Panel Data Regression 3of9 - Fixed Effects LSDV

This 3rd of 9 videos in the series shows how to run fixed effects Least Squares Dummy Variable (LSDV) regression on both Excel and EViews. It provides a practical understanding of how to design and run the model outside of the built-in EViews application. I also interpret the results.

[@LorstenTegat8272]

every video of this playlist is very easy to understand, comprehensive and just efficient. thanks a lot! great examples as well

[@PatObi]

Thank you.

[@SM27_11]

Sir, you are marvelous. Lots of respect for u from India

[@ZhongdaSun]

Thank you so much Professor Pat, it is very clear and concise!

[@xeniama9839]

Very nice explanations!

[@rutniken4746]

Thank you so much, easy to understand

[@ayodejinajeemiziaq9166]

Can you please explain how to do this from Excel before importing to eviews

[@davidmontano6540]

Hi! In this example, LSDV model would suggest dummy Variable for Firm 7 (D7) is not statistically significant. As far as I know, to correctly interpret all other dummies we need to ensure that every single dummy coefficient must be indeed significant. Is this correct? Ignoring D7 significance yields in wrong conclusions? Thank you in advanced for your effort!

[@mynameisjoejeans]

Great video prof, thank you. Can LSDV be used on endogenous variables? and Can it be used with time lags?

[@daffananda7303]

Excuse me sir, is that necessary to do the classic assumptions test (heteroskedasticity and multicoliniarity) in LSDV?

[@rohtashbhall2671]

what about dummy variable D1 to D9. Where is D3 TO D8 along with B3 to B8. Thanks

[@khotibulumam7576]

Excuse me mr. Pat Obi can i have your data if you don't mind?

[@dumitrachecristina8561]

Hello and thank you for the great videos. One question if you don't mind: although I have 24 countries and made 23 dummy variables, I am having a problem when running the equation because I get the error "Near singular matrix". Would you be so kind and help me solve the problem, please?

[@PatObi]

Sorry for the very late response. You have TOO MANY DUMMY VARIABLES. Because dummies use 1 and 0 as values, it causes the estimator to see the dummy variables in your panel as almost perfectly collinear.

[@oskarikoskela4735]

@@PatObi Hi, what is a "reasonable" amount for dummies? I'm experiencing the same "near singular matrix" problem but I'm working with data relating OECD countries (38 countries) so what dummies should I eliminate? Or should I try completely different model?

[@matheusmeneghin8651]

@@PatObi i have 11 Regions, should i do only 2 dummies or how many dummies should i do?

[@VSP4591]

You could take all 24 dummy but you have to eliminate the intercept (C). If you have 23 dummy, you may keep the intercept. In this case C will include the coefficient of the 24th variable.

[@PatObi]

@@matheusmeneghin8651 Please use a different FE method, that is, instead of LSDV method

[@AbdulJabbar-vv6hr]

if we change the base group will the value of b1 and b2 be impacted other than b0??

[@Indahem]

can we use LSDV to test the effect of bond rating on bond price performance? (for example: if company's bond rating is AAA, D1=0, otherwise is 0)?

[@ManduPonkanshi]

hi sir I only have 20yrs of data and 6 dummy variable but I am getting near singular matrix error pls help me

[@卓卓-m4c]

Excuse me sir, may I ask if LSDV is FE-OLS, I often see FE-OLS in papers recently but not sure if it is the same as LSDV?

[@TinaTina-xn9on]

If I have hundreds of firms, I have to create hundreds of dummy variables?

[@agoogleuser6452]

Excuse me sir, My panel data set is non stationery at level and stationery at first difference. What is the next step I need to follow? Thanks in advance.

[@PatObi]

Test for panel cointegration and if cointegrated, run Panel VECM. Don't forget to watch the series, Panel VAR.

[@agoogleuser6452]

@@PatObi I watching your series. Very informative and clear ❤️

[@agoogleuser6452]

@@PatObi Excuse me Sir my panel data is no cointegration. Can I use OLS regression? Is it okay?

[@datax-analytviews8009]

If am having 28 companies, how many dummys should i estimate with the model?

[@PatObi]

27, but that's too many dummy variables. Use a different approach.

[@JuStFoRqUicK]

@@PatObi what kind of approach would you suggest? perhaps adding dummies only for time might be a solution?

[@matheusmeneghin8651]

Do i Also include the controlled variables or can i ignore them for fixed effect and put just dependent and independent

[@PatObi]

You can include them in the model

[@Looooly32]

how can we add a time fixed effects in addition to the firms fixed effects? thank you.

[@PatObi]

You can create dummies for time. On EViews, you can check the box as I show in video 5 of 9.