Are Technology Shocks Nonlinear?

This paper examines the behavior of postwar real U.S. GNP, the inputs to an aggregate production function, and the associated Solow residuals for the presence of nonlinearities in their generating mechanisms. To test for nonlinearity, we implement three di®erent statistical tests: the McLeod-Li test based on the correlogram of the squared data, the BDS test based on the correlation integral, and the Hinich bicovariance test based on the third-order moments of the data. We ̄nd substantial evidence that the generating mechanism of real GNP growth is nonlinear, but no evidence for nonlinearity in the Solow residual generated under alternative assumptions. We further ̄nd that the generating mechanism of the labor input series (expressed as hours worked) is nonlinear whereas that of the capital services input (expressed several ways) appears to be linear. We conclude that the source of the nonlinearity in real output is in the labor markets rather than in exogenous technology shocks. Finally, we examine the behavior of simulated factor input series from an asymmetric adjustment model to determine whether asymmetric adjustment costs are the source of the observed nonlinearities in the labor input. JEL primary ̄eld: C22