A form of this property is, unusually, the null hypothesis in Kwiatkowski et al.’s KPSS test. Engle and Granger showed how to assess if there exists a linear combination of variables with this property, a scenario called cointegration. This property does not hold in the presence of a unit root of the characteristic equation, which can be assessed using a Dickey-Fuller test. Whether this property holds for the joint (*) CDF or just the first two moments distinguishes its “strict-sense” and “wide-sense” forms, the latter of which corresponds to the autocovariance function only depending on the lag between its two inputs. Differencing, adding a seasonality term, and de-trending are all common ways of getting a time series to have this property. (-5[1])For 10 points, (10[1])name this property of a statistical process whose distribution does not change with time. ■END■ (10[1]0[5])