Stationary data and the effect of the minimum wage on teenage employment

Abstract

One of the interesting features of time series work on the effect of the minimum wage is that the problem of possible spurious regressions when data may be non-stationary has been largely ignored. It is shown that this is potentially a serious problem. It is found that although the teenage employment to population ratio is stationary, the 'independent' variables in time series regression equations are non-stationary. Substitution of the stationary first differences of the independent variables in the regression equations for their levels, is found to greatly undermine the estimated influence of the minimum wage on the employment-to-population ratio. Estimates of the effect of the minimum wage are found to be only marginally more significant when seasonal differencing of the dependent variable and an ARCH estimation technique are employed.