22 Jan 2009
Amid all the column inches dedicated to the financial crisis, Viral Acharya and Julian Franks, believe some critical aspects have been overlooked.
Their latest research looks at capital budgeting and governance at banks and the adverse role played by government guarantees. It challenges banks to look again at how they set performance targets; whether these targets are appropriate in maximizing bank value and whether they appropriately reflect banks' business and financial risks; and how robust are the targets over the business cycle. Acharya and Franks' conclusions might unlock important and far reaching improvements in the quality of capital budgeting and governance practices at banks.
Investigating the causes and consequences of the banking crisis led us to six potentially important conclusions:
First, the required return on equity (ROE) of banks is not, and was not pre-crisis, invariant to their leverage and risks. In particular, it was not close to the flat 10 per cent ROE used by most banks as a basis for their capital allocation decisions until the Fall of 2008. We believe it was considerably higher. The measured equity betas of banks implied low or almost zero asset betas for their business risks. Such measures are simply not credible. Starting from any reasonable estimate of a bank's business risk, the cost of levered equity is unlikely to be below 20 per cent, the estimated cost of new preferred/convertible finance raised by Barclays in the Fall of 2008.
Second, banks should focus on return on assets (ROA) rather than on return on equity (ROE). In order to estimate the target ROA, banks should assess their business risks and financial leverage, and impute the related cost of capital. In the jargon of CAPM, banks should start by assessing and using "asset betas" to estimate "equity betas", and then the return on equity, rather than using equity betas to estimate asset betas. The latter practice creates the risk that during good times, banks (implicitly) employ low or virtually zero asset betas in their capital allocation decisions. Indeed, we believe that the usual models for estimating costs of capital do not apply for banks since they ignore the substantial discounts incurred while issuing equity in bad times, and importantly, embed the effect of government guarantees.
Third, explicit deposit guarantees and implicit guarantees on other loans lowered the pre-crisis cost of borrowing of banks to virtually risk-free levels and lulled banks into believing that this low cost reflected the low business risk of their asset portfolio. Such a low cost of borrowing made the cost of debt look relatively flat over a wide range of leverage, the cost of equity look high relative to the cost of debt, dissuaded banks from equity financing and resulted in excessive leverage.
Fourth, bank deposit guarantees should be charged for by regulators on a "marked-to-market" basis (that is, taking account of leverage and asset risk of the bank being guaranteed) to ensure fair pricing in bank lending decisions and to ensure that banks do not raise leverage beyond the optimum that is suitable for their business risk.
Fifth, conversely, banks should estimate and employ in their capital budgeting a cost of capital "without the guarantee". The "without guarantee" cost of capital is an important concept and measure because it helps answer the questions: What is the cost of bank capital relating to the underlying business risk and leverage, and what is the bank's true economic worth absent the guarantees? If a bank had value solely because of government guarantees, it would be insolvent if the guarantees were appropriately priced. It is essential for bank boards, top management and policy makers to know if this is the case.
Finally, these capital-budgeting issues have direct relevance even to bank assets that are outside the activities guaranteed by government. It is helpful here to state the simple point that the cost of capital of non-guaranteed lending and capital market transactions are substantially higher than that of the other activities that are subject to guarantees. We believe the failure to properly account for this distinction runs the classic risk of regulatory arbitrage where financial institutions get larger simply to fund high cost-of-capital activities with subsidized government guarantees. Indeed, we believe such arbitrage activity explains the emergence of large, complex financial institutions that have imposed systemic consequences on economies.
But what does this mean in terms of policy recommendations for banks? We believe banks need to consider four things:
1. Banks should measure their cost of capital "without the guarantee" in order to determine the extent of the guarantee's value and its impact on the capital structure of the bank and the amount and price of lending. This is an important metric when considering capital allocation decisions inside the guaranteed structures.
2. Banks should set a cost of capital that reflects the cost of capital in good times and bad times, and obtain better estimates of their effective equity beta, and thus of their cost of issuing equity capital in bad times (when equity is typically issued).
3. Banks should set a long-run target ROA - return on assets (debt + equity) - rather than short-run ROE since ROA better reflects the economic profitability of the bank's assets and long-run measurement captures better the true economic value of business decisions. They should also take account of leverage and a suitable cost of capital "without the guarantee" while setting the target ROA.
4. Banks should consider how their economic leverage and the business mix of lending and asset activities relate to their cost of capital. Banks should also analyse how interest rate spreads they charge on loans and other assets relate to their cost of capital and the target ROA.
Similarly, our primary policy implication for regulators is that the provider of bank guarantees should properly mark to market the value of these guarantees, taking account of asset risk as well as leverage of the bank being guaranteed, and charge a premium accordingly, and on a regular basis. A model here could be that of the pension regulator (in the UK) who charges different premium to each pension fund depending upon its risks.
Viral V Acharya (firstname.lastname@example.org) is Professor of Finance at London Business School and New York University Stern School of Business.
Julian Franks (email@example.com) is Professor of Finance at London Business School.
The authors prepared this memorandum for Knight Vinke Asset Management as part of that company's investigation into the causes and consequences of the banking crisis.