Multi-Horizon Forecast Comparison of Linear and Non-linear Methods for GDP Growth and Inflation

Forecasting macroeconomic variables is an important issue for decision makers. In thispaper, we analyze 13 univariate models which include linear and non-linear methods to forecastGDP growth and inflation to determine a benchmark. We not only examine the models in singlehorizon by the Diebold-Mariano test, but also apply the multi-horizon analysis from Quaedvlieg(2018). By conducting the research on 4 European countries with different fundamentals, wefind thatAutoregressive methodwith fixed lags performs the best in our sample, which is thebenchmark model that we propose. The multi-horizon superior predictive ability(SPA) test canbe an useful tool in comparing large set of forecasts which delivers reliable results.