講座題目：Do Professional Forecasters consider Others' Forecasts in their Phillips Curve Belief? Evidence from Functional Data Analysis

講座時間：2022年11月29日星期二下午18:00-19:30

講座方式：騰訊會議ID：880-827-750

密碼：221129

講座鏈接：https://meeting.tencent.com/dm/FjnslReB8rED

主 講 人：

Shixuan Wang is an Associate Professor in Economics at the University of Reading. His main research focuses on the econometrics of change-points (structural breaks), functional data analysis, and empirical applications of those in macroeconomics and finance. Additionally, he is also interested in using techniques of data analytics and machine learning for operations management. His research has been published in Annals of Statistics, Journal of Econometrics, Journal of Business & Economic Statistics, European Journal of Operational Research, among others. He serves as an Associate Editor of the International Journal of Finance and Economics (Wiley) which has an Association of Business Schools 3* ranking.

Before joining the University of Reading in 2018, he was a postdoctoral research associate in statistical forecasting at Cardiff Business School, Cardiff University, where he worked on an Engineering and Physical Sciences Research Council (EPSRC) project. In 2017, he obtained his doctoral degree at University of Birmingham with full scholarship from the Economic and Social Research Council (ESRC), UK. He was awarded a Royal Economic Society (RES) junior fellowship in 2016.

講座簡介：

We employ tools in functional data analysis to examine whether professional forecasters’ belief in the Phillips Curve is affected by others' historical forecasts. To efficiently extract most useful information, we use functional principal component analysis on the density curves of the professional forecasters’ historical forecasts, which needs a suitable transform in order to be viewed as functional data in a Hilbert space. Our results show that the first three functional principal components are nonlinear measures of disagreement, skewness, and kurtosis. Based on the functional principal component regression, 45% of professional forecasters do consider others' historical forecasts, which affects their Phillips curve belief. Specifically, professional forecasters mostly consider the disagreement of others' forecasts, while skewness and kurtosis are also occasionally used.

This is a joint work with Michael P. Clement.