollarization in transition economies? Journal of Banking and Finance 32(5): 858-869.

Mian, SL. 1996: Evidence on corporate hedging policy. Journal of Financial and Quantitative Analysis 31(3): 419-439.

Nagy, PM, Jeffrey, S and Zettelmeyer, J. 2011: Addressing private sector currency mismatches in emerging Europe. In: Prasad, E and Kawai, M (eds). Financial Market Regulation and Reforms in Emerging Markets. Brookings Institution Press: Washington DC.

O'Connell, S. 2005: Currency game is risky for the smaller players. Sunday Times.

Pindado, J, Rodrigues, L and Torre, C. 2006: How does financial distress affect small firms' financial structure? Small Business Economics 26(4): 377-391.

Pistor, K, Raiser, M and Gelfer, S. 2000: Law and finance in transition economies. Economics of Transition 8(1): 325-368.

Purnanandam, A. 2008: Financial distress and corporate risk management: Theory and evidence. Journal of Financial Economics 87(3): 706-739.

Rajan, RG. 1992: Insiders and outsiders: The choice between informed and arm's-length debt. Journal of Finance 47(4): 1367-1400.

Rajan, RG and Zingales, L. 1995: What do we know about capital structure? Some evidence from international data. Journal of Finance 50(5): 1421-1460.

Rey, H. 2001: International trade and currency exchange. Review of Economic Studies 68(2): 443-464. Rosenberg, CB and Tirpak, M. 2009: Determinants of foreign currency borrowing in the new member states of the EU. Czech Journal of Economics and Finance 59(3): 216-228.

Ross, SA. 1977: The determination of financial structure: The incentive-signaling approach. Bell Journal of Economics 8(1): 1-30.

Senbet, LW and Seward, JK. 1995: Financial distress, bankruptcy and reorganization. In: Robert/LI, Maksimovid V and Ziemba WT (eds). Handbooks in Operations Research and Management Science. Elsevier: Amsterdam.

Sharpe, SA. 1990: Asymmetric information, bank lending and implicit contracts: A stylized model of customer relationships. Journal of Finance 45(4): 1069-1087.

Sorsa, P, Bakker, BB, Duenwald, C, Maechler, AM and Tiffin, A. 2007: Vulnerabilities in emerging southeastern Europe-How much cause for concern? IMF Working Paper No. 236, International Monetary Fund: Washington DC.

Stein, J. 2002: Information production and capital allocation: Decentralized versus hierarchical firms. Journal of Finance 57(5): 1891-1922.

Stulz, RM. 1984: Optimal hedging policies. Journal of Financial and Quantitative Analysis 19: 127-140. Townsend, RM. 1979: Optimal contracts and competitive markets with costly state verification. Journal of Economic Theory 21 (2): 265-293.

Tsamenyi, M and Skliarova, D. 2005: International cash management practices in a Russian multinational. Managerial Finance 31(10): 48-64.

Viaene, J-M and de Vries, CG. 1992: On the design of invoicing practises in international trade. Open Economies Review 3(2): 133-142.

von Thadden, E-L. 2004: Asymmetric information, bank lending, and implicit contracts: The winner's curse. Finance Research Letters 1 (1): 11-23.

Ziaul Hoque, M. 2003: Flawed public policies and industrial loan defaults: The case of Bangladesh. Managerial Finance 29(2): 98-121.

(1) In East Asia, corporate debt is split about equally between foreign and domestic currencies (Allayannis et al., 2003) while in several Latin American countries the share of foreign currency debt exceeds 20% (Galindo et al., 2003). Between 20% and 75% of all corporate loans in Eastern European countries are denominated in a foreign currency (European Central Bank, 2006, p. 39).

(2) Foreign currency loans create serious challenges to policymakers. Countries with high volumes of foreign currency loans are more vulnerable to financial crises and more prone to spillover effects of country-specific shocks (see, eg, Cetorelli and Goldberg, 2011). Furthermore, foreign currency-denominated loans distort the transmission of monetary policy, influence the available credit in the economy, and therefore can impact the catching-up process of transition countries (see, eg, Gorodnichenko and Schnitzer, 2010).

(3) in contrast to these two studies, others have examined foreign currency borrowing by analyzing aggregate cross-country data (eg, Luca and Petrova, 2008; Rosenberg and Tirpak, 2009; Basso et al., 2011) or the currency denomination of debt of large firms within a single country (Keloharju and Niskanen, 2001; Benavente et al., 2003; Gelos, 2003; Kedia and Mozumdar, 2003; Cowan et al., 2005) or across countries (Rajan and Zingales, 1995; Booth et al., 2001; Allayannis et al., 2003; Cowan, 2006; Esho et al., 2007; Kamil and Sutton, 2008; and Kamil, 2009). Clark and Judge (2008) provide a review of the relevant empirical literature.

(4) We will not discuss: (1) International taxation issues such as tax loss carry forwards and limitations on foreign tax credits; (2) The possibilities for international income shifting; (3) The differential costs across countries of derivatives to create synthetic local debt; and (4) Clientele effects in issuing public bonds. These issues are clearly important when analyzing the debt structure of large corporations.

(5) See, for example, Dollar and Hallward-Driemeier, (2000). In addition, banks often cannot verify firm sales information through advanced cash management services, which are yet to be introduced there, either because banks do not offer these services (eg, Tsamenyi and Skliarova, 2005) or firms do not demand them (eg, in the survey analyzed in Brown et al., 2011, one-third of the firms report receiving less than one third of their income through their banks). Banks may also lack information on firm quality, project choice, or managerial effort, eg, incurring monitoring costs (Diamond, 1984; Diamond, 1991) or forming relationships with the firms (Sharpe, 1990; Rajan, 1992; von Thadden, 2004; Hauswald and Marquez, 2006; or Egli et al., 2006, among others). Also other financiers may face more information asymmetries in transition and developing countries (eg, Claessens et al., 2000).

(6) See Friberg and Wilander (2008). Firm risk aversion (Viaene and de Vries, 1992), currency variability (Engel, 2006), and medium of exchange considerations (Rey, 2001) may determine currency choice.

(7) if the firms' cash flows are in foreign currency, borrowing in the same foreign currency will provide a straightforward natural hedge (Goswami and Shrikhande, 2001). Mian (1996), Bodnar et al. (1998), Allayannis and Ofek (2001), and Brown (2001), among others, analyze the hedging of foreign currency exposure, using forward contracts and derivatives for example. But many developing country currencies have no forward markets; and even in those that do, there are substantial costs to hedging (Frankel, 2004). And even in developed countries, small firms rarely use derivatives to hedge their net currency exposure (Briggs, 2004; Borsum and Odegaard, 2005; and O'Connell, 2005, among others). As expected, therefore, small firms in developing countries not uncommonly default on loans in foreign currency following a deep depreciation of the local currency (Ziaul Hoque, 2003). Static capital structure trade-off theory suggests firms opt for the lowest cost debt, making the interest rate differential, that is, the deviations from the UIP, the second main determinant of the firm's choice of loan currency denomination (Graham and Harvey, 2001).

(8) As we later assume that the level of firm revenues does not change with the exchange rate, the changes in the exchange rate in our model are assumed to be real.

(9) For a richer model in which firms also differ with respect to their debt-to-income levels, see the SNB Working Paper version of our paper (Brown et al., 2009b).

(10) See Goldberg and Knetter (1997), for example, on exchange rate pass-through.

(11) Firms in our model receive both their expected income and their loan in a single, though not necessarily the same, currency. Without qualitatively affecting the main hypotheses, our model is readily extendable to include firms that receive their expected income and loans in varying proportions in multiple currencies.

(12) This is a crucial assumption in our model. If the UIP holds then the local currency earners will not have any incentive to borrow in foreign currency, as they will only bear higher costs either in terms of higher interest rate and/or in terms of prevailing distress costs.

(13) General reviews by Hodrick (1987), Froot and Thaler (1990), Lewis (1995), Engel (1996), for example. For emerging markets, see Francis et al. (2002) and Alper et al. (2009).

(14) Given our focus, we do not derive the optimality of this debt contract (see Townsend, 1979, eg).

(15) For example, this corresponds to the risk aversion of managers, as in Stulz (1984), or of firms, as in Calvo (2001).

(16) As financially distressed firm may lose customers, suppliers, and/or employees depending on the characteristics of their products and labor contracts for example, financial distress costs are also assumed to be heterogeneous across firms in Purnanandam (2008). Andrade and Kaplan (1998) estimate that financial distress costs vary between 10% and 20% of firm value (see also the review by Senbet and Seward, 1995). For small firms, both the level and dispersion of these costs are likely to be even higher (eg, Pindado et al., 2006).

(17) See the SNB Working Paper version of our paper (Brown et al., 2009b) for a model with a continuous distribution of firms' distress costs. A discrete distribution makes the analysis more

(18) In our model, all banks are equally affected by the information asymmetry regardless of the currency in which they lend. Most domestic and foreign banks in Eastern Europe, for example, offer loans in both local and foreign currency to local firms (see Brown et al., 2011 and Brown et al., 2012). If financiers lend only in their own currency, existing models predict that: (1) Firms may borrow first in the local and then in the foreign currency, after having exhausted internal funds, if local financiers have better information about the firm than foreign financiers (pecking order hypothesis); (2) Firms with high monitoring costs may borrow more locally in the local currency (Diamond, 1984); and (3) Better firms may borrow in the foreign currency to signal their quality, if foreign currency debt is more expensive (Jeanne, 1999) or entails more regulatory scrutiny hence higher distress costs (Ross, 1977).

MARTIN BROWN [1], STEVEN ONGENA [2] & PINAR YESIN [3]

[1] University of St Gallen, Rosenbergstrasse 52, St. Gallen 9000, Switzerland.

[2] CentER--Tilburg University and CEPR, PO Box 90153, 5000 LE Tilburg, Nethertands.

[3] Swiss National Bank, PO Box 2800, 8022 Zurich, Switzertand.