Content area
Full Text
Abstract
This paper examines predictive power of the confidence indicators for developments in industrial output, producer prices and employment in the Czech and Slovak Republics, Hungary, and Poland (V4 countries). The Granger Causality tests are used for establishing potential causation between the confidence indicators and real economy data. The best OLS models with autoregressive terms complemented by confidence indicators are selected and their predictive accuracy is tested against the ARMA benchmarks with the Diebold-Mariano test. All OLS models performed better than the naïve ones . We conclude that the actual CI variables seem to reflect future patterns of economic development in next 1 - 2 months, and not just opinions by economic agents based on current or past economic trajectories.
Keywords: forecasting, confidence indicators, ARMA models
JEL Classification: E37, C53
Introduction: Confidence Indicators
Short-term forecasts ('nowcasts') provide policy makers and business agents with valuable knowledge on current and near-future trends in national economy. There is high demand on timely and reliable information on output, prices and employment in sectors and industries of a national economy. Timely and reliable information is no easy to get. Most important economic data are published with significant delay of two or three months. The January data on industrial production, for example, are published in mid-March in most EU Member Countries. Confidence ('soft') indicators (CI) are alternative source of information on near-future trends. They usually are published by end of current month and present expectations by businesses on developments in next three months. Confidence indicators account for a number of advantages over hard statistical data: (a) early release (one-three months before publications of hard data for most time series); (b) limited amount of follow-up corrections and revisions, and (c) signals on expected economic activity in key sectors of national economy provided by relevant economic agents (mostly business leaders in particular economic sectors). The industry confidence indicator therefore brings more and timely information about the evolution of gross domestic product and/or of industrial production index (Gagea, 2012; 2014).
There is plethora of research evaluating performance of forecasting models with the confidence indicators in OECD Member Countries. Most studies concentrate on short-term forecasts of gross domestic product (GDP), and use industrial confidence indicators (ICI) and Economic Sentiment Indicator (ESI) variables (provided...