A STRIKING change in empirical macroeconomics in the 1980s has been the development of an alternative way to think about aggregate trends and cycles. Traditionally, aggregate series such as gross national product have been modeled as stationary processes about a deterministic trend. All aggregate fluctuations were thus short-run phenomena with no bearing on the long-run behavior of the economy. Starting with the work of Charles Nelson and Charles Plosser, however, empirical workers have developed considerable evidence that suggests that some component of aggregate activity follows a stochastic trend-the long-run path of the macroeconomy is permanently affected by contemporary events. This perspective means that not only are trend-cycle decompositions extremely difficult,i n that the same structural stochastic elements affect both underlying time series, but that, in addition, the feedback mechanisms from current activity to long-run growth render the traditional distinction meaningless.