Endogenous independent variables in multilevel models (event studies)

[Elena Komarova]

Good afternoon! I run an event study where I estimate the abnormal stock return to a firm's corporate event with the 2-level random intercept model (xtmixed command). Some of my explanatory variables (at a firm level) can be potentially endogenous. Could you recommend me which approaches I could use to correct for endogeneity in Stata? On other other hand, would the event study methodology on its own useful to rule out potential endogeneity of explanatory variables?

Thank you in advance!

[Phil Bromiley]

You'll increase your chances of a useful answer by following the FAQ on asking questions - provide Stata code in code delimiters, readable Stata output, and sample data using dataex. Also, remember that most of us are not from your area.

I don't use the mixed procedure. There are many different ways to attack this - eregress, mixed, treatment models, etc. However, some don't implement random effects (although you can easily implement fixed effects). GSEM can certainly do it. User written cmp will also probably work. You don't say what the two levels are. If it is firm and year, folks often organize panels by firm and then include i.year among the regressors.

[David Benson]

To answer the last part of your question, "does the event study methodology on its own useful to rule out potential endogeneity of explanatory variables?" The answer is probably no. Many of the corporate events researchers care about (firing a CEO, making an acquisition, forming an alliance, repurchasing their stock, etc) are likely endogenous, and so you will need to find some way to try to correct for it.

Michael Roberts & Toni M. Whited have a great paper ("Endogeneity in Empirical Corporate Finance") discussing a number of econometric techniques aimed at addressing such endogeneity problems (i.e. instrumental variables, difference-in-differences estimators, regression discontinuity design, matching methods, panel data methods, etc) see https://papers.ssrn.com/sol3/papers....act_id=1748604

[Elena Komarova]

Thank you so much for your useful answers!