Writing

[Mesfin Yasin]

Hi everyone,

I’m using the following regression in Stata:
reghdfe depvar indvar contrvar, vce(cluster Firmcode) absorb(Firmcode Year) In describing my model, I wrote that it "uses a high-dimensional fixed effects panel regression clustered at the firm level, accounting (absorbing) for both firm-level and year-level effects." However, my supervisor asked me to clarify what "accounting" or "absorbing" actually means in this context.

Could someone help me refine this explanation? Specifically, how can I better describe what "absorbing" firm and year effects mean in the context of high-dimensional fixed effects regression? Is this the correct way of writing, clustered at the firm level, including firm and year-level fixed effects?

Any advice or suggestions would be greatly appreciated!

Thanks in advance!

[George Ford]

"The regression model is a two-way fixed effects regression with fixed effects for firm and year, which control for unobserved time-invariant firm characteristics and temporal shocks. Standard errors are clustered at the firm level."

Note you are using the within estimator. Look it up.

Basically, the means are subtracted from the variables, so all the firms and time periods are on the "same" level. Say, for instance, a year was really bad for everyone. The year fixed effect remove that common effect. Or, if firms differ a lot in size, the sizes are normalized. This helps isolate the relationship you're interested in from unobserved firm-specific and time-specific confounders