Exploring The Short- And Long-Run Links From Bank Competition To Risk - Reconciling Conflicting Hypotheses?

The subject of bank competition and risk has returned to the fore with the financial crisis of 2008-9, with a common view being that competition between financial institutions during the preceding boom was at the core of the crisis. This in turn implies that the benefits of banking competition for economic growth and efficiency need to be placed in the balance. On the other hand, there is an extensive literature, generally estimated on pre-crisis data, which finds conflicting results on the relation between competition and risk. This follows on the one hand the so-called "franchise value" or "competition-fragility" approach - that more competition reduces the value of a banking licence and requires firms to take more risk as a result; and on the other hand, the "competition-stability" view that with low levels of competition banks may charge excessively high rates of interest on loans and hence generate adverse selection and moral hazard on theirloan books. Both types of results have been found in the empirical literature.

In this paper, using a dataset of 6000 banks for the EU-27 covering 1998-2012, we break new ground by looking at the short run dynamics of the relation between competition and risk as well as the long run equilibrium relation, by taking account of a full 6 year period since the onset of the crisis in 2007, as well as a comparable period before it; and by comparing and contrasting results using two competition indicators, the H statistic (which measures the degree to which changes in input prices affect bank revenues) and the Lerner index (a measure of the banks' price-cost margin). The risk measure used is the Z score, which measures the number of standard deviations a banks' rate of return on assets must fall for the bank to become insolvent.

We find for both measures of banking competition that in the pre-crisis period, a change in competition entails an increase in risk. On the other hand there are conflicting results for the longer term impact of changes in competition, with the H statistic showing heightened competition diminishes risk, while for the Lerner index the long run effect of heightened competition is to increase risk. Testing for the reason for differences in long run effects we find that the H and Lerner differ in their impact on the volatility of profits, a key input to bank risk.

In terms of policy, the work consistently suggests that considerable caution is warranted by regulators in the initial period after a rise in competition, since the indicators show consistently that a rise in bank risk accompanies it. On the other hand, in the longer term, there is a need for caution in drawing policy conclusions from risk-competition studies without careful consideration of the likely impact of a given policy shift, given the conflicting results for H and Lerner. So for example it needs to be considered what effect policies such as separation of retail and wholesale banking or certain macroprudential policies will have on bank margins (as shown by Lerner) as opposed to pass through of input prices to revenue (shown by the H statistic). An impact directly on margins is shown to be deleterious to risk, while enhancing pass through generally tends to enhance soundness.

-Ends-

For further information contact:

Brooke Hollingshead, Press Officer
T: +44 (0)207 654 1223

E Philip Davis, Principal Research Fellow
T: +44 (0)20 7222 7665
E: P.Davis@niesr.ac.uk