he topic of cointegration and related topics of nonstationarity and unit root T econometrics .have been the center of considerable attention in the applied econometric literature over the last several years. A partial listing of articles includes: Engle and Granger (1987), Engle and Yo0 (1987), Granger (1986), Hendry (1986), and Campbell and Shiller (1988). This article explores the application of cointegration techniques to the study of daily futures and cash prices on live cattle.
This article is presented in four sections. First, the basics of cointegrationeconometrics are reviewed. A discussion follows on how cointegration reflects upon economic interrelationships in general and, more specifically, how it reflects on issues of interest to futures market researchers (e.g., informational efficiency, causality, forecasting, and basis relationships). Then, cointegration methods are applied to daily data for slaughter cattle cash and futures prices and the implications of the results with respect to the above issues are discussed.
SOME BASICS ON COINTEGRATION
A series of data indexed by time (a set of data in which order of observation is important) is said to be integrated of order d if it requires d first differences to reduce the resulting series to stationarity (e.g., d = 2 if X ( t ) -X(t -1) -X ( t -1) + X(r -2) = Z(t) is stationary). Here, stationarity means that the characteristics of the times series are describable in terms of the time separating observations and not the particular time of the observations. Researchers find that many economic time series appear to require first differencing (d = 1) to achieve stationarity (Gould and Nelson (1974), Granger (1986)).
The standard approach to univariate time series analysis of data integrated of order d , is to model the dth differenced data as either an autoregression, a moving Word processing and editorial assistance were provided by Liisa Menzel. Thanks to Robert Shiller and two anonymous reviewers for comments on an earlier draft. The views expressed are solely those of the authors. This research was supported by the Texas Higher Education Commission, ARP project number 7321.