CEU eTD Collection (2021); Wang, Xinqi: Choosing the Best Model - An Out-of-sample Testing for Gradual Physical Impact Model

CEU Electronic Theses and Dissertations, 2021
Author Wang, Xinqi
Title Choosing the Best Model - An Out-of-sample Testing for Gradual Physical Impact Model
Summary Physical risks are often categorized into gradual (slow-onset) impacts and extreme weather events. This article estimates the gradual physical impacts using the relationship between increasing temperature and aggregate productivity (GDP) on 37 countries from 1961 to 2019. I follow the econometric model developed by Burke et al. (2018) and Burke and Tanutama (2019) to first perform an in-sample forecasting of a static and a dynamic model. I confirm that country-specific linear and quadratic time trends are statistically significant in the model. I then perform an out-of-sample forecasting with rolling window to compare models and to check whether a dynamic panel model is statistically more appropriate than a static panel model. I later confirm the results with a Giacomini and White test that a dynamic panel model is indeed a better model to use, and this outperformance of the dynamic panel model is statistically significant comparing with the static panel model.
Supervisor Bogel, Gyorgy
Department Economics MSc
Full texthttps://www.etd.ceu.edu/2021/wang_xinqi.pdf

Visit the CEU Library.

© 2007-2021, Central European University