This preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
Unformatted text preview: Imbens, Lecture Notes 21, ARE213 Spring ’06 1 ARE213 Econometrics Spring 2006 UC Berkeley Department of Agricultural and Resource Economics Panel Data II: Fixed Effects In this lecture we consider the same setup, with a linear model: Y it = X it β + c i + ε it , with c i an unobserved individualspecific, timeinvariant component. However, compared to the random effects discussion we relax the assumption that c i is independent of the observed covariates X it . We continue to maintain the exogeneity assumption on the residuals: E [ ε it  X i 1 , . . . , X iT , c i ] = 0 , and in fact for inference we make the stronger assumption E [ ε i ε i  c i , X i ] = σ 2 · I T . We consider a couple of estimators. The first is based on simply adding a Ndimensional vector of timeinvariant covariates Z , with its j th element for unit i in period equal to Z it,j : Z it,j = 1 { i = j } . Then if we define c to be the vector with typical element c i , we can write Y it = X it β + Z i c + ε it . The first estimator is just the least squares estimator for this regression function: min c,β i,t ( Y it X it β Z i c ) 2 . Imbens, Lecture Notes 21, ARE213 Spring ’06 2 The estimators for both β and c are unbiased. However, the estimators for c are not con sistent. As we get more and more observations, we do not get more information about c i ....
View
Full
Document
This note was uploaded on 08/01/2008 for the course ARE 213 taught by Professor Imbens during the Spring '06 term at Berkeley.
 Spring '06
 IMBENS

Click to edit the document details