regdhfe display empty samples

[Frank Wilkins]

Hi All,

I am doing a diff-in-diff regression in a panel dataset.
As I want to include both firm fixed effects and year*industry fixed effects I used the regdhfe code. More specifically:

reghdfe dmarketshare lagcash1 CUT21 cash1xCUT21 a(gvkey sicfyear)

dmarketshare: the year change in marketshare per firm
lagcash1: the variable of interest: lagged cash holdings of the firm
CUT2: A treatment dummy
lagcash1xCUT2: interaction variable
gvkey: fixed effects for firm
a,(gvkey sicfyear): include fyear*industry fixed effects



But it always displays:

Empty sample, check for missing values or an always-false if statement
r(2000);

[Joseph Coveney]

Have you verified that there are at least some observations that do not have any missing value on any of the variables (help misstable)?

This error message also occurs if a variable is string instead of numeric. You might want to check for that, too.

[Carrie Zhou]

Have you verified that there are at least some observations that do not have any missing value on any of the variables (help misstable)?

This error message also occurs if a variable is string instead of numeric. You might want to check for that, too.

Thank you Joseph! It is really helpful. It turns out that I was using a string variable for clustering.