Nonlinear Error Correction: The Case of Money Demand in the UK (1878-1970)

Alvaro Escribano
Department of Statistics and Econometrics, Universidad Carlos III de Madrid

Abstract

In this paper we define a broad class of nonlinear error correction models. We discuss the advantages and disadvantages of the different notions of the concept of trend and its relationships with long-memory, long-range dependence, persistence, etc. By using this concepts we formally characterize the time series properties of variables generated by these nonlinear models. We suggest a semi-parametric approach to model the unknown nonlinear adjustment function and discuss alternative parametric approximations. Estimation and testing in this class of models are reviewed and the problems that might arise with linear estimation procedures are highlighted. As an application we discuss in detail the estimation of the money demand in the UK since 1978-1970.