Do Long Swings in the Business Cycle Lead to Strong Persistence in Output?

In this article we extend the Hamilton Markov chain regime switching model of real aggregate output to include the possibility of an occasional long regime. We model this long regime feature with a fat tailed duration distribution where the duration length can have high levels of variability and take on extreme values. We show that an economy with a fat tailed expansion and/or contraction duration leads to long memory behavior in aggregate output. In addition, we nd empirical support for our theoretical ndings with estimates of the degree of occasional long swings in the US business cycle when the length of the expansion and contraction regimes are de ned by the National Bureau of Economic Research recession dates. Our estimates of the tail index for the length of US economic booms and busts closely correspond to the magnitude of the long memory parameter estimated by Diebold and Rudebusch (1989) and Sowell (1992) for real US output, and suggest that a forecast of aggregate output based on a long history of past observations will be inferior to a projection that uses observations from the current regime.