ABSTRACT
Panel data sets provide a rich environment for researchers to investigate issues which could not be studied in either cross-sectional or time series settings alone. A key attribute of this type of data is that it provides a methodology to correct for the omitted variable problem inherent in cross-sectional data analysis. In a typical panel data study there are a large number of cross-sectional units, for example, a large number of individuals, firms or even regions/countries, and only a few periods of time. Researchers collect data on the characteristics of the cross-sectional units over a relatively short period of time, to investigate changes in behaviour or potential of a typical cross-section unit. A key feature of the panel data set is that it provides observation over time on the same cross-sectional units. This is in contrast to a pooled cross-section/time series data set, discussed in Chapter 9, where, typically, a randomly selected cross-section data set is observed a number of times over time, in order to increase the number of observations and to improve the degrees of freedom of the regression. This chapter focuses discussion on panel data of the type described above, providing an introduction to their applications and estimation in practice. Key topics
The nature of panel data and its applications
Using panel data to correct for the omitted variable problem
The fixed-effects regression models
The random-effects regression models
Panel data regression in practice
