The Reliability of Credit Risk Models

At a lunchtime meeting organised by the Centre for Economic Policy Research, Professor William Perraudin, Professor of Finance at Birkbeck College London and Special Advisor to the Bank of England, considered the reliability of the current generation of credit risk models.

A credit risk modelling revolution is going on in major international banks. In the 1980s, banks became more sophisticated in their hedging and pricing of interest rate risk. This boosted efficiency in government bond markets and improved financial stability as banks were better able to manage maturity imbalances in their books. Now, new modelling methods are changing the way banks understand and handle credit risk. 

What are the implications? First, being able to measure risk permits one to allocate scarce economic capital more precisely. So, the new modelling techniques have led some banks to retreat from capital-hungry lending activities. For others, one may expect that loan pricing will increasingly be determined by hurdle rates of return driven by credit risk and capital allocation models.

Second, discrepancies between regulatory capital charges and the levels of economic capital that banks themselves would prefer to hold are increasingly obvious when risks can be quantified and compared. Regulatory arbitrage (in which banks engage in cosmetic transactions solely designed to reduce regulatory capital) is not a new phenomenon; but the wide-spread use of credit risk calculations and hurdle rates based on return on equity has certainly encouraged its development.

Should policy-makers ‘go with the flow’, as they did with market risk on trading books, letting banks use their own internal risk management models to calculate regulatory capital? Should senior bankers place faith in credit risk models, allowing them to determine how economic capital is allocated?

Perraudin gives his personal views on the reliability of the current generation of credit risk models and the scope they offer for the more efficient allocation of regulatory and economic capital. Drawing on an extensive programme of research on credit risk, he argues:

• that commonly employed models are insufficiently conservative especially when applied outside the US market
  

• that detailed bottom up calculations which inevitably leave out importance sources of risk must be supplemented with top down measures and prudent supervisory and managerial judgements
  

• that some of the crucial ingredients of credit risk models, in particular agency ratings, should be viewed in a constructively sceptical way.

Notes for Editors:

The Centre for Economic Policy Research is a network of over 500 researchers based throughout Europe, who collaborate through the Centre in research and its dissemination. CEPR helps its Research Fellows and Affiliates develop projects, obtain their funding, administer them and disseminate their results. The Centre’s research ranges from open economy macroeconomics to trade policy, from the economic transformation of Central and Eastern Europe to regionalism in the world economy. The views expressed in CEPR publications and meetings are those of the authors and speakers. CEPR takes no responsibility for these views, and does not take any institutional policy positions. 

For further information about CEPR, please contact Rita Gilbert, Tel: (44 20) 7878 2917 or email: rgilbert@cepr.org.

The research underlying this presentation was supported by a ROPAs grant from the ESRC.