Hello all,
I am currently trying to estimate a DiD model for the first time and I need some advice, as I am no expert in statistics.
I want to analyze the differences in returns within one sector within different countries. As a time dummy I want to use the date of new regulatory rules that were introduced in country X but not country Y (hence I would use binary variables here). As a control group I am plannig to use the country where the regulation stayed the same (hence indicated by 0 and 1).
I have several questions regarding the procedure:
1. How can I statistically test if the trends (for both countries) were relatively similar before the introduction of the regulation in country X? Is this also necessary?
2. What range (measured in days) should I use for the estimation? Will there be issues if I choose an interval that is too wide (e.g. +/- 365 days)?
3. Is it necessary that I have the same number of stocks for both countries?
4. Would you also recommend using fixed effects? In this case, should I use date and stock fixed effects?
I am aware how to create the two dummy variables and the interaction variable (time x treated), however I am not 100 percent sure, how i should use the dependent variable. Is it sufficient if I use xtset stock_id date and then use the return variable as the dependent variable?
Thank you for your help! And please excuse my limited knowledge in the statistical methods.
I am currently trying to estimate a DiD model for the first time and I need some advice, as I am no expert in statistics.
I want to analyze the differences in returns within one sector within different countries. As a time dummy I want to use the date of new regulatory rules that were introduced in country X but not country Y (hence I would use binary variables here). As a control group I am plannig to use the country where the regulation stayed the same (hence indicated by 0 and 1).
I have several questions regarding the procedure:
1. How can I statistically test if the trends (for both countries) were relatively similar before the introduction of the regulation in country X? Is this also necessary?
2. What range (measured in days) should I use for the estimation? Will there be issues if I choose an interval that is too wide (e.g. +/- 365 days)?
3. Is it necessary that I have the same number of stocks for both countries?
4. Would you also recommend using fixed effects? In this case, should I use date and stock fixed effects?
I am aware how to create the two dummy variables and the interaction variable (time x treated), however I am not 100 percent sure, how i should use the dependent variable. Is it sufficient if I use xtset stock_id date and then use the return variable as the dependent variable?
Thank you for your help! And please excuse my limited knowledge in the statistical methods.
