Estimating Bank Trading Risk: A Factor Model Approach

Risk in bank trading portfolios and its management are potentially important to the banks' soundness and to the functioning of securities and derivatives markets. In this paper, proprietary daily trading revenues of 6 large dealer banks are used to study the bank dealers' market risks using a market factor model approach. Dealers' exposures to exchange rate, interest rate, equity, and credit market factors are estimated. A factor model framework for variable exposures is presented and two modeling approaches are used: a random coefficient model and rolling factor regressions. The results indicate small average market exposures with significant but relatively moderate variation in exposures over time. Except for interest rates, there is heterogeneity in market exposures across the dealers. For interest rates, the dealers have small average long exposures and exposures vary inversely with the level of rates. Implications for aggregate bank dealer risk and market stability issues are discussed.