Gloria Gonzalez-Rivera

Gloria Gonzalez-Rivera

University of California, Riverside USA

Gloria González-Rivera is Professor of Economics at the University of California Riverside. She received her Ph.D. from the University of California San Diego writing her dissertation under the supervision of 2003 Nobel Laureate Professor Robert F. Engle. Her research focuses on the development of econometric and forecasting methodology with applications to financial markets, volatility forecasting, risk management, and agricultural markets. She is the sole author of the textbook Forecasting for Economics and Business (Pearson, 2012). Professor González-Rivera is a Fulbright Scholar. She was awarded the UC-Riverside University Scholar Distinction (2007-2011) as well as several teaching awards at UC-San Diego. She is Associate Editor for the International Journal of Forecasting, guest editor for the special IJF issue on predicting rare events, and she is currently editing volume 36 of Advances in Econometrics. Her research has been funded by the US National Science Foundation, the Giannini Foundation, California Native Indian Gaming Association, and the UC-Agricultural and Natural Resources Network, among others. She served as Chair of the Economics Department (2003-2008) and Vice-Chair (2010-2011) of the Graduate Council of UC-Riverside. She has been elected President of the International Institute of Forecasters (her tenure starts on May 2016).

Research stay at UC3M: DEPARTMENT OF STATISTIC

Project: Specification and Forecast Evaluation of Time Series Models. Extensions of the Autocontour Methodology

This project focuses on relevant extensions of the econometric testing methodology introduced by González-Rivera and her co-authors (2010, 2011, 2012, 2014). Within the context of dynamic specification and forecast evaluation of time series models, González-Rivera et al. (2010, 2011) introduce a device -the autocontour- that is the basis for the construction of a battery of specification tests for the joint hypothesis of i.i.d-ness and conditional density functional form. Since the shape of the autocontour is very sensitive to departures from the null in either direction, the autocontour-based (ACR) tests enjoy superior power vis-a-vis other proposed statistics. However, there are technical limitations to the implementation of the original tests and González-Rivera and Sun (2014) proposed a generalization (G-ACR) that is applicable to univariate and multivariate models alike as well as in-sample and out-of-sample environments. In this project, we propose to investigate further improvements that will permit the statistical comparison of several competing specifications. The key idea is to construct a scalar measure that summarizes the degree of uniform-ness of each model (i.e. location of the transformed observations within the maximum autocontour) and, based on that, to build appropriate statistical tests whose properties will be analyzed. We will explore the connection of these ideas to the literature on scoring rules and information theory.

Stay period: SEP 2015 - FEB 2016