A Structural Break in U.S. GDP?
26 Pages Posted: 13 Aug 2003
Date Written: July 2003
The volatility of growth in U.S. real GDP declined dramatically in the mid-1980s. Viewed through the lens of linear autoregressive models, this phenomenon appears to be the result of a structural break in the innovation process that drives GDP fluctuations. We present an alternative model that is structurally stable over the entire post-war period. This model characterizes growth as following non-linear trajectories that fluctuate stochastically between alternative periods of general acceleration and deceleration. The specific trajectories can differ across regimes due to randomness in the parameters that govern their behavior; but the underlying distribution from which these parameters are realized is stable. The model also allows the variance of growth-rates innovations to differ across regimes, as a function of their corresponding regime-specific trajectories. We find no evidence of a structural break when viewing GDP growth through the lens of this model. Moreover, using an encompassing exercise, we show that this model can account for the evidence favoring a structural-break interpretation of the volatility reduction under linear autoregressive processes. The model also encompasses general patterns of business-cycle activity.
Keywords: business cycles, encompassing, regime switching, error correction, conditional heteroscedasticity
JEL Classification: C22, C51, C52
Suggested Citation: Suggested Citation