I am currently researching the effect of mobile banking availability on bank profitability for 2015 - 2019. My data is a data panel with n = 30 and T = 5 (N = 150 data). In my model there are 8 independent variables and one of them is the mobile banking variable. I use a dummy variable for the mobile banking variable. 1 if the bank has mobile banking and 0 if the bank does not have mobile banking, in that year. For example, if a bank launched mobile banking in 2017, then in 2015 - 2016 it would be 0, and in 2017 - 2019 it would be 1. Is it possible to use such a dummy variable? what estimation method is most suitable?
yes, it is possible to use dummy variable. Least squares dummy variable (LSDV) method can be used for estimation.
Maheswar Sethi is correct. If your dependent variable (profit) is in numbers, you also might want to log it. Furthermore, you should get cluster-robust standard errors to get correct t-values for all your regression coefficients.
It depends on several factors
1. Type of dependent variable you are using i.e. limited discrete dependent variable, continues variable etc.
2. Is your model a dynamic model or not
3. Check for endogeniety, heteroscedasticity etc.
So depending upon above factors you will decide which estimation technique would b more appropriate for your model i.e. fixed / random effect model, FMOLS, GMM, logit probit model etc.
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics
- Regression