Forecasting with Second-Order Approximations and Markov-Switching DSGE Models
- PDF / 591,777 Bytes
- 25 Pages / 439.37 x 666.142 pts Page_size
- 58 Downloads / 190 Views
Forecasting with Second-Order Approximations and Markov-Switching DSGE Models Sergey Ivashchenko1,2,3,4 · Semih Emre Çekin5 · Kevin Kotzé6 Rangan Gupta7
·
Accepted: 22 October 2019 © Springer Science+Business Media, LLC, part of Springer Nature 2019
Abstract This paper considers the out-of-sample forecasting performance of first- and secondorder perturbation approximations for DSGE models that incorporate Markovswitching behaviour in the policy reaction function and the volatility of shocks. The results suggest that second-order approximations provide an improved forecasting performance in models that do not allow for regime-switching, while for the MS-DSGE models, a first-order approximation would appear to provide better out-of-sample properties. In addition, we find that over short-horizons, the MS-DSGE models provide superior forecasting results when compared to those models that do not allow for regime-switching (at both perturbation orders). Keywords Regime-switching · Second-order approximation · Non-linear MS-DSGE estimation · Forecasting JEL Classifications C13 · C32 · E37
1 Introduction Dynamic Stochastic General Equilibrium (DSGE) models are frequently used by academics and researchers at public institutions for policy-analysis and forecasting
Electronic supplementary material The online version of this article (https://doi.org/10.1007/s10614019-09941-8) contains supplementary material, which is available to authorized users.
B
Kevin Kotzé [email protected]
Extended author information available on the last page of the article
123
S. Ivashchenko et al.
purposes.1 Most of these models make use of a framework that is based on linear first-order perturbations, where it is assumed that the sample of data that is used in the estimation is not subject to any regime-switching behaviour. In a recent review of research that has been conducted using DSGE models, Christiano et al. (2018) note that the extensive application of first-order approximations for the model solution may be motivated by the fact that these models appear to provide an accurate characterisation of the effects of small shocks that arose during the post-war period in the United States. In addition, these techniques also allow researchers to make use of linear filters and estimation methodologies that are not subject to the computational complexity of non-linear counterparts. However, despite the attractive features of linear models that employ first-order approximations for the model solution, there are those who suggest that these models may fail to capture many of the non-linear features that are present in macroeconomic data. For example, Stiglitz (2018) notes that the use of linear approximations in such a macroeconomic model would be inappropriate, as it would not provide an accurate description of the effects of large shocks. In addition, FernándezVillaverde et al. (2016) and Schmitt-Grohé and Uribe (2004) have suggested that the use of higher-order perturbation techniques could provide improvements in terms of the accuracy
Data Loading...
