I have another question and hopefully you can help me (as already done several times)
I am conducting an analysis on the determinants of NPLs in European Banking SYstem.
In my sample I have observations for different banks in the same State.
After done a dynamic panel analysis with the xtdpdgmm command on the whole sample, I want to see if there is some changes when I consider a subsample, for example for big banks; banks belonging to distressed countries, banks with a high value of npl.
In order to do so, for obtaining high npl banks I have done the following steps:
1) Calculate, by each bank, its median NPL value.
2) Calculate, by each state, its median NPL value. This is done using data for all banks in a given country.
3) Create a dummy which takes the value 1 if the median NPL by bank is greater than the median NPL by country.
Code:
egen bank_npl_median=median(npleur) , by(bank_id) egen country_npl_median=median(npleur), by(ISO) g high_npl=1 if bank_npl_median>country_npl_median replace high_npl=0 if high_npl==.
Then, when I analyze the determinants for such subgroups, I have basically just ad "if high_npl==1" after the regression. Is this correct ? Interacting this dummy with all the determinants will lead to the same results, am I right??
thank you in advance
0 Response to Subsample in Panel data By Median
Post a Comment