On the Estimation of Panel-Data Models With Serial Correlation When Instruments Are Not Strictly Exogenous

In recent years, researchers in many disciplines, including economics, accounting, finance, and marketing, have increasingly relied on panel data to model the behavior of individuals and firms. They have done so because panel data allow them to control for temporally persistent unobserved differences among individuals or firms that in many instances may bias estimates obtained from cross-sections.