Consumer Survey Data and short-term forecasting of households consumption expenditures in Poland

The main aim of the paper is an investigation whether consumer tendency survey (CTS) data can be useful for short-term forecasting of real private consumption expenditures in Poland. A consumer sentiment index and as well balances for individual questions (from consumer survey conducted by Research Institute for Economic Development of Warsaw School of Economics – RIED WSE) were used in the paper. This is one of the first attempts to model private consumption using RIED consumer survey data or even generally any consumer survey data in Poland. The data are quarterly and covers the 1996-2007 period. An investigation of predictive relationship between CTS results and real private consumption was done by using in-sample and out-of-sample approach. Under the first one unconditional and conditional causality analysis was performed. Unconditional approach is a standard bivariate Granger causality test, where appropriate pairs of variables were tested: real private consumption expenditures and particular CTS indicators. In the case of conditional Granger causality test aside from consumer survey results other macro variables were used, i.e.: lagged dependent variable, real gross disposable income of households, net financial assets of households, retail sales and wage-pension fund. It was realized by doing redundancy tests for survey indices in ARMAX equation specified for consumption expenditures and other variables mentioned above. The research is also extended by assessing whether this predictive in-sample conditional relationship has changed over time by running a series of recursive regressions and doing tests outlined above. The out-of-sample analysis focused on testing recursive one-step forecast errors of restricted (without CTS variables) and unrestricted ARMAX models.