Panel Data (7): Fixed effects model in STATA

[@ektasrivastava1678]

very helpful videos sir

[@RESEARCHHUB]

Thanks!

[@stephenzamore3969]

Thank you.

[@redabennouna2007]

Svp, j'ai effectué le test Hsio j'ai obtenu que mon panel à un effet individuel, ensuite j'ai effectuer les 2 modèles et le Test Hausman ma présenter que mon panel suis un modèle aléatoire. Est ce que après je peux faire le modèle MCG et Pvar. Ou il me faut avoir un effet fixe ? Et merci 🙏

[@nanakwamegyasi5905]

Please do a video about Dynamic Panel model and Instrumental variables in Stata

[@jenli0426]

Hi, thank you for the video! May I know how do I create year-month * country fixed effect? My year-month variable is in '2010Jan' form, and i try to create by adding i.country*yearmonth, and no observation is produced. Thank you for your time！

[@stephenzamore3969]

Hi, Thank you for watching our videos. First, is your time frequency monthly or yearly? If it is monthly, then you need to give value labels of your time variable (e.g., 1 = 2010Jan, 2 = 2010Feb, etc, Stata can do it automatically - check out the command "encode"). Stata does not recognise "2010Jan" as numeric (This is string and it should be red in colour). Second, your syntax "i.country*yearmonth" seems problematic. Are you creating an interaction variable or is part of the regression? If it is part of the regression, the syntax should be "i.country#c.yearmonth". I will be able to provide specific guides if you answer my questions.

[@yousefghafo]

Hello, I’m having trouble writing the equation for fixed effects with clustered standard errors . Could you possibly help me out ?

[@esayasmequanint9539]

Thank you!

[@georgioskanistras4453]

Hello there. I have unbalanced data meaning that fo example for firm i , at time 2010 I have the value for revenue, costs but not for assets. Then for company 2 I have for 2012, 2011 all the data but for 2010 i miss one value of the observation for 2010...how to deal with this issue in Stata ?

[@stephenzamore3969]

This is basically unbalanced panel and you can use it as it is. If your T is not large, I suggest you run OLS using the averages of the variables.

[@andreab2114]

Very interesting! What if the dependent variable is in z scores within each entity? My dependent variable is an index that was computed for each country for 20 years, but it is expressed in standardized scores, so each group mean will be 0.

[@99evan76]

5:13 stata start

[@MaroofKhan-cr9ky]

Its very very informative but the screen is not visible so we are not able to see do file etc

[@stephenzamore3969]

Thanks for the feedback. It might help if you get a bigger monitor. But we will take this feedback into consideration when making new videos.

[@inesba2897]

Thank you for the video, have you please an idea how to perform difference in difference method (DID) for panel data (in stata)?

[@stephenzamore3969]

I have not looked into DID yet. You can check at the statalist here www.statalist.org/forums/forum/general-stata-discussion/general/1382228-difference-in-difference-in-difference-estimation-in-stata

[@commenting..2567]

at 1:39 you say that ui is correlated with the error term, but isn't ui the error term? And do you mean it is correlated with the predictor variable? Thank you

[@RESEARCHHUB]

Hi, u_i is unobserved individual/firm/industry/country (depending on your unit of analysis) specific effect, v_it is the error term after controlling for the unobserved specific effect.

[@commenting..2567]

@@RESEARCHHUB Ah okay, thank you so much for your reply

[@commenting..2567]

@@RESEARCHHUB In other words can I say that the main assumption for FE is that the error term is correlated with the unobserved variable u_i?

[@RESEARCHHUB]

@@commenting..2567 yes, exactly. We allow it in FE model but in RE models the assumption is that they are not correlated.

[@onwejoshuachukwuma4395]

How can I get the do file

[@renatusmbamilo9709]

Hi Stephen Zamore, can you share with me your ppts? I find them very resourceful.