Much of what is known about the evolution of the U.S. economy is based on statistical descriptions of the behavior of individual aggregate time series or of the relations among key series. This project continues and expands previous work on quantifying the sources of aggregate fluctuations. It consists of three parts. The first is primarily empirical, and focuses on the role of human capital accumulation (learning-by-doing) as a mechanism whereby short-run fluctuations have permanent effects. The second is also primarily empirical and concerns the predictability of recessions and expansions (as opposed simply to fluctuations in levels of growth rates) and the implications of this predictability for economic theory. The third is primarily one of theoretical econometrics and develops techniques for inference about orders of integration of time series variables. This project makes important methodological and substantive contributions to economics. The first part of the project provides one of the first systematic empirical investigations of the link between long-run growth and shocks traditionally viewed as purely temporary such as transitory shifts in monetary policy or in preferences. This has important implications for our understanding of the sources and the transmission of economic fluctuations. The second part builds on recent joint work aimed at developing a revised Index of Leading Economic Indicators. It could provide better predictions of future recessions and expansions. The third part yields new tools for applying time series methods to economic problems.
