Normal07.8 磅02falsefalsefalseEN-USZH-CNX-NONEMicrosoftInternetExplorer4Topic：Autocontour-Testing in Unstable EnvironmentPresenter:Sun Yingying

Time:Oct.23 15:30-16:45

Location:EMS-A511

Abstract:

In this paper, we propose the out-of-sample density forecasting evaluation method in the presence of the instabilities based on Generalized Autocontour Method. We define the instabilities as time variation in the density function of a stochastic process. These variations include changes of mean, variance and/or the functional form of the underlying density function. To take care of the instability, we evaluate one subsample of the evaluation sample, using data from t−rol+1 up to t, to evaluate the assumed predictive density. According to Generalized Autocontour, for one subsample, we can obtain three different types of statistics to take care of the instabilities. We construct the Sup-type statistic and the Avg-type statistic, which explore the supreme and average behavior of the environmental instabilities. The asymptotic distributions of the statistics constructed in this paper are functional of standard Brownian motions and have good finite sample properties. We have applied our tests to evaluate the density forecast performance of U.S. inflation produced by linear and Markov-switching Philips Curve. We conclude that Markov switching Philip Curve can provide a good out-of-sample density forecast for U.S. inflation in the presence of instabilities.

Bio：Dr. Sun earned her Ph.D. in economics from University of California, Riverside in May 2013. She received a master degree in economics from Wuhan University in 2007, and a bachelor degree in management from Huazhong Normal University in 2005. Her research focuses on econometric theory, time series analysis, and econometric forecasting. Her recent research has been published in the Journal of International Forecasting.