Hello,
I trying to investigate how a policy change may impact households saving choices by using panel data. My empirical strategy is a diff-in-diff and an OLS with an included interaction term, i.post#i.treated. My dependent variable, which is active saving in shares, is very zero-inflated. Almost 90 percent of the observations are equal to zero. However, the variable is continuous and can be both negative and positive, so empirical models related to count data are not optional in my case. I was wondering if the distribution of my dependent variable could cause any particular problems regarding the estimates and my empirical strategy?
Any help/answers will be highly appreciated!
I trying to investigate how a policy change may impact households saving choices by using panel data. My empirical strategy is a diff-in-diff and an OLS with an included interaction term, i.post#i.treated. My dependent variable, which is active saving in shares, is very zero-inflated. Almost 90 percent of the observations are equal to zero. However, the variable is continuous and can be both negative and positive, so empirical models related to count data are not optional in my case. I was wondering if the distribution of my dependent variable could cause any particular problems regarding the estimates and my empirical strategy?
Any help/answers will be highly appreciated!
