Wednesday, March 1, 2023 at 11:15am to 12:45pm
Uris Hall, 498
Robert Sherman, CalTech
IN A CORRELATED RANDOM COEFFICIENTS LINEAR PANEL DATA MODEL:
A FUNCTIONAL FIXED POINT APPROACH TO IDENTIFICATION AND ESTIMATION
We develop a linear regression panel data model allowing random coefficients to be correlated with regressors not only within periods but also across periods. The random coefficients are modeled as sums of independent possibly nonidentically distributed past and current shocks. This structure allows feedback, in the sense that future regressors can be correlated with all past shocks, and also allows all lagged dependent variables to be regressors.
For each time period, we identify all marginal first moments of the random coefficients. These moments have causal interpretations. The identification results rest on a novel functional fixed point argument and lead to natural estimators of these moments. We provide simulation evidence of the usefulness of this approach.