Forecasting of recessions via dynamic probit for time series: replication and extension of Kauppi and Saikkonen (2008)

摘要

In this work, we first replicate the results of the fully parametric dynamic probit model for forecasting US recessions from Kauppi and Saikkonen (Rev Econ Stat 90(4):777-791, 2008) [which is in the spirit of Estrella and Mishkin (Rev Econ Stat 80(1):45-61, 1998) and Dueker (Rev Fed Reserve Bank St Louis 79(2):41-51, 1997)] and then contrast them to results from a nonparametric local-likelihood dynamic choice model for the same data. We then use expanded data to gain insights on whether these models could have warned the public about approach of the latest recession, associated with the Global Financial Crisis. Finally, we also apply both approaches to gain insights for 2018.