Firm, industry and event fixed effects in a regression model

[Thiago Fernandez]

I am doing an event study on the impact of ESG on stock returns. I have performed the event study itself and got the values for abnormal returns as well as cumulative abnormal return.

In addition to this, I need to regress the cumulative abnormal returns on a trend, some event-related variables, firm related variables, and firm and industry fixed effects. Thus, the model looks as follows:

CAR = alpha + beta x Trend + gamma*X_i + lambda*Y_i,t + delta_j + theta_s

where, X_i is a vector of variables related to the event, Y_i,t is a vector of variables related to the targeted company at time t. Delta is firm fixed effects, and theta is industry fixed effects.

I know that for the industry fixed effects I can use dummy variables, but how do I account for firm fixed effects? What does firm fixed effects include?

[Joro Kolev]

If you include firm fixed effects, your regression will drop all firms for which there is only one event.

[Phil Bromiley]

Welcome to Stata list. Following the FAQ on asking questions will increase your chances of useful answer – provide Stata code in code delimiters, readable Stata output, and sample data using dataex.

If you include company fixed effects, it essentially provides a dummy variable for each company. The estimate is than within the company over time. This controls for and means you can't estimate parameters on any stable company characteristics. Since publicly traded companies almost never change reported industry, including company fixed effects will control for industry.

Please read the panel data analysis part of the PDF documentation and xtreg. This will help you a lot in understanding what you're doing.