Business Survey Data in Forecasting Macroeconomic Indicators with Combined Forecasts

Abstract

Current economic crisis shed dark light on the possibilities of creating a valuable and reliable short and medium term forecasts with the use of the most commonly applied econometric models in the structural or autoregressive form (SVAR, VAR), but also models of the general equilibrium (CGE, DSGE). The models failed to forecast especially at the verge of the crisis when the information on upcoming peak in the business cycle would be of the highest value. This situation was a stimulus to undertake research oriented at creating a family of models that would react faster and with higher precision to dynamic changes in the economic environment. As a result it is expected that a family of models will be specified, identified and estimated. They should provide leading and more accurate information on basic macroeconomic variables – GDP, unemployment and inflation. Each of the specifications will be subject to two objectives: (1) the minimum ex-ante forecast error and (2) immediate and reliable accessibility of data. The database applied in the procedure will comprise of time series from the Research Institute of Economic Development (RIED) on sentiment in manufacturing industry, households, trade and construction. The series on economic activity in Poland cover the period of 1995–2009.