Identification, Estimation, and Inference in Cointegrated Systems

Abstract

Chapters 3 and 4 provide technical background for the empirical applications conducted in Chapters 6–8. In this Chapter we examine identification, estimation and testing hypotheses in models characterized by long-run relations that may prevail in nonstationary time series data. The discussion does not break new ground, but brings together important points that span in a wide range of literature. At the same time, there is no attempt to incorporate all contributions that have been made in this area. The discussion is “complete” only in the sense that it provides a unified approach to dealing with the econometric issues inherent in the consideration of nonstationary macroeconomic aggregates. We tackle issues in the sequence they are typically encountered in applied work. The discussion spans issues of structural identification, estimation and inference, and structural stability while implications for dynamic analysis are discussed in Chapter 4. The discussion is aimed at those readers wishing to acquaint themselves with an econometric approach for empirical analysis of models characterized by integrated and cointegrated variables. In later chapters the techniques are illustrated with applications that pertain to a long-run money demand relation, but the methodology may be applied in a number of areas. Chapter 3 begins by providing background on nonstationarity and the implications that unit roots have for estimation of a stable long-run money demand function. In section 3.2 procedures designed to detect unit roots are discussed.