|
on Forecasting |
| By: | Pinkwart, Nicolas |
| Abstract: | We present a comprehensive disaggregate approach for short-term forecasting economic activity in Germany by explicitly taking into account the supply or production side and the demand side of GDP. The GDP figures calculated by the two sides usually yield different results and the official GDP release is somewhere in between. We make use of this statistical procedure by separately modeling the two sides of GDP in a system of bridge equations at the most disaggregate level available and combining the resulting two aggregate GDP forecasts. Comparing several specification schemes in an out-of-sample forecast evaluation setup, we are able to find informative forecasts for most of the underlying GDP components. We then show first, that both approaches already yield informative aggregate forecasts for forecast horizons of up to 28 weeks and second, that combining the production side and the demand side projections substantially improves the forecast performance, in particular for the shorter forecast horizons. |
| Keywords: | German Economy,GDP,Disaggregation,Forecasting,Nowcasting,Bridge Equations |
| JEL: | C22 C53 E32 E37 |
| Date: | 2018 |
| URL: | http://d.repec.org/n?u=RePEc:zbw:bubdps:362018&r=for |
| By: | Sergey Ivashchenko (The Institute of Regional Economy Problems (Russian Academy of Sciences), St. Petersburg, Russia; National Research University Higher School of Economics, St. Petersburg, Russia; The Faculty of Economics of Saint-Petersburg State University, St. Petersburg, Russia and Financial Research Institute, Ministry of Finance, Russian Federation, Moscow, Russia.); Semih Emre Çekin (Department of Economics, Turkish-German University, Istanbul, Turkey); Kevin Kotzé (School of Economics, University of Cape Town, Rondebosch, South Africa.); Rangan Gupta (Department of Economics, University of Pretoria, Pretoria, South Africa) |
| Abstract: | This paper compares the out-of-sample forecasting performance of first- and second- order perturbation approximations for DSGE models that incorporate Markov-switching behaviour in the policy reaction function and the volatility of shocks. These results are compared to those of a model that does not incorporate any regime-switching. 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: | C13 C32 E37 |
| Date: | 2018–09 |
| URL: | http://d.repec.org/n?u=RePEc:pre:wpaper:201862&r=for |
| By: | Jahn, Malte |
| Abstract: | Artificial neural networks have become increasingly popular for statistical model fitting over the last years, mainly due to increasing computational power. In this paper, an introduction to the use of artificial neural network (ANN) regression models is given. The problem of predicting the GDP growth rate of 15 industrialized economies in the time period 1996-2016 serves as an example. It is shown that the ANN model is able to yield much more accurate predictions of GDP growth rates than a corresponding linear model. In particular, ANN models capture time trends very flexibly. This is relevant for forecasting, as demonstrated by out-of-sample predictions for 2017. |
| Keywords: | neural network,forecasting,panel data |
| JEL: | C45 C53 C61 O40 |
| Date: | 2018 |
| URL: | http://d.repec.org/n?u=RePEc:zbw:hwwirp:185&r=for |
| By: | Adam Richardson; Thomas van Florenstein Mulder; Tugrul Vehbi |
| Abstract: | This paper analyses the real-time nowcasting performance of machine learning algorithms estimated on New Zealand data. Using a large set of real-time quarterly macroeconomic indicators, we train a range of popular machine learning algorithms and nowcast real GDP growth for each quarter over the 2009Q1-2018Q1 period. We compare the predictive accuracy of these nowcasts with that of other traditional univariate and multivariate statistical models. We find that the machine learning algorithms outperform the traditional statistical models. Moreover, combining the individual machine learning nowcasts further improves the performance than in the case of the individual nowcasts alone. |
| Keywords: | Nowcasting, Machine learning, Forecast evaluation |
| JEL: | C52 C53 |
| Date: | 2018–09 |
| URL: | http://d.repec.org/n?u=RePEc:een:camaaa:2018-47&r=for |