Date of Degree
The research presented here is a comprehensive analysis of research on the "Great Moderation" and its impact on business cycle modeling. In the presence of a less volatile aggregate economy, the methods of modeling business cycles have fundamentally changed along with the ability to detect turning points in the business cycle using standard algorithms. Chapter One lays out the historical case for modeling the business cycle in a manner placing importance on the ability of a model to replicate features observed in actual GDP data, such as the depth and length of recessions, or the average height of expansions. Chapter Two compares different business cycle models by their ability and accuracy in reproducing features of observed GDP data in simulated Monte Carlo paths. Comparisons are made by examining how volatility moderation affects business cycle modeling for the U.S., U.K., and Australia. Univariate ARIMA, structural change, and Markov-switching ("MS") models are estimated and used to simulate time paths using Monte Carlo methods. These results generally support previous findings that MS models are superior to linear models and comparable to structural change models at fitting business cycle characteristics. Tests show that to replicate business cycle characteristics, MS models must account for independent shifts in mean and volatility parameters. Substantial new evidence shows that commonly specified MS models with a simple linear structure, constant variance, or state-dependent volatility are sub-optimal and should be avoided in practice. Results indicate that models attempting to replicate business cycle features in any series should consider the importance of how volatility is modeled prior to estimation. Evidence is also presented showing that the Great Moderation may have recently ended. Chapter Three examines algorithm robustness used to conclude that independent switching models are better able to replicate business cycle features. Robustness is tested by varying the parameters of dating algorithms used to detect turning points. Evidence shows that the "window" and "censor" used for turning point selection criteria does not lead to substantial changes in the conclusions of our previous findings implying that our results are not artifacts of the algorithm, but due to the actual economic model itself.
Neveu, Andre R., "Essays on the Impacts of the Great Moderation on Business Cycle Modeling" (2009). CUNY Academic Works.