POLICY RESEARCH WORKING PAPER 2364 Perverse Effects of a Allowing banks to hold less capital against loans to Ratings-Related Capital borrowers who have received Ade qu acy S ys te m a favorable rating by an approved rating agency may result in a rating system that Patrick Honohan neither reveals risk information about borrowers nor protects the deposit insurance fund. Part of the problem is the very idea of basing portfolio risk evaluation on the sum of individual loan risks, but there are also important incentive issues. The World Bank Development Research Group Finance June 2000 POLICY RESEARCH WORKING PAPER 2364 Summary findings It has recently been proposed that banks be allowed to chiefly at times of systemic crisis. And using agencies' hold less capital against loans to borrowers who have individual ratings is unlikely to be an effective early- received a favorable rating by an approved rating agency. warning system for the risk of systemic failure, so use of But a plausible model of rating-agency behavior shows the ratings could lull policymakers into a false sense of that this strategy could have perverse results, actually security. increasing the risk of deposit insurance outlays. It is important to harness market information to First, there is an issue of signaling, with low-ability improve bank safety (for example, by increasing the role borrowers possibly altering their behavior to secure a of large, well-informed, but uninsured claimants), but lower capital requirement for their borrowing. this particular approach could be counterproductive. Second, establishing a regulatory cut-off may actually Relying on ratings could induce borrowers to increase reduce the amount of risk information made available by their exposure to systemic risk even if they reduce raters. exposure to specific risk. Besides, the credibility of rating agencies may not be damaged by neglect of the risk of unusual systemic shocks, although deposit insurers greatest outlays come This paper - a product of Finance, Development Research Group - is part of a larger effort in the group to examine the effects of financial sector regulation. Copies of the paper are available free from the World Bank, 1818 H Street NW, Washington, DC 20433. Please contact Agnes Yaptenco, room MC3-446, telephone 202-473-8526, fax 202-522-1 155, email addressayaptenco@worldbank.org. Policy ResearchWorkingPapers are also posted on the Web atwww.worldbank.org/ research/workingpapers. The author may be contacted at phonohan@worldbank.org. June 2000. (14 pages) The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, evenz if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the view of the World Bank, its Executive Directors, or the countPtes they represent. Produced by the Policy Research Dissemination Center Perverse Effects of a Ratings-related Capital Adequacy System by Patrick Honohan World Bank Development Research Group, The World Bank. Author's e-mail: phonohan@worldbank.org. I am grateful to Jerry Caprio and Giovanni Majnoni for helpful comments. Perverse Effects of a Ratings-related Capital Adequacy System by Patrick Honohan World Bank 1. Introduction It has recently been proposed' that banks should be allowed to hold less capital against loans to borrowers who have received a favorable rating by an approved rating agency. The objective is an improved allocation of bank capital and risk, and lower exposure to bank failure. But a plausible model of rating agency behavior shows that this strategy can have perverse results, actually increasing the risk of deposit insurance outlays. A number of different critiques are relevant.2 For instance, while capital is intended to cope with unexpected loan-losses, ratings are more strongly related to expected default rates, than to the standard deviation or risk of default (and of course they don't take account of correlations between different loans in the portfolio).3 There is also the question of how important it is for rating agencies to retain credibility by posting ratings for bank loans that will prove to be expost accurate. After all, in contrast to the situation with rating tradable bonds (where rating agencies mediate the conflicting interests of investors and issuers), bank and borrower share an interest in reducing the burden of regulatory capital by securing a favorable rating for the borrower. Furthermore, one may ask just how reliable the rating agencies would be, especially in developing countries where information costs are high and where the ratings profession is in its infancy if it exists at all. Will there be issues of market power and concentration? In the June 1999 consultative document "A New Capital Adequacy Framework" of the Basel Committee on Banking Supervision. 2 There have been a number of other recent critiques of the incentive structure facing rating agencies, cf. Kuhner (1999), Partnoy (1999). For an overview of the regulatory use of rating agencies, see Adams et al. (1999). 3 While standard deviation of default percentages from B rated securities in the US is 3.8 times higher than those of AA rated securities; the mean default percentage of B rated securities is fully 11.7 times higher than that of the AA rated. Note also that the Basel proposal does not closely tie the differential capital requirements to varying default percentages. For example, lending to corporates rated A are shown in the proposal as requiring 5 times the capital required for those rated AA, whereas historical experience indicates a mean default rate only 75 per cent higher and a standard deviation only 60 per cent higher. Cf. Adams et al. (1999), Caprio and Honohan (2000). 2 But our purpose here is to abstract from most of these questions to focus on the incentive structure in a model of competitive banking and rating, with free entry and substantial access by market participants to information. We show that even here there can be perverse effects from the introduction of more liberal capital requirements for lending to firms that receive favorable ratings. We outline a model of bank lending under uncertainty to borrowers of different abilities. Well-informed and competitive banks comply with minimum capital requirements. They have access to an infinitely elastic supply of deposits, and these deposits are insured by a public insurer, whose expected payout will be the focus of our attention. The capital requirements are differentiated according to whether the borrower has been (favorably) rated or not. The expected default percentage of rated borrowers is set to satisfy a certain threshold by rating agencies. Borrowers of different abilities choose the risk profile of their projects based both on systemic and specific risk factors. Low ability borrowers may choose a safer profile in order to secure a rating. Our model reveals three main underlying problems. First, an issue of signaling, whereby low-ability borrowers may alter their behavior in order to secure a favorable rating and hence a lower capital requirement for their borrowing, even though they continue to place banks, and the deposit insurer, at risk. Second, the establishment of a regulatory cut-off may actually reduce the amount of risk information made available by raters. Third, even though rating agency credibility may require the average default percentage of rated companies to correspond to understood thresholds in normal times, they will not normally have an incentive to factor-in the risk of unusual systemic shocks, but it is these that cause the major surges in fiscal (deposit insurance) outlays. The paper is organized as follows. Section 2 presents the risk and return opportunities of different classes of borrower. Section 3 describes the debt contract and the assumptions regarding the rating agencies. Section 4 analyses the banks' decisions, explores the conditions under which banks may fail, defining the efficient capital requirement, i.e. the minimum requirement to avoid failure and deriving the competitive equilibrium interest rate. Section 5 explores the impact of ratings-related capital requirements on contracts and behavior. Section 6 provides two specific examples to confirm that perverse results can arise. Leaving the model, section 7 provides a brief intuitive discussion of the nature and source of the distortions. Section 8 concludes. 2. Risk and return for the borrowers Our model distinguishes between systemic risk - associated with the potential for bank failures, and specific risk. Systemic risk is modeled by assuming that there are two possible states of the world s, s =0 is the good state and and s = 1 is the bad state. The probability of the bad state or systemic crash Prob {s = 1 }= xr. Each period is a one-shot event with no consequence for subsequent periods. 3 The entrepreneurs in the model are risk-neutral, and each entrepreneur k is endowed with a specific ability or type ak s [ao,a,]. Each can choose risk parameters Uk> 0 and vk, with -uk < Vk. The entrepreneur must invest $1 in the project, which she does not have and so must borrow. (We need to have different abilities/types because otherwise all entrepreneurs will behave the same way and there will be no reason for rating and capital differentiation) The gross return on the project to an entrepreneur of ability a and chosen risk parameters u, v is aX(u,v,s) if the project succeeds, which happens with probability p(u,v,s); otherwise the project fails and the return is 0. (Note the implication thatpiaL. For a given repayment contract D, type H will choose lower risk: u(aH;D) < u(aL;D) and v(aH,D) < v(aL;D). Assuming that these abilities can be observed, the contracts will be differentiated and the no-profit condition will determine two contracts DH < DL. We now introduce the concept of having capital requirements depend on ratings. Specifically, we postulate that the regulator sets the capital requirement in respect of a borrower who has received a favorable rating at a lower rate yo < Yi. Further, suppose that these regulatory capital requirements are the efficient levels corresponding to types H and L respectively, i.e. yo (yr) is the rate determined by equation (6) above for the risk choices made by H (L). There may not be an exact correspondence between ability and rating. In particular, the prospect of a better loan contract may induce low ability borrowers to choose a risk profile so cautious that it secures them a favorable rating. The bank will not be fooled by this, but it too will take advantage of the lower capital requirement. Thus we need to take account of four different contracts D(a,y) satisfying the (competitive) no-profit condition P; they correspond to the two levels of ability and the two capital requirements. We may ignore the high capital high ability contract, which only becomes relevant if the cost of rating is so high that even high ability borrowers choose not to be rated, in which case the rating agencies play no further role. The other three contracts are D(aH,yo) < D(aL,yo) < D(aL,yi). The middle one is the interesting and problematic contract offered to low ability borrowers L who have obtained a rating. Since the loan can qualify for the lower capital requirement, the bank can break-even with a lower-interest contract than the no-profit equilibrium contract for low ability entrepreneurs and high capital. But use of this middle contract places the deposit insurer at risk. Since yo(yL) are the efficient capital requirements for ability level H(L), the system is no longer sure to satisfy the safety condition F unless the L borrowers do not find it desirable to obtain a rating. If they do get a rating, and borrow at the middle contract D(aL,yo), the lower capital yo will be insufficient in the bad state to protect any bank which has lent to them.4 Here is where the role of the rating agencies comes to the fore. Recall that they can acquire information about the borrowers at cost c. Will this cost be sufficiently small, from the L borrower's point of view, to be worth paying in terms of the lower repayment? 4The entrepreneur will respond to the lower interest rate by lowering risk, but not by enough to avoid violating the safety condition F. Furthermore, we can see from (2) above that the cost c incurred by the borrower in obtaining the rating will tend to increase the risk level chosen by the borrower. 7 From the assumption (made above) that the rater's reputation requires unbiased default probabilities for rated borrowers, the mean5 default rate of rated firms will be no greater than the externally targeted rate OoŽOH. Denoting the proportion of low-ability firms by A, and the choice of default rate of high ability firms by OH then the low ability firms can be rated provided their default rate OL satisfies (1-X)OH + XOL < So; i.e. if _0 _ )H>0 (7) OL < 0, = S (1A ),S 7 Thus, provided the high ability borrowers are choosing a default rate strictly lower than the target rate So, the low ability borrowers will in practice be set a more lenient target 01> Oo. However, lending to the low ability borrowers with low capital will be an equilibrium only if the best contract satisfying the no-profit condition gives the low ability borrower a better return (net of the costs of achieving a rating, and taking account of the impact on gross return of the risk choices made to secure the rating) than does the best high capital contract. We now examine the conditions under which this will be the case. To get a rating, the low ability entrepreneurs will have to choose risk parameter values satisfying either (8a) or (8b), depending on whether ratings are based on specific risk alone (good state) or on unconditional risk: I - u < 01 (8a). 1-u - 7ZV < 01 (8b). Subject to this, the borrower will choose the most favorable balance between specific and systemic risk u and v. For example, at an interior solution from a continuous range, the chosen u,v will satisfy a first-order condition corresponding to (2): Xv,- D, = 7E(X- Dy) (8') Writing the choice of u and v satisfying the rating-constrained contract (8a or b), (8') as u(aL,D(aL,yo),01), v(aL,D(aL,yo),01), the condition for the low ability entrepreneur to opt for a rating is: { 1- 01} {azlXu(aL,D(aL,yo),01), v(aL,D(aL,yo),01)] - D(aL,yo) - c} > {1 - 6(aL,D(aL,yl))} {aLX[u(aL,D(aL,yl)), v(aL,D(aL,yl)] - D(aL,yl)} (9) Evidently, this depends sensitively on the pay-off function X as well as on the loan contract terms. In two special cases discussed in the next section, we see that the low ability borrower L may opt for rating, and how this can work out to the disadvantage of the deposit insurance fund. 5 As discussed below, this may be either the unconditional default rate, or the default rate conditional on the good state. 8 6. Two specific examples In order to provide a more precise and explicit illustration of what can go wrong when lending to rated firms requires less capital, this section examines two special cases for the return function X(u,v), still where there are just two ability types aL < aH, (and where, for simplicity, the resource cost c of lending is negligible). Both examples exploit the fact that, unless the capital requirements are high enough to avoid bank failure in the bad state, the choice of higher v entails higher expected deposit insurance outlay, without having any effect on the bank's return. Special case (i) First suppose that X(u,v) only depends on the default probability 0 = u+xrv, and that there are just two possible choices of 0: 00 < 01, (though firms can freely choose u, with v deduced along the corresponding trade-off). In this special case we also assume that, as discussed above, the rating agency rates firms based only on their "good state" default probability u. Provided the safe choice So gives the better expected gross return (i.e. before repayment to the banks), there will then exist a range of values for D such that L will choose high risk 01 and H will choose low risk 00. (This can be seen from Figure 1 which plots the net return 0,(X(0,) - D/a) against Dla for each of the two values of i). The crossover point defines the critical value d*. Then for values of D such that: - < D <-, aL aH the low ability borrower will continue to choose high risk, while the high ability borrower chooses low risk. By equation (5') above a restrained (high capital) banking system will not, therefore offer the same contract to both types. But a bank which is allowed to maintain low capital against a rated borrower will be happy to offer a low D, based on a low u, even if that borrower retains high risk by shifting between u and v retaining the overall default rate 0 while increasing the size of the deposit insurer's loss in the bad state. Special case (ii) Alternatively, suppose there are two possible values for each risk parameter uo < ul and vo < v,, giving four alternative risk choices. Suppose that any risk choice other than (ul,vl) qualifies a borrower for a rating. The potential social impact, of a lighter capital requirement being allowed for rated borrowers depends on whether the ligher capital requirement induces low ability borrowers to switch to vl. If a low ability borrower chooses uo in order to get a lower D offered by the bank, as a result of the no-profit condition (5), can this result in the 9 borrower switching from vo to vl, thereby increasing the expected outlay of the deposit insurer? The answer is yes. Figure 3 indicates the different configurations that are possible. Writing X(u,,vj) as Xv, etc., it plots the expected return OU,(Xt-D) to the borrower, depending on the risk choices and the terms of the lending contract D. For any given D, the borrower chooses the highest value - the outer envelope. The solid lines thus show the choices involving high systemic risk. The plots in both panels satisfy Xi I> Xio > XoI > Xoo and Oil> Oio > 001> 000. In the upper panel, the borrower's preferred choice of systemic risk is always v, and does not depend on D. (This will happen, for example, if the probability 7r of the bad state is very low). But, as illustrated by the lower panel of Figure 3, there can be a range of values of D which induce low systemic risk vo whereas for lower values of D high systemic risk v, is chosen.6 What is happening here is that, if the gain in return for higher systemic risk v is sufficiently higher when specific risk u is lower, then a more favorable loan contract (low value of D) could induce the low ability firm to opt for higher systemic risk vi, while still retaining its rating by switching to the low value of specific risk uo.7 Thus, in this example, by lowering the competitive repayment contract D offered by banks to rated borrowers, the introduction of ratings-related capital requirements actually increases the systemic risk v and hence the expected outlay of the deposit insurer. 7. Discussion: distortions from the rating system We now see the nature of the costs potentially introduced by the rating system. First is the fiscal (deposit insurance fund) cost incurred in the bad state when banks fail because they offered the low-ability borrowers the contract based on low capital D(aL,yo). This contract may have been good for the bank ex ante, but embodies too little capital for the bank to survive in the bad state. The level of capital yo is the efficient capital requirement for firms of type H, and is not enough to protect the bank from failure if it lends to firms of type L, even if those firms have adjusted their risk levels to satisfy the target failure rate. Second, we note that the establishment of a regulatory cut-off may actually reduce the amount of risk information made available by the raters. The rater could announce the precise value of u and v for each borrower, but doing so will reduce the number of borrowers who can get a favorable rating in the sense that we have used it, i.e. falling within a group whose average risk of default is no greater than the regulatory threshold. 6 The highest values of D induce high systemic risk also, but for these values the borrower remains unrated. 7 Incidentally, the lower specific risk will also help lower the value of D, according to (5) above. 10 Third, there is the consideration of the real economic resources c employed in the rating. Not only is this a direct social cost, but it also induces riskier behavior by all borrowers who pay it, regardless of ability. A fourth possible bias arises if the rating agency bases the ratings on specific risk only. It is clear that this might be relevant when we consider that neither the bank nor the firm has any interest in ensuring that the rating agency is actually giving an accurate rating. Rather the contrary, as both stand to gain from a favorable rating. The situation here is different to that in bond market ratings, where the rating agency mediates the conflicting interest of issuers and investors, and thus the agency's credibility depends on their ratings having predictive power. To be sure, we have assumed that some such credibility factor remains relevant for rating agencies in the banking sphere, perhaps because they need to retain their regulatory approval. Therefore we assumed that ratings are not systematically biased. But rare systemic crises (the "bad" state) may not be taken into account in this attempt to avoid being systematically biased. In particular, the agency could choose to ignore risk choices v and to give a rating to firms with u < 01 < u+7rv, i.e. firms whose default probability conditional on the good state is adequate, but whose unconditional default probability is below the regulatory threshold. The average default rate of such firms will not exceed u except and until the bad state is realized - at which date there will be a banking crisis anyway. If the rating agency's franchise extends only to the next banking crisis anyway it has nothing to lose by awarding ratings to the less demanding standard. 8. Conclusion Allowing banks to lower capital backing for loans made to highly-rated borrowers may result in a rating system that neither reveals risk information concerning borrowers, nor protects the deposit insurance fund. Part of the problem is the whole idea of basing portfolio risk-evaluation on the sum of individual loan risks, but there are also important incentive issues involved. Our model formalizes the idea that the largest outlays of deposit insurers are chiefly in times of systemic crisis, and gives reasons why use of agencies' individual ratings is unlikely to be an effective early-warning system for the risk of systemic failure. As such, the use of ratings is likely to lull policymakers into a false sense of security. While harnessing market information for improving bank safety is an important and valuable concept (for example, by increasing the role of large, well-informed but uninsured claimants), this particular approach could be counterproductive. Reliance on ratings could induce borrowers to increase their exposure to systemic risk even if they reduce exposure to specific risk. This danger highlights the hidden dangers of allowing banks to hold less capital against rated borrowers. 11 References Adams, Charles, Donald J. Mathieson and Garry Schinasi (1999), International Capital Markets: Developments, Prospects, and Key Policy Issues (Washington DC: IMF). Caprio, Gerard and Patrick Honohan (2000), "Reducing the Cost of Banking Crises: Is Basel Enough?", Presented at the AEA Meetings, Boston, January. de Meza, David and David Webb (1999), "Wealth, Enterprise and Credit Policy", Economic Journal, 109, April, 153-163. Kuhner, Christoph (1999), "Rating Agencies: Are They Credible", University of Munich, mimeo http://papers.ssrn.com/paper.taf.'ABSTRACT ID=139412) Partnoy, Frank (1999), "The Siskel and Ebert of Financial Markets: Two Thumbs Down for the Credit Rating Agencies", Washington University Law Quarterly, 77, 619-712. 12 D l/(l-u) , ..~~~~~~~~~~~~~~~~~~2 Y* Y Figure 1: Equilibrium interest contract for loans of risk level u,v, as a function of capital requirements y. (1)(X(O)-D/a) (b-00)X(00) (L-00)X01) d* X(80) X(01) D/a Figure 2: Special case (i) -- Choice of risk level where only overall default risk matters. Only for low values of D/a will firms choose safety; therefore at a given value of D, only sufficiently high ability firms will choose safety. 13 O(X-D) (1-001)(Xol-D) , (I-°lo )(Xio-D) a (lI-oo)(Xoo-D) Xoo Xo I X1' D 6(X-D) X (l-01)(XoX-D) (I4z)4z(101 ) -D) (1-0oo)(XoooD144 \ XOO Xoi Xio xI,, D Figure 3: Special case (ii)- Choice of specific and systemic risk as D varies. Lower panel illustrates possible switch from low risk vo to high risk v, as D declines. 14 Policy Research Working Paper Series Contact Title Author Date for paper WPS2340 Currency Substitution in Latin Pere Gomis-Porqueras May 2000 M. Puentes America: Lessons from the 1990s Carlos Serrano 39621 Alejandro Somuano WPS2341 The Tyranny of Concepts: CUDIE Lant Pritchett May 2000 R. Widuri (Cumulated, Depreciated Investment Effort) Is Not Capital WPS2342 What Can We Learn about Country Martin Ravallion May 2000 P. Sader Performance from Conditional 33902 Comparisons across Countries? WPS2343 Ownership and Performance of David A. Grigorian May 2000 D. Brown Lithuanian Enterprises 33542 WPS2344 Designing Direct Subsidies for Vivien Foster May 2000 S. Delgado Water and Sanitation Services: Andr6s G6mez-Lobo 37840 Panama-A Case Study Jonathan Halpern WPS2345 Information and Modeling Issues in Andres G6mez-Lobo May 2000 S. Delgado Designing Water and Sanitation Vivien Foster 37840 Subsidy Schemes Jonathan Halpern WPS2346 The Middle Class Consensus and William Easterly May 2000 K. Labrie Economic Development 31001 WPS2347 Terror as a Bargaining Instrument: Francis Bloch May 2000 P. Sader A Case Study of Dowry Violence in Vijayendra Rao 33902 Rural India WPS2348 Taxing Issues with Privatization: Jack M. Mintz May 2000 G. Chenet-Smith A Checklist Duanjie Chen 36370 Evangelia Zorotheos WPS2349 Trade, Foreign Direct Investment, Kamal Saggi May 2000 R. Bonfield and International Technology 31248 Transfer: A Survey WPS2350 Multilateral Trade Liberalization and Maurice Schiff May 2000 L.Tabada Political Disintegration: Implications 36896 for the Evolution of Free Trade Areas and Customs Unions WPS2351 Environmental Policy and Time Peter W. Kennedy May 2000 Y. D'Souza Consistency: Emissions Taxes Beno^it Laplante 31449 and Emissions Trading Policy Research Working Paper Series Contact Title Author Date for paper WPS2352 How Stronger Patent Protection in Carsten Fink May 2000 L. Tabada India Might Affect the Behavior 36896 of Transnational Pharmaceutical Industries WPS2353 The Sao Mateus-Jabaquara Jorge Rebelo May 2000 S. Van Veldhuizen Trolleybusway Concession in Brazil Pedro Machado 38722 WPS2354 When the Bureaucrats Move out of Yi Chen May 2000 M. Yafi Business: A Cost-Benefit Ishac Diwan 34649 Assessment of Labor Retrenchment in China WPS2355 Greed and Grievance in Civil War Paul Collier May 2000 A. Kitson-Walters Anke Hoeffler 33712 WPS2356 Bureaucratic Delegation and Political Philip Keefer June 2000 P. Sintim-Aboagye Institutions: When Are Independent David Stasavage 37644 Central Banks Irrelevant? WPS2357 Evaluating Carbon Offsets from Kenneth M. Chomitz June 2000 J. Ancrum Forestry and Energy Projects: 33512 How Do They Compare? WPS2358 Why Infrastructure Financing Daniela Klingebiel June 2000 M. Salehi Facilities Often Fall Short of Their Jeff Ruster 37157 Objectives WPS2359 Reducing Carbon Dioxide Emissions Will Martin June 2000 L. Tabada Through Joint Implementation of 36896 Projects WPS2360 Corruption and the Composition of Beata K. Smarzynska June 2000 H. Sladovich Foreign Direct Investment: Shang-Jin Wei 37698 Firm-Level Evidence WPS2361 The Welfare Effects of Private Sector George Clarke June 2000 H. Sladovich Participation in Guinea's Urban Water Claude Menard 37698 Supply Ana Maria Zuluaga WPS2362 A Transitory Regime: Water Supply Claude Menard June 2000 H. Sladovich In Conakry, Guinea George Clarke 37698 WPS2363 Would Collective Action Clauses Barry Eichengreen June 2000 S. Kpundeh Raise Borrowing Costs? An Update Ashoka Mody 39591 and Additional Results