Panel Data Analysis: Static Models

Abstract

Panel data are constructed through survey conducted at several points in time using the same cross section units. A panel consists of a set of multiple entities from which information on similar issues is collected over time. Panel data can take care of inter-individual differences and intra-individual dynamics by mixing cross section and time series components. Panel data econometric models examine unobserved heterogeneity by estimating cross section-specific effects, time effects or both. These effects may be non-stochastic or stochastic. In a fixed effects model, unobserved heterogeneity varies across cross section dimension or time period non-stochastically, whereas a random effects model considers stochastic variation of the unobserved character in the data across individual or time period in terms of error variance components. A one-way error component model captures only one type of unobserved heterogeneity by including one set of dummy variables, while a two-way model takes care of both cross section-specific and time-specific heterogeneity by taking two sets of dummy variables. This chapter discusses different types of panel data model in a static framework.