Modely predstihových indikátorov pre Solvenskú republiku - Aplikácia empirických a teoretických prístupov

Abstract

Real gross domestic product (GDP) is frequently used as a headline indicator for business cycle approximation measuring the acceleration and deceleration of aggregate economic activity. The official GDP estimate is published with a delay of several weeks after the reference period. Macroeconomic indicators published ahead of the official GDP figure or passing through the turning points of the business cycle with a significant lead are called leading indicators. The leading indicators are carrying potential signals of the upcoming phase of the business cycle and are used in practice as a short-term forecasting tool. Fiscal and monetary authorities must follow also the economic interpretation of actual economy development in order to achieve their set targets. With the aim to interpret the actual economy in terms of economic theory a structural vector-autoregression model (SVAR) is estimated for the Slovak economy using main macroeconomic aggregates allowing identification of demand and supply shock, as well as specific fiscal and monetary shocks. The model is embedded with leading indicators to increase its forecasting ability in the short-term period. The forecasting ability of the SVAR model is compared with two econometric models without insight into economic theory – based on interpretation of the business cycle as observed variable (vector-autoregression model, VAR) and unobserved variable (dynamic factor model, DFM). Econometric models are estimated in quarterly frequency in the sample from 2002 to 2016. Out of sample forecasts of Slovak GDP for three quarters ahead are tested in the period from 2008 to 2016. The SVAR model specified by econometric theory produces the least biased and most accurate forecasts in the forecasting period among all models. However, at all forecast horizons the combination of all three types of models produces the most accurate forecasts. The forecast combination in the horizon of three quarters produces forecast with a deviation below 0.1 percentage points of GDP quarterly growth with an average square error within 0.2 percentage points of GDP growth. Following the estimation of the SVAR model in the sample from 2002 to 2016, the supply shocks are the dominant source of fluctuations of aggregate economic activity, persistent demand shocks play a minor role. From the policy shocks point of view, the monetary policy shocks are more significant before the entry of the Slovak Republic into the euro area in 2009, in the sample from 2010 to 2016 we observe more dominant influence of fiscal impulses, suggesting a substitutionary character of both policies.