This paper examines the returns of European bank stocks between years 2000 and 2018 using different models, data and tools. We start with a comparison of the standard market models used for understanding the returns of financial instruments. We expand the models with additional factors, for which we can assume a significant impact on the bank returns, mostly in relation with interest rates. By comparing statistical criteria we determine the factors and models with the best explanatory power for a given set of banks. We also compare the efficiency of models on sets of big and smaller banks, European and U.S. markets, and individual European countries. We note a significant difference between the different subsets of banks, especially between more and less developed markets, as in general, all models have a higher explanatory power when applied on banks operating in more developed stock exchange markets, and factors reflecting the general state of the market tend to be more significant. On the other hand, interest rate factors tend to be relatively more significant in less developed markets. Similarly, market factors tend to be more important for big banks, while for smaller banks, the bank specific factors are more significant, and the models have a lower explanatory power.
We continue with an application of the Bayes approach to model selection on the aforementioned set of factors in order to determine the best model for a given set of banks. We use the same approach on the sets of banking supervision data in Europe, made available on an annual basis by the European Banking Authority. Using this method we determine a set of data points and geographical exposures that might have an impact on the returns of European banks.
|
