The Role of Time Series Analysis in Econometric Model Evaluation

Abstract

This chapter discusses the role of modern methods of time series analysis in the evaluation of econometric models. Despite the fact that econometric models are frequently based on time series data, classical regression and related methods are almost always used in parameter estimation and hypothesis testing. The chapter presents an approach to econometric model evaluation that draws heavily on time series methods generally and spectral methods in particular. It presents the general approaches of classical econometrics and time series analysis. This comparison provides the motivation for the view that time series methods can play an important role in econometric model evaluation. The chapter discusses several basic concepts of univariate time series analysis including the power spectrum and explains how these can be used in model evaluation. It presents multivariate time series analysis and an analysis of aggregate consumption data that illustrates the use of time series techniques to evaluate a simple model.