# Time-fixed effects with lagged variables and monthly dummies with Stata.

I'm studying Finance and currently doing my master thesis where I'm trying to relate a mutual fund's performance with its characteristics. A lot of my thesis revolves around econometric and stata usage and, unfortunately, I'm having some doubts, so I am asking for your help.

I'm guiding myself through a paper and right now I have to do a time fixed effect regression with time dummies (monthly) and lagged explanatory variables. This explanatory (independent) variables are the fund characteristics (just as age, dimension, fees, etc) that I'm trying to relate to performance (the dependent variable).

The paper reads: "All the regressions include time-fixed effects (quarter dummies) to account for cross-sectional dependence, and t-statistics are clustered at the fund level to account for autocorrelation in fund performance. All explanatory variables are lagged one-quarter." and this is what I'm trying to mimic, just using monthly dummies instead.

What commands do you think I should use? I've already tried:

"xi: xtreg depend L.independ i.date, fe robust"; this way I get somewhat strange results.

I've also tried "xtreg depend L.independ, fe i(date) robust" but I get the error "time variable not set" even after doing tsset. I do not get that error if I remove the L. (for the lagged effect) and sure I could jut generate a new lagged variable and use that instead but since I get different results in both ways, I'm unsure about which should I use.

I hope I have been clear about my problem but please ask me if there are still any doubts.

I'm guiding myself through a paper and right now I have to do a time fixed effect regression with time dummies (monthly) and lagged explanatory variables. This explanatory (independent) variables are the fund characteristics (just as age, dimension, fees, etc) that I'm trying to relate to performance (the dependent variable).

The paper reads: "All the regressions include time-fixed effects (quarter dummies) to account for cross-sectional dependence, and t-statistics are clustered at the fund level to account for autocorrelation in fund performance. All explanatory variables are lagged one-quarter." and this is what I'm trying to mimic, just using monthly dummies instead.

What commands do you think I should use? I've already tried:

"xi: xtreg depend L.independ i.date, fe robust"; this way I get somewhat strange results.

I've also tried "xtreg depend L.independ, fe i(date) robust" but I get the error "time variable not set" even after doing tsset. I do not get that error if I remove the L. (for the lagged effect) and sure I could jut generate a new lagged variable and use that instead but since I get different results in both ways, I'm unsure about which should I use.

I hope I have been clear about my problem but please ask me if there are still any doubts.

## All Answers (3)

Vera Rocha· Copenhagen Business SchoolFirst of all, from your description, I am not sure if you should estimate a fixed effects panel data model. You only mention fixed effects for time, and this could be done by inserting monthly dummies. Check which model the paper uses - random effects? fixed effects? population-average? Try to estimate using "xtreg", without "fe" option.

Also, you mention " t-statistics are clustered at the fund level to account for autocorrelation in fund performance." - so try to estimate your model with cluster-robust standard erros, clustered at the fund-level (add the option "vce(cluster XXX)" where XXX should be the variable that identifies each of the mutual fund in your data).

I hope this could help you.

Vera

Barnabas Kiiza· Makerere UniversityBarnabas Kiiza

James Thomas Bang· Saint Ambrose UniversityCan you help by adding an answer?