Two proxies for changes in interest rates are used: the change in the three-month Treasury Bill (TB) yield (DYT3) and the difference between the return on long-term U.S.
where: [r.sub.i] is the excess return on a stock portfolio i; EMKTSP is the excess return on the S&P 500 stock index (the market factor); DYT3 is the change in the three-month TB yield; TPREM is the return on U.S.
Based on the evidence from previous studies showing the negative relationship between bank stock returns and changes in interest rates [e.g., Flannery and James, 1984; Aharony et al., 1986; Kane and Unal, 1988], the coefficients of DYT3 and TPREM are expected to be negative and positive, respectively, in the regressions.
Beginning-of-month CD and TB yields used to compute the monthly values of DYT3 and DCDTB are obtained from Salomon Brothers' Analytical Record of Yields and Yield Spreads .
The coefficients of DYT3, TPREM, DPREM, and DCDTB for the BHC portfolio all have the expected signs in the 1969-89 period and they show that BHCs in general exhibited statistically significant extra-market sensitivities to the interest rate and default risk factors.(5) For the subperiods, the DYT3 and TPREM coefficients indicate that the extra-market interest rate sensitivity of the BHC portfolio was not significant in the pre-October 1979 period, while it was during the post-October 1979 period.
Moreover, they show that the significant sensitivities of the BHC portfolio to the two default risk variables signify an additional dimension of the market's perception of the higher risk of the banking industry that is not evident from the coefficients of EMKTSP, DYT3, or TPREM (i.e., the Stone model market and interest rate risk proxies).
The significant correlations between the first component and DYT3, TPREM, DPREM, and DCDTB are reflective of the significance of the coefficients of these interest rate variables for the BHC portfolio in the multifactor model analysis (see Table 1).
The second PC is, however, significantly correlated with DYT3, which suggests that it may be related to the influence of short-term interest rate changes.
The third PC in the 1969-79 subperiod is significantly correlated with DYT3, DPREM, and DCDTB but the correlations are opposite in sign to those observed for the 1969-89 period.