77ie European Journal of Finance 1, 257-278 (1995)

Incomplete contracts, renegotiation, and the choice between bank loans and public debt issues A. BAGLIONI Banca Commercial ttatiana - Research Department, Catholic University, Milan, Italy

In a two-period model where an investment project is funded with standard debt, the probability distribution of final cash flow is determined, at the interim date, by an unverifiable state of nature together with a choice by the controlling party (entrepreneur or creditor). With a control allocation contingent on a noisy default signal, renegotiation may improve efficiency in two ways: (i) reduce excessive risk-taking - due to the entrepreneur's moral hazard - through debt forgiveness; (ii) avoid the costs of financial distress eissociated with excessive liquidation or underinvestment by debt-holders, by letting them receive an equity stake in the firm. Such efficiency gain is an advantage of bank loans over publicly traded debt, given that the former are more easily renegotiated than the latter. The difference between the two types of debt is increasing in the degree of contractual incompleteness (noise present in the default signal) and in the portion of project value accounted for by future discretionary investment options. Keywords: incomplete contracts, debt renegotiation, financial distress, intermediation 1.

INTRODUCTION

The recent literature on financial contracting has shown the relevance of incomplete contracts^ in order to understand the origin of debt: the latter is seen as an efficient instrument to cope with such an incompleteness. For instance, Aghion cind Bolton (1992) show that the contingent control allocation allowed by a standard debt contract may be more efficient than a unilateral control allocation to the entrepreneur (non-voting shares) or to the investor (standard equity), in a model where there is a conOict of interest between the two parties and where renegotiation plays a crucial role in improving the 'An excellent survey on incomplete contracts is the one by Hart and Holmstrom (1987). A fundamental building block of such a theory is the idea of 'unverifiable information', that is information which is not verifiable by third parties in a contract (e.g. legal courts), so that it cannot be included in the contract, although it may be symmetrically observed by the two parties signing the contract. As we are going to see, this idea enabled the theory of finetncial contracting and intermediation to make some progress beyond those theories focusing on asymmetric information between borrower and lender: see, for example, the classics by Leland and Pyle (1977) and Diamond (1984, 1991); for a recent survey on banking theory, see Bhattacharya and Thakor (1993). 1351-847X © 1995 Chapman & Hall

258

Baglioni

equilibrium. The same authors (1989) stress the role of incomplete contracts for the capital structure theory: a particular capital structure, by inducing a correspondent governance structure, may influence investment decisions.^ Similarly, Zender (1991) shows how a combination of debt and equity is able to avoid underinvestment or risk-shifting in the presence of private information about investment: again, the incentive to invest inefficiently is coped with through a transfer of control from one investor (shareholder) to the other (debtholder), contingent on the realization of a publicly observable signal: a default situation; thus, bankruptcy is veiluable because it gives control over the firm to the party which is eilso the residual claimant. While the focus of the above literature is on issues like security design and the financial structure of the firm, it is of interest to understand the role of incomplete contracting and the value of renegotiation in the choice between placing debt on the open market and funding through a bank loan. Therefore, we apply the above approach in order to assess the vjilue of renegotiation under debt finance and to understand the role of financial intermediaries, namely of banks, which make the renegotiation process easier. Renegotiation of bank loan contracts has been explored in detail by Gorton and Khan (1993): in their model, the bank is entitled to exercise a call provision and to engage in renegotiation with the entrepreneur, pointing to a violation of loan covenants by the latter. Through renegotiation, the bank may be able to prevent the entrepreneur from adding a risk (mean preserving spread) to the ongoing project. Their model is extremely rich in showing the different outcomes to the renegotiation process when the entrepreneur retains control: the bank may liquidate the project, raise the interest rate, forgive some of the debt, or do nothing. The model presented here tries to extend the analysis to the case where there is a transfer of control to the investor during the life of the project, where renegotiation may reduce the indirect costs of financial distress due to a suboptimal investment policy. The implications of incomplete contracts for the choice between privately and publicly placed debt are examined by Berlin and Mester (1992), on the assumption that the former type of debt may be more easily ren^otiatied than the latter. In their model, the value of the option to renegotiate comes from the fact that, thanks to it, initial debt covenants can be made more severe, because they can be relaxed selectively in those states where they pose inefficient constraints; as harsher covenants help reducing the agency costs of debt, firms with a low creditworthiness - facing heavier «^ncy problems - will rely more than others on private/intermediated debt, on which more restrictive covenants may be initially written and eventually revised when necessary. Kahan and Tuckman (1993) find empiriccil evidence consistent witb some of Berlin and Mester's results. edlocation of residual rights of control (the so-called 'governance structure' determining who in the firm is entitled to Xske decisions in all unforeseen or unspecified contingencies) is the central issue edso in the theory of vertical integration developed by Grossman and Hart (1^6). More recently, the efficient governance structure of banks and the implications for their regulation have been explored by Dewatripont and Tiroie (1993, 1994).

The choice bettveen bank loans and public debt issues

259

Closely related to the present work are those studies showing the difficulty of renegotiating publicly traded debt, relative to bank loans.^ For instance, Giammarino (1989) models a situation where informational asymmetries between corporate insiders and debtholders may prevent the two parties from agreeing upon a mutually beneficial debt restructuring in case of financial distress. Diamond (1993) stresses "the difficulty of restructuring public debt issues, owing to free-rider problems and information asymmetries"; he assumes that there is a strictly positive probability that debt on the open market is not renegotiated even when it is in everyone's interest to do so. The incentives to freeride by public debtholders (the so-called 'holdout' problem) have been modelled also by Gertner and Scharfstein (1991), showing how a firm in financial distress finds it easier to renegotiate its debt with a bank (or a small syndicate of banks). Finally, problems of coordination among creditors are studied by Herring (1989). Empirical evidence in support of the above theoretical view is provided by Gilson et al. (1990) with US data, while Hoshi et al. (1990 a,b) provide evidence relative to Japan, showing that banks do in fact reduce the cost of financial distress and help relaxing liquidity constraints, thanks to their role in lessening information and incentive problems in the credit market; on the other hand, the cost of financial distress turns out to be higher for those firms relying more heavily on the open market as a source of finance, because both informational asymmetries and free-riding problems make it more difficult for public debtholders to allow for debt restructuring. Evidence on changes of ownership and control over distressed firms is reported by Gilson (1990). Finally, Smith and Warner (1979), in their analysis of bond convenants, point to the difficulty and higher costs of renegotiation in case of public debt issues, relative to privately placed debt. An area of research strictly linked to the above literature is the one concerning customer relationships between banks and firms, stressing the long term borrower-lender relation as a typical feature of bemkfineinceas opposed to public debt issues.^ Customer relationships enable the two parties to deal with contractual incompleteness, thanks to implicit agreements and to renegotiation. Thus, they may help in solving problems like credit rationing or underinvestment^; they may provide the firm with a stable source of finance, implying financial assistance in case of distress^; they may reduce the variability of the interest rate paid by the firm, by making it less sensitive to the current performance of the firm itself and to the market level of interest rates^. ^A recent survey of theoretical and empirical work onfineincialrestructuring is the one by John and John (1992). •"Also the approach taken by Chemmanur and Fulghieri (1994), although not framed in terms of customer relationships, stresses the advantage coming from the fact that beinks are players with a longer time horizon than bond-holders: thanks to that, they have more incentive to build up a reputation for employing a high ievel of resources in order to take the right decision between liquidating and renegotiating the debt contract with a financially distressed firm. 'See Sharpe (1991) and Rajan (1992). «See Hellwig (1989, 1991) and Mayer (1988). ^See Baglioni (1992). Osano and Tsutsui (1985) provide empirical evidence indicating that this sort of implicit contract is prevalent in the Japanese bank ioan meu-ket.

260

Baglioni

Coming back to the focus of our work, we try to address this specific problem: given that the instrument chosen to finance a project is stmidanj debt, is it more advantageous for the firm undertaking the project to place such a debt on the market or to apply for a bank loan? And which of the two alternatives leads to the more efficient investment strategy? In this perspective, given the continent control allocation allowed by the debt contract, it is relevant who is the debtholder, as this c£ui inSuence the final outcome of such an allocation with regard to decisions to be taken, such ais the choice eimong different investment strategies or liquidation versus continuation in c£ise of financial distress. The answers to such questions should hopefully provide some contributions to the theory of financial intermediation. In particular, the answer that we try to provide in this work comes from the assumption that a bank loan is more likely to be renegotiated, when necessary, than a publicly traded bond^: in fact, the bamk-firm relationship is typically framed in a way that facilitates the revision of the conditions set in the initial contract, thanks to the following features. First, some specific covenants included in the contract, like the 'call provision' zmd the 'prepayment option', entitle both parties to tri^er unilaterally a renegotiation of the existing conditions. Second, the information production carried out by the bank through its monitoring activity enables it to observe some inside information, which may be valuable in the renegotiation process. Third, the customer relationship between bank cind firm seems to be much more flexible than the financing relation between bond-holders cind firm, enabling them to remedy - at least partially - the incompleteness of contracts through implicit cigreements (sustained by incentive compatibility rather than by legal enforcement) and through the revision of conditions previously set. To the contrary, corporate securities placed on the open market lack such features (except maybe^ for the 'prepayment option') and in addition they give rise to important coordination problems among security-holders in the event of renegotiation: therefore, they seem much more difficult to be renegotiated when this becomes necessary in order to improve efficiency. Consequently, the advantage of a bank loan over publicly traded debt resides in the possibility of exploiting the efficiency gains allowed by renegotiation in some circumstances. In particular, the two cases here examined are when: (i) moral hazard, caused by debt financing together with limited liability, creates the incentive for the entrepreneur to undertake a risk which lowers the expected present value of the firm; (ii) a situation of financial distress produces a transfer of control to the debt-holder, having the incentive to turn down any kind of risks, including those able to increase the expected NPV of the firm: such an incentive comes from the fact that he is interested in maximizing the expected value of its claim rather than the expected value of the firm. While the first kind of distortion may produce excessive risk-taking, the second one may ^Another kind of iuiswer may derive from the consideration that different classes of creditors, like banks rather thcin bond-holders, differ with respect to their ability to exercise their rights of control. 'The caution is justified on the grounds that the fiexibility embedded in a bank credit line seems unlikely to be fully replicated through bond covenants.

The choice bettveen bank loans and public debt issues

261

lead to a suboptimal 'too conservative' investment strategy relative to first best, for example through excessive liquidation of existing projects or by missing valuable new opportunities (underinvestment). In both cases the renegotiation of the existing debt contract (when feasible) may lead to an improvement of the equilibrium, by restoring the incentive to take the first best decision. Namely, in the first case the investor may prevent the entrepreneur from excessive risk taking, by forgiving some of its debt (interest and/or principal); in the second one the entrepreneur may be willing to share the potential profits of the firm with its creditor, thereby inducing him to exercise efficiently rights of control: in other words, the first best choice by the debt-holder may become incentive-compatible by letting him have an equity stake in the firm in addition to its original debt claim. Thus, the opportunity of renegotiating the initial financing conditions turns out to be beneficial in two directions: reducing the moral hazard effects of limited liability; avoiding some costs of financial distress, when control rights are (at least partially) transferred to creditors, having the incentive to make inefficient investment (or disinvestment) choices. The value of renegotiation comes from the incompleteness of contracts, meaning not only that it is impossible to include every future contingency in financial contracts, but also that the control allocation implied by standard debt cannot be contingent on the true state of nature, but only on a noisy signal, namely a default situation. If, on the contrary, control rights were assigned on the basis of the true state, each party - entrepreneur and investors - could be given control exactly in those states where they have the incentive to take the most efficient action: then, the first best outcome would be feasible, without any need of renegotiating. Unfortunately, such an allocation is not possible: even when the true state is symmetrically observed by all the parties involved in the firm, it may not be verifiable by third parties and it may be very difficult to be foreseen or simply described ex ante, so that it cannot be included in the contract. What may be included and legally enforced is a default on some due payment, triggering a transfer of control to debt-holders. Then, default plays the role of a signal determining the assignment of control rights. The higher the noise of the signal, the higher the chance that control is given to the 'wrong' party, that is the one having incentive to take an inefficient action: it is precisely in those circumstainces that renegotiation becomes valuable, by restoring the incentives to act efficiently. In the specific context of our model, the party in control has the opportunity of adding some risk to the ongoing project: the state of nature determines whether such available risk has a positive or negative expected value. Given risk-neutrality, the efficient choice would be to add risk in the first case while not doing so in the second one*". If control were assigned to the entrepreneur in the first state and to the creditor in the second one, the first best action would cilways be compatible with the incentives of the controlling party, so that first best equilibrium would be achieved. To the contrary, control is given to the '"As we will see below, the choice of not adding risk may have the meaning, as a limit Ccise, of liquidating the project.

262

Baglioni

entrepreneur unless a situation of financial distress occurs (meaning a default on an interim pajmient together with a downward shtft in future project cashflows), in which case the debt-holder i^ts control. Although the occurrence of financial distress may be correlated with the true state, in general it will provide only a noisy signal, such a correlation being less than perfect. Thus, we have that the advant^e of intermediated debt over publicly traded debt derives from the possibiiity of renegotiating more easily, and that renegotiation in turn acquires value from the incompleteness of financial contracts: consequently, intermediation itself is more valuable the higher is the degree of contractual incompleteness. In particular, the less accurate is the default signal which triggers a transfer of control to creditors, the less efficient is the contingent control allocation implemented through a debt contract, so that the hi^er is the efficiency g^n expected from intermediation. If an entrepreneur, evaluating different sources of funding his/her project, may expect to benefit from at least some of such a gain, he/she may be willing to bear the cost of being monitored by a financial intermediary - namely a bank rather than placing debt on the open market. Another implication of the model is that the higher is the portion of firm value accounted for by future discretionary investment decisions rather than by assets already in place, the higher is the value of the renegotiation option: therefore, the firm is expected to rely more heavily on intermediated debt rather than on public debt issues. The following model is focused on the advantages of intermediated debt coming from the renegotiation option. Of course, it should not be interpreted as indicating that intermediated debt is always superior to public debt issues: other variables, affecting the entrepreneur's choice between the two sources of finance, are not explicitly accounted for in the model, but they have to be kept in mind when balancing the different factors affecting such a choice: for example, the cost of information production by a financial intermediary. Despite their importance, such variables are not included into the model as the analysis, focused on the value of the renegotiation option, would not be significantly enriched by such inclusion. The paper is organized as follows. In the following section we give an outline of the model; in Section 3 we describe the first best equilibrium obtained with a perfect default signal; in Section 4 we analyse how a control allocation contingent on a noisy default signal may produce suboptimal decisions and how renegotiation may help in restoring efficiency; in Section 5 the entrepreneur's choice between intermediated and publicly traded debt is considered; finally Section 6 summarizes the paper. 2. THE MODEL

We have a two-period model (summarized in Table 1). At the banning of the first period (t^, em entrepreneur hcis to choose how to fund his investment project, which requires cin initial input of 1, from external sources. He has only two alternatives: (a) applying for a bank loan or (b) issuing debt on the open market. As we are not interested in the security design issue, we take as given

The choice bettveen bank loans and public debt issues

263

that the financial instrument is standard debt, which may be justified thanks to assumption (iv) below. Both kinds of debt matures at ^2 (that is at the end of the second period), and both require also an interim (coupon) payment at t^ (end of first period): we label Dj and Dj the face value of such payments. The only difference between the two types of debt contract is that the bank loan may be renegotiated at the interim date (t^'), while the debt on the open market cannot, for reaisons we explained in the Introduction. The project yields two cash flows, one at the end of each period: R^ >D^ at t{, i?2 > D2 at ^2. unless a negative shock, creating a situation of financial distress, occurs (see below). At the end of the first period (t^, the controlling party (entrepreneur or creditor) has the opportunity of adding a risk to the ongoing project", which we define as e,, a continuous random variable whose density function is defined over the interval [e^,e^|] with e^ < 0 and t^ > 0. Such a risk may be of two types, depending on the reeilization of the state of nature / e {/i,/}, which may take the values h ('high') or / ('low'), with probabilities p^ and (1 - p^), respectively: if / = h, then the risk is e^, with density function /(e^) and with a positive expected value: E (eJ = e,, > 0; if / = /, the risk is e,, with density function g (i^ and negative expected value: E (Ei) = e, < 0. The choice about taking the risk e, may have two alternative interpretations. First, it may be seen as a liquidation versus continuation decision: in this perspective, the controlling party may decide, at t^, for an early liquidation of Table 1. The timing of the model D amount invested in the project = amount borrowed = 1 att^: D interim (coupon) payment D, is due n state /• e {h, 1} is observed by both entrepreneur and investor; the controlling party decides whether to add risk E, or not n verifiable signal s takes value equal to: nd (1st cash flow = ft^ > D.,) or d (1st cash flow = 0 => financial distress, control is transferred to creditor) D renegotiation of bank loan may occur at fj:

D final payment Dg is due D 2nd cash flow is equal to: f?2 > D2 or R2 + Ej (if risk is added at f.,); in case of financial distress: R'2
264

Baglioni

the project, with a liquidation value'^ equal to /?2 (or to R^, in case of financial distress we describe below); as an alternative, he may decide to continue the project during the second period, with a risky cash flow equal to /?2 + e, (or to R'2 + e,, in case of financial distress). In the second interpretation the decision to be taken - given a continuation of the project in the second period - concerns two alternative investment strategies, where one implies an additional risk (e,) relative to the other. As we make the assumption that all agents (namely: entrepreneur and investors) are risk neutral, it is clear that in state / = h the risk is worth taking, as it increases the expected value of the firm, while the opposite is true in the other state (/ = [); this would be the first best strategy, which would be selected by an agent maximizing the expected present value of the firm. Under the first interpretation given above, the first best strategy would have the meaning of continuing the project in the second period only if its expected value is greater than its liquidation value at f,, while liquidating if the opposite were true. Under the second interpretation, such a stratefflf would lead to taking the riskier investment choice only when it has a larger expected vcilue than the safer one. Unfortunately, neither the entrepreneur nor the investor (bank or bondholder) have always the right incentives. In particular, it may happen that the entrepreneur, because of limited liability, takes advantcige from undertaddng the risk even in the low state, thereby lowering the expected value of the firm by i,; on the other hand the investor, which is interested in the expected value of his debt claim rather than in that of the firm, may be unwilling to add risk to the ongoing project even in the high state, thus missing the opportunity of increasing the expected value of the firm by e^. As we are going to see in the next section, if a control allocation contingent on the relevant state / were available, then the first best strategy would be implemented: it would be sufficient to assign the decision over adding or not the risk to the entrepreneur in state i = h and to the investor in state / = /. But such an allocation is not feasible, «is we assume that both the state of nature /€ {h, 1} and the action taken (add or not add risk), although symmetrically observed by the entrepreneur and by the investor at ^1 (that is when the action has to be taken), are not verifiable information, so that they cemnot be included in the contract. Therefore, the two parties cannot contract on a given strategy contingent on /, and nor can the adlocation of control rights be contingent on state /. The above contractual incompleteness finds only a pteirtial remedy in the availability of a verifiable signal of the true state /, namely the occurrence of a negative shock producing a situation of financial distress. Such a shock may take place at t^: when this happens, the two project cash flows are modified as follows: R\ = 0 and /?2 = /?2 + A, with A < 0 such that /?2 < A where D = Di+ D2. That is: the entrepreneur is not able to meet the interim payment £>! and the second period cash flow is not sufficient to repay the total '^In this case (liquidation) the cash flow R2 (or /f'2) would be paid by the project at f,, instead of /j; however, its present value is unaffected, thanks to assumption (ii) below.

The choice bettveen bank loans and public debt issues

265

outstanding debt at ^2- When such a shock happens, we say that the signal s takes on the value s = d (default); otherwise, it takes the value s = nd (not default). As we have assumed that this signal is verifiable information, a control allocation contingent on its realization is feasible, where control over the project is assigned to the entrepreneur unless s = rf at ^i, in which case it is transferred to the investor: this is the typical control allocation of a standard debt contract. Clearly, such a control allocation would implement the first best strategy only if the verifiable signal s were perfect, providing full information about the true state /. In general, this will not be the case, as some noise will be present in the signal. To be more specific, we assume that the probability distribution of s conditional on / is such that: Prob(s = nd/i = H) = a^ > ^

Prob(s = nd/i = 0 = a, < 5

We define P = a,, - a,, belonging to the closed interval [0, 1], as a measure of contractual incompleteness, that is of the noise present in the signal: when P = 0 the signal provides no information at all about the true state /, whereas when p = 1 it provides perfect information. In genercil (0 < P < 1), the observation of the value taken by the signal s at t^ will provide some information about the state of nature'^, but not sufficient to distinguish perfectly between its two realizations {h, I}. Therefore, the control allocation contingent on s will not, in general, achieve the first best outcome. In particular, it may still happen that control is given to the entrepreneur in state / = / and to the creditor in state / = h, creating the possibility of inefficient actions being taken. Then, the availability of renegotiation becomes crucial: when control is assigned to the party having the 'wrong' incentive, the revision of initial conditions may lead to the most efficient choice being selected, by making it incentive compatible. That's why the expected efficiency gain of renegotiation, and consequently the value of intermediation, are increasing in the degree of contractual incompleteness (see Proposition 4 below). Additional assumptions of the model are the following. (i) There is perfect competition among creditors at t^, so that the equilibrium face value of debt is determined by the zero-expected profit condition imposed on the debtholder; consequently (see Section 5), all the expected - as of tQ - efficiency gain from renegotiation goes to the entrepreneur, (ii) The cost of funding for banks (and the opportunity cost of funds for investors) is the risk-free rate", which we set equal to zero for simplicity. This is also the rate at which agents discount future Ccish-flows. (iii) In the case of financial distress (s = d), the entrepreneur cannot avoid insolvency by making use of alternative sources of funding. ''If the value s = nd is observed, the posterior probability of the true state being i = h increases, while the opposite is true when s = d. '••Remember the above cissumption that all agents are risk-neutral.

266

Baglioni

(iv) The lender does not observe the actual project cash flows, unless in the financial distress situation, where he is given control over the firm; this ex post asymmetry of information between the investor and entrepreneur makes the standard debt contract the optimal financial instrument. In the following, we will proceed backwards. In Sections 3 and 4 we concentrate on what happens at t^, that is at the beginning of the second period. First, we will examine the benchmark Ccise where s is a perfect signal (P = 1): in this case the first best action is always selected, without the need of renegotiation. Second, we consider the more interesting case of a noisy default signal (P < 1), where renegotiation plays a crucial role in restoring the incentive compatibility of first best actions, thus improving the equilibrium. Then, in Section 5 we will see how the face value of debt is determined at /Q (beginning of first period) and we will examine the entrepreneur choice between funding his project through a bank loan rather than through a debt issue on the open market. 3.

PERFECT SIGNAL (p = 1): FIRST BEST

We are at the beginning of the second period (ti). If the signal is perfect (meaning that a,, = 1 and a, = 0) the allocation of control rights is such that the party in control has Jilways the incentive to take the most efficient action, without the need of renegotiating. In order to prove that, we have to examine the following two cases. Let us begin with the 'high' realization of the state of nature / = h: given this, the signal takes the value s = nd with probability 1 and control is always assigned to the entrepreneur. The latter will have to decide whether to add the risk if, to the ongoing project or not, by comparing his expected profit in the second period in the two cases'^:

J

(/?2 + e, - D^fCz^dif,

> /?2 - ^2

(1)

D^Rz

The above equation is equivalent to the following DT-RI

f J

0

(2)

which is clearly always satisfied. Therefore, the entrepreneur will efficiently choose to add risk to the project, thus increasing its expected value by

U> 0). "Remember that s = nd implies that first period Ccish flow is Ri > D^, so that the first debt payment £>, has been paid at f^ DT-RI "By adding and subtracting from the left-hand side: J (R2 -D2

The choice between bank loans and public debt issues

267

Let us now turn to the other case, where the 'low' state / = / occurs. Given this, the signal takes the value s = dlox sure and control is transferred to the creditor: he will not add risk to the ongoing project if the following is true: D-R\

f -

EM

- -

f - -

R'2 > (e, + R'^g(eMe, + D g(ef)de/ (3) where the right-hand side is the expected creditor's payoff if the risk e, is added to the project and the left-hand side is his expected payoff if no risk is added'^. The right-hand side of the above expression needs some explanation. The return to the creditor is bounded above by D, meaning that when the project cash flow R'2 + il takes values higher than the outstanding debt (D), he gets just D and he is not entitled to receive the part of the cash flow exceeding D. This may be justified on the grounds that we are dealing with a situation of financial distress, where creditors exert some significant control power on the firm - like deciding over a (partial) liquidation of assets or limiting investment plans - but where they are not yet the new owners of the firm. The situation we are describing is the one taking place before a formal bankruptcy procedure is completed or even begun; thus, it applies to all situations where informal agreements are negotiated between firms and their creditors in order to avoid a formal bankruptcy; or alternatively to such cases where, during a bankruptcy procedure, relevant decisions are to be made, related to asset liquidation or implementation of investment plans. Therefore, although some control rights are transferred to them, creditors are not residual claimants on project return'*. It is immediate to prove that condition 3 always (strictly) holds. Consider that: «« K 2 > ' ^ 2 " ' " ^ / "^ Em

which in turn is larger than the right-hand side of Equation 3. Therefore, the controlling creditor never has the incentive of adding risk to the project and such behaviour is efficient, given that adding risk would lower the expected value of the project by £/(< 0). We are then able to state the following: PROPOSITION I: When the default signal s e {d, nd} is perfect (^ = 1), then the first best action is always selected by the controlling party (at t,) in each state, that is: add risk in state i = h and not add risk in state i = I; such an equilibrium is feasible through a standard debt control allocation contingent on the verifiable signal s, without renegotiation. '^Remember that the default signal s = d mesins that the first period cash flow is 0, so that total debt coming due at t2 is D = Di + D2, and that the second Ccish flow of the project (if no risk is added) is R'2 < D for sure. "This assumption about creditor's return in ccise of fincUicial distress has no consequence here, as it does not affect his decision; on the contrary, it will be crucial in paragraph 4.2.

268 4.

Baglioni NOISY SIGNAL (P < 1): THE VALUE OF RENEGOTIATION

In this section we consider the more interesting case where contracts are incomplete, meaning that the verifiable default signal s is not perfect. In particular: given a 'low' state / = /, there is a positive probability (a^) that control remains to the entrepreneur; on the other hand, given a 'high' state / = h, there is a positive probability (1 - a^) that control is transferred to the creditor. These are the two cases where the controlling party may have a private incentive to act inefficiently, that is to add to the ongoing project the risk i, while turning down the risk e^: then, the right incentives may be restored through renegotiation of the initial contract, so that a more efficient equilibrium may prevail. Let us examine two such ceises in turn. 4.1 RenegotMing to avoid e x c e ^ v e risk-iiricing: the bsMik f ( M ^ ^ part of the debt

We begin with the case where, at t^, the 'low' state / = / is observed by both parties while there is no financial distress: the verifiable signal takes the value s = nd, so that the entrepreneur has to decide whether to add risk or not to the project. He will add risk if:

J J

, >/?2-Z)2

(4)

Dr-R2

which is equiveilent to'^: „ „

D2-R D2-R2

J

i i d i , >-e,

(5)

£

Unlike Equation 2, the above condition is not always met: it depends - among other things - on the level of outstanding debt D2. The higher is the face value of the debt to be paid at t2, the 'more likely' condition 5 is met, in which case the entrepreneur hcis the incentive to add risk, although such a risk lowers the expected value of the project by e,: this is nothing else but the moral hazard distortion induced by limited liability^". If Equation 5 is not satisfied, then the entrepreneur will choose not to add risk: there is no morail hazard. In order to have a more interesting ccise to examine, we assume that the face value of debt D2 is high enough that condition 5 is met, so that the entrepreneur is subject to moral hazard. More precisely: we assume that (i) there is a positive Dl such that Equation 5 holds with equality and (ii) D2 > D^. Consequently, the enterpreneur inefficiently chooses (at t^) to add the risk e, to the project. Notice that enterpreneur's choice not only is suboptimal with respect to the maximization of the expected value of the firm, but also creates a conflict of interest with the creditor: the latter would be better off if no risk were added, as the following inequality holds: "At this point, the interpretation of Equation 4 and the derivation of Equation 5 should be trivial (see Equations 1 and 2). ^''See Jensen and Meckling (1976) and Stiglitz and Weiss (1981).

77ie choice between bank loans and public debt issues D2-R2

D2 J

269

£«

(/?2 + ei)g(ei)d8, + D2

J g(e^de,

(6)

where the right-hand side is the expected value of creditor's claim if the risk e, is added to the project, while the left-hand side is the expected value if no risk is added^'. If the debt contract is a bank loan, the incentive for the entrepreneur to act more efficiently may be restored through renegotiation over the face value of outstanding debt D2. We assume here that all the bcu'gaining power in the renegotiation process is assigned to the bank. This may be justified on the grounds that the bank, thanks to the inside information produced and to the customer relationship established, enjoys some monopoly power over the entrepreneur^^. In any case, such an assumption is neutral with respect to the efficiency gain from renegotiation and to the first period choice of the entrepreneur between funding the project through a bank loan rather than through the open market: as we cire going to see in Section 5, assuming perfect competition among creditors implies that all the expected gain - as of ^Q - from the availability of renegotiation goes to the entrepreneur, independently of the distribution of bargaining power in the renegotiation process. The bank mcikes a take-it-or-leave-it offer to the entrepreneur, proposing a new face value of debt, say D'2, which is incentive compatible with the entrepreneur not adding risk to the project. The bank will propose the highest possible Vcdue of D'2, subject to the incentive compatibility constraint of the entrepreneur: from Equation 5 it is clear that such a value is D'2 = D2 (< ^2). which makes the entrepreneur indifferent between the two alternatives^^ Thus, the bank should forgive part of the debt (P2-D'^ in order to induce the entrepreneur to act efficiently (that is: not taking the risk ij). Of course, we have to check whether it is rational for both parties to participate in such a renegotiation, that is whether their expected payoffs with renegotiation are (weakly) greater than without. It is immediate to verify that the entrepreneur participation constraint is strictly satisfied. Such a constraint is:

I

ddi,

(7)

Now, by definition of D^, from Equation 4 we have: £«

EM

- D; + i;)g(i;)di, > J hR "Remember that in such a case the second period cash flow is going to be ^2 * ^2 ' o ' sure. ^^Of course, this is not an issue taken explicitly into account in our model. See Sharpe (1990) and Rajan (1992) for a detailed treatment. ^'We make the standard assumption that when one party is indifferent between two alternatives, he wiii choose the one consistent with the renegotiation outcome.

270

Baglioni

Therefore, the entrepreneur participation constraint is slack (while the incentive compatibility constraint is binding). On the other hand, the participation constraint of the bank is the following:

r..

f

-

- -

f

- -

D2 > I (/?2 + efyg(e,)de, + ^2 J SQ^L^^I which is equivalent to^'': ^^

- ^2- J

(9)

(10)

where G(ej) is the distribution function of e,. Equation 10 is the feasibility condition for renegotiation, setting a lower bound to the value of D^: this means that the highest face value of debt which is incentive compatible with the entrepreneur not adding risk must be at least as large as the right-hand side of Equation 10; otherwise, there is no new face value £> 2 incentive compatible with the entrepreneur acting efficiently and at the same time making the bank (weakly) better off with renegotiation than without. Therefore, the bank is not always able to prevent the entrepreneur from excessive risk-taking^^. When the feasibility condition 10 is met, both parties are (weakly) better off by renegotiating over the face vadue of outstanding debt, setting it equal to D^, so that excessive risk taking by the entrepreneur is avoided. The efficiency gain from renegotiation equals -e,, which is the increase in the expected value of the project (as of t{) thanks to renegotiation. It is very Qasy to check that the sum of the expected gains (as of tj from renegotiation by the two parties is exactly equal to the overall efficiency gain: -e,. From Equation 9, the expected payoff of the bauik increases by: D2-R2

tu

j D-R2

while from Equation 7 the expected profit of the entrepreneur increases by (12) D^R2

The sum of expressions 11 cind 12 equals -e,. The discussion in this subsection leads to the following: PROPOSrnON2: Given that the verifiable default signal s e {d, nd] is noisy (P < 1), when at ti control is assigned to the entrepreneur (s = nd) in state i = 1, excessive risk taking may be avoided by lowering the face value of outstanding debt to D^, D2-R7

integrating by parts:

E,g(ej)de,.

result is in line with the one by Gorton and Kahn (1993).

The choice between bank loans and public debt issues

271

provided the feasibility condition (10) is met. The efficiency gain from renegotiation Qvhen feasible) equals -e,. 4.2 Renegotiating to reduce the cost of financial distress: the bank receives part of the equity

Let us now turn to the case where, at /,, the 'high' state i = h occurs and it is observed by both parties (that is: the available risk is i^ with positive expected value Eft), while the verifiable signal takes the value s = d: this means that the firm is in financial distress, due to the shock which lowers projected cash Oows in the following way: first period cash Oow is /?', = 0, so that the firm defaults on the first debt payment Dj, and the second period cash flow (if no risk is added) is going to be R'2 < D, where D = Dj + D2 is now the outstanding debt to be paid at ^2- Due to the default situation, control is transferred to the debtholder, which has to decide whether to add risk to the project or not. He will add risk if: D-^'2

JJ

EM

(Eft + /? 2)/(ift)di, + DJJ f(i,)di,

> R'2

(13)

D-R2 DR

Em

which is equivalent Eft > J (/?'2 + eft-D)/(e^deft

(14)

DHR'2

The above condition has a clear interpretation. Notice that the right-hand side is the expected profit of the entrepreneur if the risk e,, is added to the project, whereas its profit would be zero if no risk were added (as ^2 < ^)- therefore, such an integral is the expected entrepreneur's gain from adding risk. Condition 14 says that the expected value of such a risk (eJ has to be high enough that 'something is left' to the creditor in terms of expected gain, once the entrepreneur's gain has been taken into account. If that condition is met, then the creditor is induced to act efficiently, that is to undertake the risk, thereby increasing the expected value of the project by e,,; in addition, by doing so he creates the possibility for the firm to recover from financial distress and to be solvent at ^2: this happens for all realizations of Eft > D-R'2.

In the following, we will exeimine the case where condition 14 is not met, so that - without renegotiation - the creditor has the incentive to act inefficiently, that is to turn down the risk. Such a behaviour is not only inefficient, but also creates a conflict of interest with the entrepreneur, who is clearly better off if the risk is 2*By adding and subtracting from the left-hand side: J ( D-R;

"As we already noted, the entrepreneur's expected profit is equal to the right-hand side of Equation 14 if the risk is added to the project, and zero otherwise.

272

Baglioni

Here there is a role for renegotiation between bank and entrepreneur, if the debt contract is a bank loan: the entrepreneur is willing to pass on to the bank a portion e e [0, 1 ] of his residual claim on the project, in order to induce the bank to act efficiently; in other words, the bank receives an equity position in the project in addition to the original debt claim, and this makes adding risk incentive compatible. As we are going to show below, it is always rationad for the entrepreneur - given the situation of financial distress - to do so: thus, renegotiation is always feasible in this case. By the way, notice the symmetry with the renegotiation examined in the preceding subsection: in that case the bank was ready to forgive part of his claim (P2 - D*^ in order to avoid the entrepreneur's excessive risk taking; here it is the entrepreneur who is willing to give up a share e of his own claim on the project, in order to avoid a 'too conservative' policy by the bank. Obviously, the equilibrium value of e = e* will depend on the distribution of bargaining power between the two parties: in particular, if all the bargaining power is assigned to the bank, then e"^ = 1, whereas in the opposite case e* e (0, 1). Again, however, the distribution of beirgcuning power is immaterial to the efficiency gain of renegotiation and to its feasibility. That's what we are going to show in the remaining part of this subsection. Let us begin with all the bargaining power being assigned to the bank, making a take-it-or-leave-it offer to the entrepreneur. The bank proposes the highest possible value of e subject to the entrepreneur's participation constraint: E«

>0

(15)

D-R\

Such a value is clearly e* = 1. On the other hand, it is easy to see that the incentive compatibility and the particii>ation constraints of the bcmk are the same, that is: D-Rz

en

J J

)di, > R'2

(16)

D-Rz D-Rz

which, for e = e* = 1 becomes:

which, of course, is always true with strict inequality. The equilibrium value e* = 1 might be striking, but it is not. Consider that the entrepreneur's residual claim on the project is worth nothing if no risk is added, as the final cash flow is bound to be R'2 < D for sure: that's why he is willing to pass it on to the bank and, in the limit case where all the bargaining power is assigned to the bank, the latter is able to get all of such a claim, thereby extracting all the surplus from renegotiation (eJ. The interesting feature of renegotiation here is that the equity claim is worth nothing if it stays with the entrepreneur, whereas it has a positive value of e,, if it is transferred to the bank, to the change in incentives it produces.

The choice between bank loans and public debt issues

273

If the distribution of bargaining power is reversed, so that it is the entrepreneur making a take-it-or-leave-it offer to the bank, his proposal will be the lowest possible value of e subject to the incentive compatibility and participation constraints of the bank, which are both given by Equation 16. Such a value is given ^* e*= 1

^ E«

I (R'2 + ih-D)f(i,;)dii,

(18)

D-R'2

which clearly belongs to the open^^ interval (0, 1). It is immediate to see that with e* € (0, 1) the entrepreneur's participation constraint 15 is (strictly) satisfied. Moreover, the entrepreneur extracts all the surplus from renegotiation: in fact, its expected profit increases by i,,,^" while the expected payoff of the bank remains equal to R'2.^^ We may summarize the above discussion in the following: PROPOSITIONS: Given that the verifiable default signal s e {d, nd} is noisy (P < 1), when at t, control is assigned in state i = h to the bank (s = d) and the latter has incentive to act inefficiently (that is not add risk if, to the project), a number e* e (0, 1] exists such that: the entrepreneur may induce the bank to act efficiently by transferring to it a share e* of his own residual claim on the project; such a renegotiation is always feasible and the efficiency gain is equal to if,, independently ofthe distribution of bargaining power between the two parties; to the contrary, the value of e* and the allocation ofthe renegotiation surplus between the two parties are affected by the distribution of bargaining power When a situation offincincialdistress produces a (partial or total) treinsfer of control rights to debt-holders, the latter may have incentive to follow a 'too conservative' policy, meaning an excessive liquidation of existing projects or missing valuable new opportunities (underinvestment^^: such a constraint imposed by creditors on the investment policy of the firm may be seen as a cost of financial distress, which lowers the expected value of the firm below its first best equilibrium value^^. Proposition 3 tells us that a remedy may be found in a renegotiation between the firm and the debt-holder (specifically: the bank) which entitles the latter to receive an equity stake in the firm: this creates the right incentive for the bank to act efficiently. and subtract

(R'2+e.^f(i^de.^ from the left-hand side of Equation 16 and solve for e. D-R'2

n fact: we are examining the case where Equation 14 holds with the inequality strictly reversed • e* > 0; on the other hand, we have the assumption that e^ > 0 => e* < 1. see this, substitute the value of e* from Equation 18 into Equation 15. definition of e*, constraint 16 above is binding. (1977) analyses another source of underinvestment, nameiy the incentive for the entrepreneur (shareholder-manager) to turn down positive NPV investment options when the level of debt exceeds such NPV. 'Earner (1977) classifies this profit loss - due to mismanagement of the firm during the bankruptcy process - as "indirect bcinkruptcy costs".

Baglioni

274

5. FIRST PeUOO @Mltt.»mUM: THE CHOICE BCTV^EN BAIiH< LOANS AND PUBUCLY TflADlD

From the preceding two sections, it should be clear that the ability of renegotiating enables the two parties - entrepreneur and debtholder - to reach a more efficient equilibrium as to the choice between adding or not the available risky opportunity (e,, / e [h, /}) to the ongoing project. In particular, when in state i = I the entrepreneur retains control, he may have the incentive to inefficiently add risk to the project (moral hazard): this excessive risk-taking may be avoided through a renegotiation which lowers the face value of outstanding debt (Proposition 2). On the other hand, when in state / = h control is transferred to the creditor due to a default situation, he may have the incentive to inefficiently turn down a valuable investment opfK)rtunity: this cost offinfuicialdistress may be avoided through a renegotiation which results in the creditor receiving an equity stake in the firm (Proposition 3). Under the assumption we made in Section 2 of perfect competition among creditors, the face value of debt D = D1+D2 will be determined at t^ by the zero-expected profit condition imposed on the debtholder. With an (opportunity) cost of funds equal to the risk-free rate, such a condition is the following, if renegotiation is not an available option:

I - l =0(19)

Di + J (7?2 + ei)g(e2)de/ + ^2 J Zm

D2-R2

With the opportunity of renegotiating, such a condition becomes^^:

f

. . .

f

D, + J (/?2 + e^/(e,^de;, + D2 J

- 1= 0

(20)

+ (l-Pft)a,(Di

Imposing the above conditions implies that all the expected efficiency gain from renegotiation goes to the entrepreneur, either directly through an appropriation of some renegotiation surplus at ^1 or indirectly through the pricing of his debt^^ at t^. Therefore, at t^ he will get the difference between the expected values of the project with and without renegotiation, which are, respectively: '••Equation 20 is written under the hypothesis that the bank retains all the bargaining power: as should be clear from the text, such an hypothesis is immaterial to the final result, that is the increase in entrepreneur s expected profit (net of the cost of funding) thanks to renegotiation. '^The proportion between the two ways will depend, of course, on the distribution of bargaining power between the beink and the entrepreneur.

The choice between bank loans and public debt issues

275 ft) (21) '2

(22)

Not surprisingly, the difference between the two is: i

(23)

which is the sum of the expected efficiency gains from renegotiation when the entrepreneur retains control in state / = / and when control is trcmsferred to the creditor in state / = h, weighted by the probabilities of such events occurring. The above expression is increjising in the degree of contractual incompleteness, being inversely related to P (as defined in Section 2): so the higher is the noise present in the verifiable signal s E {d, nd}, the lcirger is the benefit that the entrepreneur may get from renegotiation (in the limit, the value of renegotiation reaches its maximum for p = 0, while for p = 1 it is equal to zero). When (at ^Q) the entrepreneur has to choose between the two sources of funding his project, the opportunity of renegotiating weights in favour of the banking channel, under the assumption that issuing debt on the open market makes such an opportunity not (or less likely) available. The model, therefore, provides a reason why an entrepreneur may find it optimal to rely on intermediated debt rather than on the market. The benefit of intermediation, of course, has to be balanced with the cost of information production by the bank, which is not explicitly taken into account in the model. We may summarize our discussion in the following: PROPOSITION 4: If renegotiation is allowed at t,, the expected value of the project, as of to, is increased by (J-p^
As we already said, the gain from renegotiation is twofold: reduce excessive risk taking, due to morad hazard, when the entrepreneur retains control; avoid some costs of financial distress, namely excessive liquidation or underinvestment, when control is transferred to creditors. Such a gain is larger when the contingent control allocation implied by the debt contract is less efficient, that is when the verifiable signal provided by default is less correlated with the underlying true state determining the prospects of the firm. Therefore, a testable implication of the model is that in those cases (firms, industries) where the occurrence of finemcial distress is less correlated with real economic performance, borrowers should rely more on intermediated debt, relative to the open market. Another implication derives from the fact that the efficiency gain from renegotiation is proportional to (-e^) and to e^,, that is to the scale of the available risks (as measured by their - absolute - expected values). This implies that the higher is the portion of the project expected value due to the option to

276

Bagioni

undertake such risks, the higher is the vaiue of the renegotiation made possible by intermediated debt; consequently, those firms whose value is more accounted for by future discretionary investment decisions rather than by assets already in place, should rely more heavily on intermediated debt^. 6.

SUMMARY AND CONCLUSIONS

In this work, the incomplete contracts approach to financial contracting was employed in order to explore the benefits of renegotiation under debt finance. With a project final cash flow determined - at the interim date - by an unverifiable state of nature together with an action taken by the controlUng party (entrepreneur or creditor), the allocation of control ri^its becomes crucial. In fact, if control were contingent on the true state (or equivalently on a perfect signal) the first best action would always be selected. To the contrary, given a control allocation contingent on a noisy default signal, two kinds of inefficiency may arise: (i) when entrepreneur retains control, he may have incentive to take suboptimal risky strategies, due to moral hazard; (ii) when control is transferred to creditors, they may be induced to inefficiently liquidate existing projects or to turn down valuable investment opportunities, thus generating indirect costs of financial distress. In both cases, renegotiation may be helpful in restoring efficiency, by maMng first best decisions incentive compatible. In the first case, a reduction in the face value of debt may avoid moral hazard; however, the creditor is not always able to control the entrepreneur's excessive risk-taking (as renegotiation might not be feasible). In the second one, excessive liquidation or underinvestment by the controlling debt-holder are avoided through a renegotiation resulting in the creditor adding an equity stake in the firm to its original debt claim. We assumed that bank loans are more easily renegotiated than publicly traded debt, thanks to some features like specific convenants (call provision and prepayment option), information production cstrried out by banks Jind typical bank-firm customer relationships, in addition to coordination problems among corporate bond-holders. Therefore, the efficiency gains of renegotiation are an advantage of intermediated over publicly traded debt, so that, in the entrepreneur's choice between the two sources of funding, they weight in favour of the first one. The value of intermediation increases with the degree of contractual incompleteness, measured by the noise present in the default si^pial: the higher the noise, the less efficient the contingent control allocation implied by the debt contract, so the lar^r the gain from renegotiation. Another implication of the model is that the renegotiation surplus - and consequently the incentive to rely on intermediated debt - is increasing in the portion of project value accounted for by future discretionary investment options rather than by assets in place. impdication parallels the one drawn by Myers (1977) from his model on underinvestment, namely that leverage should be inversely related to the ratio of future growth opportunities to total firm value.

The choice between bank loans and public debt issues

277

ACKNOWLEDGEMENTS

I would like to thank all participants in the European Finance Association 21st Annual Meeting (Brussels, August 1994), and in peuticular my discusszuit S. NagfU'ajan for his precious comments. Many thanks also to U. Cherubini and M. Esposito, to the participants in a seminar at Bemca Commerciale Italiana Research Department, and to an anonjnnous referee. The usual discledmer applies. REFERENCES

Aghion, P. and Boiton, P. (1989) The financial structure of the firm and the problem of control, European Economic Review, 33, 286-93. Aghion, P. eind Boiton, P. (1992) An incomplete contracts approach to financial contracting. Review of Economic Studies, 59, 473-94. Bagiloni, A. (1992) Bank-firm customer relationships as dynamic game equilibria, Intemational Review of Economics and Business, 39, 943-58. Berlin, M. and Mester, L.J. (1992) Debt convenants and renegotiation, Joumal of Financial Intermediation, 2, 95-133. Bhattacharya, S. and Thakor, A.V. (1993) Contemporary banking theory, Joumal of Financial Intermediation, 3, 2-50. Chemmanur, T.J. and Fulghierl, P. (1994) Reputation, renegotiation, and the choice between bank loans and publicly traded debt. The Review of Financial Studies, 7, 475-506. Dewatripont, M. cind Tiroie, J. (1993) Efficient governance structure: implications for beinking regulation, in Capital Markets and Financial Intermediation, C. Mayer and X. Vives (eds) Cambridge University Press. Dewatripont, M. and Tiroie, J. (1994) The Prudential Regulation of Banks Cambridge, MA. MIT Press. Diamond D.W. (1984) Financial intermediation and delegated monitoring. Review of Economic Studies, 51, 393-414. Diamond D.W. (1991) Monitoring and reputation: the choice between bank loans and directly placed debt, Joumal of Political Economy, 99, 689-721. Diamond D.W. (1993) Bank loan maturity and priority, when borrowers can refinance, in Capital Markets and Financial Intermediation, C. Mayer and X. Vives (eds) Cambridge, UK, Cambridge University Press. Gertner, R. and Scharfstein, D. (1991) A theory of workouts and the effects of reorganization law, Joumal of Finance, 46, 1189-222. Giammarino, R. (1989) The resolution of financial distress. Review of Financial Studies, 2, 25-47. Gilson, S.C. (1990) Bankruptcy, boards, banks, and blockholders, Joumal of Financial Economics, 27, 355-87. Gilson, S.C, John, K. and Lang, L.H.P (1990) Troubled debt restructurings, Joumal of Financial Economics, 27, 315-53. Gorton, G. and Kahn, J. (1993) The design of bank loan contracts, collateral, and renegotiation, NBER Working Paper no. 4273, Cambridge, MA, February. Grossman, S.J. and Hart, O.D. (1986) The costs and benefits of ownership: a theory of vertical and lateral integration, Joumal of Political Economy, 94, 691-719. Hart, O. and Holmstrom, B. (1987) The theory of contracts, in Advances in Economic Theory. 5th World Congress, T.F Bewley (ed.) Cambridge, tJK, Cambridge University Press.

278

Baglioni

Hellwig, M. (1989) Asymmetric information, financial markets, and financial institutions. European Economic Review, 33, 277-85. Hellwig, M. (1991) Banking, financial intermediation and corporate finance, in European Financial Integration, A. Glovannini and C. Mayer (eds) Cambridge, UK, Cambridge University Press. Herring, R.J. (1989) The economics of workout lending, Joumal of Money, Credit and Banking, 21, 1-15. Hoshi, T, Kashyap, A. and Scharfstein D. (1990, a) Bank monitoring and investment: evidence from the changing structure of Japanese corporate banking relationships. in Asymmetric Information, Corporate Finance, and Investment, Hubbard R.G. (ed) Chicago, IL, University of Chicago Press pp. 105-26. Hoshi, T, Kashyap, A. and Scharfstein D. (1990, b) The role of banks in reducing the costs of financial distress in Japan, Joumal of Financial Economics, 27, 67-S8. Kahan, M. and Tuckman, B. (1993) Private vs. Public Lending: Evidence from Convenants, New York University Salomon Center, Working Paper Series, S-93-54, November. Jensen, M.C. and Meckling, W.H. (1976) Theory of the firm: managerial behavior, agency costs and ownership structure, Joumal of Financial Economics, 3, 305-60. John, K. and John, T.A. (1992) Coping with finsuiciai distress: a survey of recent literature in corporate finance. Financial Markets, Institutions and Instruments, 1, 63-78. Leland, H. and Pyle, D. (1977) Informational asymmetries, financial structure, and financial intermediation, Joumal of Finance, 32, 371-87. Mayer, C. (1988) New issues in corporate finance, European Economic Review, 32, 1167-89. Myers, S.C. (1977) Determinants of corporate borrowing, Joumal of Financial Economics, 5, 147-75. Osano, H. and Tsutsui, Y. (1985) Implicit contracts in the Japanese bank loan market, Joumal of Financial and Quantitative Analysis, 20, 211-29. Rajan, R.G. (1992) Insiders and outsiders: the choice between informed and arm's-length debt, Joumal of Finance, 47, 1367-400. Sharpe, S.A. (1990) Asymmetric information, bank lending, and implicit contracts: a stylized model of customer relationships, Joumal of Finance, 45, 1069-87. Sharpe, S.A. (1991) Credit rationing, concessionary lending, and debt maturity, Joumal of Banking and Finance, 15, 581-604. Smith, C.W. and Warner, J.B. (1979) On financial contracting. An analysis of bond covenants. Journal of Financial Economics, 7, 117-61. Stiglitz, J.E. and Weiss, A. (1981) Credit rationing in markets with imperfect information, American Economic Review, 71, 393-410. Warner, J.B. (1977) Bankruptcy costs: some evidence. Journal of Finance, 32, 337-47. Zender, J.F (1991) Optimal financial instruments. Journal of Finance, 46, 1645-63.