Possibilities of Estimations of Short-term Macroeconomic Aggregates Based on Business Survey Results

Abstract

The aim of the article is to construct a model for estimating the quarterly gross value added (GVA) of the national economy (GDP) based on the results of business surveys (so-called confidence indicators) in industry, construction, commerce and services (incl. banking sector), and to set the forecast for four quarters ahead. The suitability of the applied approach is assessed using pairwise dependencies for individual sectors. In the case of both pairwise and multidimensional dependencies, the authors proceed from a linear dynamic model, which is a combination of ARIMA models (or SARIMA models) in conjunction with regression analysis, where the variables explained are time-shifted. The quality of the estimated models is proven to be very high. The analysis shows a significant link between the sector's gross value added and sectoral confidence indicators. Significant predictors of the GVA of the national economy and GDP show explanatory variables of confidence indicators in industry and construction, whereas indicators of confidence in trade and services were statistically insignificant. Timely knowledge of these indicators in conjunction with linear dynamic models allows better and faster predictions of quarterly GVA and GDP than with conventional time series models.