‘Capital requirements and the behaviour of banks’

‘Capital requirements and the behaviour of banks’

C. Mayer. Comments 1171 Comments ‘Capital requirements of banks’ by Jean-Charles Rochet and the behaviour Colin Mayer Affiliation?, City, Country...

258KB Sizes 2 Downloads 47 Views

C. Mayer. Comments

1171

Comments

‘Capital requirements of banks’ by Jean-Charles Rochet

and the behaviour

Colin Mayer Affiliation?, City, Country

The determination of capital requirements for European banks has become a subject of particular concern to regulators involved in the integration of financial markets and the creation of a single European currency. Different forms of bank regulation conflict in a single market in which banks can freely trade and establish branches across borders. Problems of coordination of regulation have been addressed through international negotiation; a greater degree of consensus has emerged in the imposition of solvency ratios for banks than in most areas of financial regulation. The Basle Accord specified risk based solvency ratio requirements that have in large part been followed by the European Community. This paper is concerned with the reasons for and the effects of capital requirements. It has the commendable feature of bringing rigorous theoretical analysis to bear on an important policy area without resorting to the ad hoc assumptions that have afflicted most of the literature. The analysis is in three stages. In the first stage, the paper discusses a ‘complete markets’ model of banks with the possibility of bankruptcy occurring. The second stage is a portfolio model without bankruptcy and the third stage combines the first two in a portfolio model with bankruptcy. Stage One - the complete markets model For bankruptcy to have any effect on bank behaviour in a complete markets model, it is necessary to assume that interest rates fail to reflect fully risks of bankruptcy. Bankruptcy makes the bank’s pay-off function convex. As a consequence, for investments of given mean return, the equity value of banks is greater the higher is the level of risks of a bank’s assets. Capital requirements are neither necessary nor sufficient for reducing risk taking by banks in a complete markets model. They are not sufficient

1172

C. Mayer, Comments

because increasing capital does not alter bank behaviour: it is always optimal for a bank to choose the highest risk investments. They are not necessary because properly priced insurance internalizes the externality associated with returns in bankrupt states. Stage Two - a portfolio model without bankruptcy Relaxing the complete markets assumption, it is necessary to consider the relation between risk and return. Since the model takes net worth as given, this can be done in the context of an asset pricing model. Applying a Capital Asset Pricing Model, for example, the effect of a capital requirement can be viewed as a shift of the bank’s chosen portfolio along the efficient market line (KM in fig. 4 of the paper) towards the riskless asset (K). Provided that the chosen portfolio remains on the market line, there is an unequivocal reduction in the risk return ratio as measured by the ratio of standard deviation to mean returns. However, if by imposing capital requirements the relative price of different risky assets is changed then the risk return ratio may increase, as shown by the move from B to C in fig. 5 of the paper. Whether this happens depends on the balance between income and substitution effects: with increasing relative risk aversion the default probability of low net worth banks is higher. The paper therefore correctly concludes that it is important to ensure that risk weights in capital requirements are ‘market based’. The solvency ratios that are specified in the Basle Accord have gone someway towards imposing risk based capital requirements. However, they only account for credit risk and not for market risk arising, for example, from interest and exchange rate fluctuations. Furthermore, the credit risk requirements do not come close to approximating true market risk measures. A major obstacle to imposition of market based risk requirements is the information which is required. Market values of assets of mutual funds, brokers and dealers can in general be measured. However, information theories suggest that a primary rationale for the existence of banks is that they can acquire information and monitor firms at lower cost than outside investors. The market values of bank assets are not therefore readily observable. Thus while risk based capital requirements can be imposed on some non-banks, they cannot readily be determined for banks. Indeed, market based capital requirements are imposed on brokers and dealers in the U.K. but not on banks. This recommendation is therefore more applicable to non-banks than banks. Stage Three - a portfolio model with bankruptcy With bankruptcy,

the relation between utility of shareholders

and asset

C. Mayer, Comments

1173

returns is no longer uniformly convex or concave. For low values of capital requirements, banks will tend to choose risky investments for the reasons described in Stage One. However, this will be deterred at higher values of capital requirements because of risk aversion. Combining this with the results of Stage Two suggests a rule that capital requirements should be set according to market risk weights and at a sufficiently high level to deter choice of risky assets. However, as Stiglitz and Weiss have noted in the context of credit rationing, in the presence of diminishing risk aversion, the imposition of high capital requirements may lead to the selection of agents with lower levels of risk aversion and therefore preferences for higher risk investments. The above description makes clear that capital requirements are viewed in this analysis as playing an incentive role. Recent literature on capital structure suggests that it is difficult to consider firm financial policy in this way. The reason is that incentives can be provided in other forms without any implication for chosen financial structure. Another way of putting this is that this model does not appear to have much to do with banks. It could be applied to any firm where there is a risk of default and debt may not be correctly priced. It does not address the issues that are most commonly considered in the context of bank regulation, namely risks of runs and contagion. In that respect, the complete markets analysis is curious. If there were complete markets, then there would be no risk of runs: liquidity demands could be specified and liquidity withdrawals could be distinguished from insolvency. A complete markets assumption is not attractive in the context of a model of bank runs and contagion. Justification for the imposition of capital requirements usually comes from a combination of externalities across banks through risks of contagious bank failures and the fact that banks play important functions in offering payments services and maturity transformation of short-term deposits into long-term loans. Widespread bank failures are therefore unacceptable. Provided that they have adequate financial resources, banks have incentives to avoid contagious failures by rescuing failing banks themselves. Where they are unable to do so then liquidity may have to be injected through lender of last resort facilities and bail-outs. The former has monetary and the latter fiscal consequences. A desire to avoid the monetary and fiscal consequences of bank failures leads to the imposition of liquidity and capital requirements on banks. Capital requirements are therefore determined by considerations of systemic rather than individual bank risk. This single bank model is therefore unable to take account of factors which appear most relevant to the determination of optimal capital requirements.

1174

R. Repullo, Comments

To summarize, this is a careful analysis of the risk reducing function of capital requirements. However, both the structure of the model and the recommendations suggest that it is more relevant to the regulation of nonbanks than banks.

Comments

‘Capital requirements of banks’ by Jean-Charles Rochet

and the behaviour

Rafael Repullo CEMFI,

Madrid, Spain

The purpose of Jean-Charles Rochet’s paper is to contribute to the recent academic debate on the effects of capital regulation on risk-taking behaviour by banks. The paper considers a simple static model of bank behaviour under two alternative assumptions about the underlying market structure, namely complete and incomplete markets. The complete contingent markets assumption poses immediately the question of why should there by any financial intermediaries (such as banks) in the first place. Since this issue is not addressed in the paper, I am not sure about how should one take Propositions l-4. For this reason, I am going to focus my comments on the second part of the paper. Of course, the incomplete markets model is not without problems, since in this case banks’ profits in different states of nature cannot be aggregated into a single index, and so profit maximization is not well-defined. Thus, the paper asumes that the bank’s objective function is to maximize the expectation of a von Neumann-Morgenstern utility function u(.) that describes the preferences of the risk-averse owner-manager. The argument of this utility function is the bank’s final net worth, a random variable given by

R,(x) = j7.x + [ROD- C(D)] + K,(

1 + R,),

where XE [WNis the portfolio of the N risky securities, (random) vector of excess returns of the risky securities,

fi =r? - 1. R, is the ROD-CC(D) are the