Record Details

A NEW MODEL WITH REGIME SWITCHING ERRORS: FORECASTING GDP IN TIMES OF GREAT RECESSION

Ekonomika

View Archive Info
 
 
Field Value
 
Title A NEW MODEL WITH REGIME SWITCHING ERRORS: FORECASTING GDP IN TIMES OF GREAT RECESSION
 
Creator Bartkus, Algirdas
 
Subject —
univariate time series models; forecasting and prediction methods; autocorrelated errors; threshold models; GDP forecasting
—
 
Description This paper investigates the possibility to obtain better GDP forecasts in the early stages of Great Recession. Here, predictive performance refers to exclusively out-of-sample forecasts. Based on exploratory data analysis and general-to-specific modelling, this paper proposes a univariate predictive threshold model for the small open economy that outperforms its linear counterparts and correctly determines the course of events. This model does not explain any causal links; however, based on a set of economic arguments, it sets forward an idea regarding how a forecaster can act when principal determinant factors, responsible for a sudden, yet lasting change, are unknown, unmeasurable or cannot be influenced by national policy makers. A major dissimilarity between usual threshold models and the model presented in this paper is that while variables act differently under different conditions in the former, in this model, due to economic reasons, errors act differently. Alternatively, this paper can be viewed as a comparative GDP prediction study.
 
Publisher Vilniaus universiteto Ekonomikos fakultetas / Vilnius University Faculty of Economics
 
Contributor —
 
Date 2016-10-03
 
Type info:eu-repo/semantics/article
info:eu-repo/semantics/publishedVersion
Peer-reviewed Article
 
Format application/pdf
 
Identifier http://www.zurnalai.vu.lt/ekonomika/article/view/10122
10.15388/Ekon.2016.2.10122
 
Source Ekonomika; Ekonomika 2016 95(2); 7-29
1392-1258
1392-1258
 
Language lit
 
Relation http://www.zurnalai.vu.lt/ekonomika/article/view/10122/8178
 
Rights Autorinės teisės (c) 2016 Ekonomika