Growth States and Shocks

Abstract

GDP growth typically vibrates with modest variation around a mean of a few percent per year, but periodically, mean growth undergoes a major shift, vibrating thereafter around a new level. I present a transmission mechanism with nonlinear dynamics that endogenously translates random sectoral shocks into just this sort of behavior, creating what might be thought of as multiple growth states. Small shocks cause vibration within a state. Sufficiently large shocks cause a state-change. Behind the nonlinear dynamics lies essentially the same dynamic externality that drives the "new growth theory" models. The relative balance of endogenous versus exogenous growth determines whether the economy will have multiple stable growth states. The model can generate data which looks much like the data generated by a Markov process of the sort identified in Hamilton (1989) or a trend-breaking process as in Perron (1989). The model has two output processes. Input factors are drawn into the "leading process," where learning-by-doing further increases that process's technological lead. If the "leading process" is also the inherently high-growth process, then growth is fast for both technologies. Shocks to preferences and technologies cause endogenous switching of the leading sector role between the high-growth and slow-growth processes.

JEL Classification: E32