Abstract

This paper investigates the relation between credit and market risk over long investment horizons. We split credit risk into transition and spread risk so that results can be directly related to ratings-based credit risk models which adopt this decomposition. We find that spread risk for high credit quality exposures exhibits variable but generally negative correlation with interest rate changes. For low credit quality spreads, the correlation is markedly negative. Transition risk is also negatively correlated with interest rate changes in that VaRs are distinctly higher when calculated using a transition matrix based on years of data in which interest rates fall.

Introduction

VaRs for Market and Credit Risk

Since the mid-1990s, banks have made extensive use of VaR models for measuring and controlling the market risk of their trading portfolios. They have been encouraged in this by regulators who, since the 1997 Basel Accord Amendment on market risk (see Basel Committee on Banking Supervision (1996)) have permitted use of these models for the calculation of regulatory capital for trading books.

Recently, there has been much interest in a new generation of portfolio management models designed to measure the risks associated with portfolios of credit exposures. The frameworks proposed by JP Morgan (1997) (Creditmetrics), Credit Suisse Financial Products (1997) (CreditRisk+), and by the consulting firms KMV and McKinsey's have been widely discussed and analyzed. Although they may be employed for different purposes (for example, portfolio allocation), the most common use of these models is to generate VaR estimates for credit-sensitive portfolios.