Cointegration: An Introduction to the Literature

Abstract

An overview of the cointegration approach to econometric specification and estimation is provided. A non‐technical approach is adopted, and is intended to serve as an entry into this important new literature for the reader with no background knowledge of the subject but with some limited knowledge of econometrics. Particular emphases are given to the rationale for using cointegration techniques in the estimation of economic relationships, to providing intuitive explanations of the concepts and techniques, and to demonstrating their applications in practice. Reference is made throughout to other articles which explain particular methods or recent developments more formally and fully than is possible here. Finally, a simple application of cointegration techniques to the estimation of the consumption function is provided.

The relationship between stock markets of major developed countries and Asian emerging markets

“…It incorporates the presence of non-stationarity, long-term relationships and short-run dynamics in the modeling process. A detailed description of cointegration can be found in Dolado, et al (1990), Perman (1991), Hamilton (1994), Manzur, et al (1999) and Penm, et al (2003).…”

Cointegration: a novel approach for the removal of environmental trends in structural health monitoring data

“…A suitable number of lags p should be included to insure that 3 i becomes a white noise process (Perman 1993). To convert from the more traditional AR(p) model to the model form (equation (3.3)), the following substitutions should be made: let a 1 = 1 + r + b 1 , a n = −b n−1 + b n , for n = 2 .…”

Approaches to nonlinear cointegration with a view towards applications in SHM

“…A suitable number of lags p should be included in this model to insure that becomes a white noise process [4]. In this form, the characteristic roots of the process in question, and therefore the stationarity of the process, is determined by the value of .…”

“…For SHM then at least, the interest of cointegration lies in how to combine nonstationary variables to create a stationary result purged of the common trends in the original set. As one may imagine, a large body of literature is already available from econometrics on this very topic (see [4] for a good overview). One of the most common approaches used to find a stationary linear combination of multivariate nonstationary data is the Johansen procedure [5], which is a maximum likelihood based method for combining nonstationary variables whose first difference is stationary.…”