Of Forests and Trees

Abstract

My aim in this note is not to rejoin the useful comments of Beck and Williams (in this volume) in this symposium, but rather to speculate about the future place of cointegration and error correction in the discipline of political science. Stopping somewhere short of the pejorative “hegemony” Williams anticipates for the ECM framework, I agree that this modeling approach is likely to enjoy wide play in a variety of substantive arenas. As such, I believe it is critical that analysts concerned with advancing cointegration methodology do so with an eye toward not just technical issues, but also questions regarding the theory of cointegration, error correction, and time-series stationarity. Furthermore, if methodological advancements are to be of any real value, they must be accessible to political scientists who, for inexplicable reasons, care not to make econometrics the center of their beings. In short, I am concerned that we will produce a method (1) dominated by questions of measurement and measurement strategy (at the expense of theory) and (2) inaccessible to (and therefore ignored by) most analysts. Such an evolution would be tantamount to losing the error correction forest for its trees.