A Primer on Cointegration with an Application to Money and Income
For some time now, macroeconomists have been aware that many macroeconomic time-series are not stationary in their levels and that many time-series are most adequately represented by first differences.1 In the parlance of time-series analysis, such variables are said to be integrated of order one and are denoted I(1). The level of such variables can become arbitrarily large or small so there is no tendency for them to revert to their mean level. Indeed, neither the mean nor the variance is a meaningful concept for such variables.