Stochastic Technical Progress, Nearly Smooth Trends and Distinct Business Cycles

This paper investigates whether it is possible to entertain simultaneously two attractive views about US GDP. The first is that long term growth in US GDP is attributable to an empirically plausible specification of random technical progress. The second is that deviations of GDP from a fitted smooth 'trend' are mostly attributable to shocks that have only temporary effects, so that they are unrelated to the shocks to technical progress that lead to long term growth. The paper shows that these two views are not incompatible by constructing a model where stochastic technical progress (whose properties are calibrated to fit some features of US data) has essentially no effect on suitably detrended time series of GDP. The paper also studies variations in wedges between price and marginal cost that are capable of giving rise to these transitory movements.