Estimation Procedures for Pooled Cross-Sectional and Time Series Data

Abstract

In this paper we provide a didactic introduction to econometric estimation procedures for data with both cross-sectional and longitudinal variation. Several different ap proaches are discussed, and data from an earlier study by Deutsch and Alt (1977) are reanalyzed as an example.

Corruption and Related Socioeconomic Factors: A Time Series Study

“…Panel data, multi‐dimensional data covering numerous countries over several time periods, is used. The validity of utilizing Granger causality in panel data is evidenced in numerous studies –Marvell and Moody (1996), Berk et al (1979), Coupet (2003), and Erdil and Yetkiner (2005), to name a few (although only Coupet (2003) is a corruption study). Additionally, ‘recent theoretical developments in Granger causality methods have made tests using relatively short time periods possible through the use of panel data’ (Hoffman et al 2005, p. 312).…”

Modernization, age structure, and regional context: A cross‐national study of crime

“…In fact, if dummy variables for both time and country and their cross-products were included, the problem of nonindependence of measurement would be eliminated. (See Berk, Hoffman, Maki, Rauma and Wong 1979;Bennett and Lynch 1990 for a further discussion of these issues.) However, this procedure would generate an inappropriately high number of parameters to be estimated given the available degrees of freedom.…”

Specification Problems, Police Levels, and Crime Rates*

“…The multiple time series is an exceptionally strong evaluation design (e.g., Berk et al, 1979;Campbell and Stanley, 1966). One obvious advantage is that it provides a very large sample size, 980 or more even with annual data, thus permitting us to enter numerous control variables and still retain a large number of degrees of freedom.…”