Valuation of financial versus non-financial firms: a global perspective

It is common practice in many empirical studies in finance to exclude financial services firms from the samples used in different stages of the analysis. In this paper, we analyze the impact of this exclusion on common asset pricing tests by comparing the results using data both including and excluding financial sector firms in the countries with the largest financial sectors (the G7 countries plus the Netherlands and Switzerland) over the 1973-2000 period. We find that excluding financial service firms from empirical asset pricing tests can impact the corresponding inferences. It may influence both the identification of the number of risk factors found to be significant and the corresponding betas. For example, the estimated coefficients on some factors are significantly negative for the financials and significantly positive for many of the non-financials. We also show that, using a Fama-MacBeth (1973) test procedure, we fail to reject some models when financial firms are included in tests, but reject the same models when financials are excluded. Consequently we suggest researchers recognize the possible impact of the industry composition of the portfolios used in empirical studies.