Forecasting with DSGE-Model: the Case of Institutional Differences between Countries

The ECB traditionally considers price stability as its main purpose, while all other goals are considered as secondary. With the aim to support price stability the ECB accepted the "two-pillar" rule, which is based on a comprehensive risk analysis and in particular on the role of money growth in controlling price dynamics. Due to the fact that this feature often is not reflected in the modeling of monetary policy in new Keynesian models, a research question was formulated as following: whether the including of a "two-pillar" rule in the two-country dynamic stochastic general equilibrium (DSGE) model can improve the quality of forecasting of macroeconomic variables in comparison with a model that takes into account only different interest rate rules, as well as with a model with an identical interest rate rule. However, due to the need to check the quality and reliability of the received forecasts, it was decided to also focus on the BVAR model as a basic forecast model. The main goal of the work was to obtain more accurate and effective forecasts within the two-country DSGE model as a result of taking into account the institutional differences of Central banks. Current work considered several types of two-country DSGE models (for the US and the Euro Area) according to several institutional differences: a model with the same interest rate rules (same standard Taylor rules); a model with different interest rate rules (different Taylor rules) and a model that includes both different Taylor rules, as well as a "two-pillar" rule. The quality of forecasts was assessed based on the RMSE quality metric and the statistical Diebold, Mariano test (1995). As a result, it was found that including "two-pillar" rule in model actually can improve the predictive power of the DSGE model for such indicators as output gap and CPI-inflation, and gives statistically significantly better forecasts for these variables, especially for short-and medium-run horizons. However, it does not exceed the quality of the BVAR forecasts. The best forecast model of the DSGE class for inflation, calculated using PPI and the interest rate, was the model with divergent Taylor rules, but it was also not so gut as BVAR forecasts.