RDP 1999-09: Australian Banking Risk: The Stock Market's Assessment and the Relationship Between Capital and Asset Volatility Introduction

The stability of the banking sector has long been a concern of public policy. Recent events in Asia, and elsewhere, have emphasised the importance of the two-way interaction between the soundness of the banking sector and the health of the macro economy. The potential public-sector liability that may arise if a bank fails is also of concern. The nature of the public-sector liability is most transparent in countries with explicit deposit insurance or deposit-guarantee structures in place. However, Kane and Kaufman (1992) argue that even in Australia, where no deposit insurance is provided, the possibility of claims on the government is a legitimate concern of public policy.

Estimates of the probability of failure of an individual bank are useful both for individual-bank supervision and in forming an assessment of the overall stability of the financial system. If an institution fails, depositors in that institution may lose funds, and in particular circumstances, the failure could cause difficulties for other financial institutions or turmoil in financial markets. The likelihood of a bank becoming insolvent, within a given period of time, is a function of the variability in its income and its ability to withstand losses in the short run. These two determinants depend, in turn, on the volatility of the rate of return on bank assets and the bank's level of capital, respectively. Estimates of asset volatility and capital ratios can be computed from banks' annual reports. However, the backward-looking nature of accounting measures means that these figures are unlikely to correspond to the relevant economic concepts (i.e. discounted future flows of economic earnings). In principle, one way of overcoming this problem is to use market values as proxies for economic values. Unfortunately in this case, this solution is not available because most bank assets and liabilities are not regularly traded and, therefore, lack an observable market price.

However, for many Australian banks one piece of market-based information is available and readily observable: the value of shareholder equity. In an efficient share market, the market capitalisation of a firm reflects the difference between the economic (or market) value of its assets and liabilities. Thus, a model of the relationship between the market capitalisation of the firm and its economic assets and liabilities can be used to infer those economic values. Similarly, the volatility of the market capitalisation reflects the unobservable volatility of the bank's economic assets and liabilities, again suggesting the possibility of inferring one from the other. These measures need to be viewed with caution since they depend on the assumption of market efficiency. If the share market is inefficient then the firm valuation and risk measures based on this valuation will be inaccurate.

Gizycki and Levonian (1993) used such a model to calculate the economic values of asset volatility and capital ratios for 11 Australian banks between 1983 and 1993. They found that the estimated capital ratio for the banking sector rose over this period, while there was no noticeable increase in the asset volatility of banks. They also concluded that banks with more volatile assets tend to maintain higher capital ratios and that there is a positive relationship between the two variables over time.

Since 1993, a number of developments that have occurred in the Australian banking sector suggest that an update of the Gizycki and Levonian (1993) study is warranted. Firstly, five more banks have listed on the Australian Stock Exchange. Secondly, the regulatory requirements applied to banks since the late 1980s have changed considerably. Any effect these changes have had on the relationship between banks' capital ratios and their asset volatility will be made clearer over the longer sample period. Thirdly, risk-management techniques have become more sophisticated, giving rise to the possibility that the interaction between banks' leverage and asset volatility may have changed. When the sample is extended, we find that both banks' asset volatility and capitalisation grew strongly during the 1990s. We also find that the net effect of the growth in these two variables has been a decline in the overall riskiness of Australian banks and, thus, a more stable financial system.

In addition to updating the results of Gizycki and Levonian (1993), this paper contains three key extensions. Firstly, call options on banks' shares are used to generate alternative estimates of the risk measures. Secondly, the probability that each bank will close (based on our estimated model) is used as a further measure of risk. Thirdly, an attempt is made to unearth the direction of any causation between a bank's asset volatility and its capital ratio. The introduction of the 1988 Capital Accord marked a movement from flat rate to risk-based capital adequacy requirements. Beyond any shift in the overall level of capital, this change in the regulatory regime altered the incentives facing banks. We, therefore, consider the impact of the 1988 Capital Accord on the capital-risk relationship and test for asymmetry in that relationship.

The paper proceeds as follows. Section 2 outlines the contingent-claim model that is used to estimate each bank's capital ratio and volatility of assets, the probability that each bank will close and the expected creditor losses in the event of bank closure. In Section 3, the data, assumptions and the estimation technique used are outlined. The results of the analysis are then presented in Section 4. In addition to presenting average levels of capital ratios, asset volatility and probability of failure we consider the probability distribution of the likelihood of losing a giving share of the banking system, which is taken to be an indicator of overall system stability. Section 5 provides a theoretical and empirical discussion of the relationship between asset volatility and capital. Section 6 concludes.