Industry fixed effect in an ols regression

[Najiba Ramdani]

Dear Statlist,

I am conducting an event study for a sample of 153 companies announcing layoffs between 2008 and 2016.

I am using a regression between cumulativ abnormal return (CAR) and different variables using the command.
regress CAR variables1 variable2 etc, r
(The different variables correspond to firm performance, size of the firm, dummy variable if the firm belong to the sector service or no ...)

However I was thinking about the relevance of using industry or year firm fixed effect using the following command
regress CAR variables1 variable2 i.INDUS i.year, r

I was thinking of industry fixed effect since 50% of the companies belong to the same industry.

However I am not sure I am having the right reasoning.

I have found a lot of information about panel data fixed effect but not much about fixed effect in OLS regression that's why I am asking.

Thanks a lot,

Najiba

[Joro Kolev]

There is nothing to stop you from including industry and year fixed effects as you showed in your post.

[Carlo Lazzaro]

Najiba:
if you have N>T panel data, your first try should be -xtreg-.
That said, you can explore industry and time fixed effects with pooled OLS, too (however, if a panel-wise effect exists, and to investigate whether it actually exists, I would consider -xtreg,fe-).

[Phil Bromiley]

I haven't read event studies recently, but traditionally you would not need a fixed effect. The abnormal return is essentially a change measure controlling for the performance of the market as a whole. While a firm fixed effect would make sense if you are working with stock price, with abnormal returns you are essentially looking at changes in stock price.

If you do go to with fixed effects model, you should consider fixed effects at the firm level.

[Phil Bromiley]

Also, do test for significance of the fixed effects. If they are not significant (and I wouldn't think they would be) then you don't need them and you save degrees of freedom by omitting them. You might consider average abnormal returns for the industry (Omitting the firm of interest) as a control.