Random Coefficient Models for Time-Series–Cross-Section Data:

This article considers random coefficient models (RCMs) for time-series–cross-section data. These models allow for unit to unit variation in the model parameters. The heart of the article compares the finite sample properties of the fully pooled estimator, the unit by unit (unpooled) estimator, and the (maximum likelihood) RCM estimator. The maximum likeli-hood estimator RCM performs well, even where the data were generated so that the RCM would be problematic. In an appendix, we show that the most common feasible generalized least squares estimator of the RCM models is always inferior to the maximum likelihood estimator, and in smaller samples dramatically so. 1 Random Coefficient Models and Time-Series–Cross-Section Data The use of time-series–cross-section (TSCS) data has become very common in the study of comparative political economy.1 TSCS data consist of time-series observations at regular intervals on a fixed set of units. The paradigmatic comparative political economy studies (e.g., Garrett 1998) have about 30 annual observations on about 20 advanced industrial democracies; the dependent variable in these studies is typically a measure of economic policy or outcome, with the independent variables being the economic and political determinants of those outcomes or policies. Whereas both the number of units and the length of the time series vary from study to study, the numbers in the Garrett stud