Because the unobserved managerial effort and the unobserved x have area 2 it
specific components OLS would yield biased estimates and panel data techniques
should be used. Random effects estimation requires that the area effects are not
correlated with x Ž . Baltagi, 1995 . But the omitted x are correlated with x 1it 2 it 1it
by assumption and the omitted managerial effort variable is correlated with x1it
Ž . directly and indirectly via x from the manager’s optimisation problem. Thus 2 it
the endogeneity of managerial effort indicates that we should use fixed effects
procedures to estimate the cost function