Cross Sectional analysis or creating subsamples?

#1
Hi there,

I would like to investigate whether the announcement of a rating change has an impact on the return on shares of different companies. Therefore, I use the methodology of the event study. Afterwards I would like to perform a multivariate regression to investigate the determinants of possible abnormal returns. For example, I would like to use regression analysis to investigate whether the stock prices of larger companies might react stronger/weaker to a rating change than smaller companies.
I can test some of the hypotheses both through the event study as well as through cross sectional regression. In order to test whether, for example, the shares of financial companies react more strongly, with event study methodology I can simply subdivide the sample into subsamples (financial companies and non-financial companies). However, I could also investigate this using a dummy variable in multivariate regression (1 for financial company and 0 otherwise). My question now is whether it is simply a matter of taste how to proceed here or whether there might be advantages and disadvantages for a method.

Thanks in advance