Do inflation expectations matter in a stylised New Keynesian model? The case of Poland

This paper estimates different versions of a stylised New Keynesian model of the Polish economy, in which alternative measures of inflation expectations are used. They include: model-based (rational) expectations as well as survey measures of inflation expectations formed by consumers, enterprises and financial sector analysts. After estimating the models we verify to what extent the use of different measures of inflation expectations affects the assessment of the monetary transmission mechanism, the exchange rate pass-through and the sacrifice ratio. Simulation results show different responses in all the analysed areas. For example, the maximum reaction of CPI inflation to the interest rate impulse is almost twice bigger if the direct measures of financial sector analysts are used instead of model-consistent expectations. Also the model with survey-based measures of producer inflation expectations displays stronger response of inflation to monetary policy impulse than the model, in which rational expectations are assumed. Estimates of the exchange rate pass-through from the models with survey-based expectations are very similar to each other, but stronger than in the model with rational expectations. The sacrifice ratio seems similar in the case of all versions of the New Keynesian model except its version with consumer inflation expectations that shows significantly larger output loss resulting from a permanent reduction of the inflation target than the other models. Differences in the assessment how monetary factors affect macroeconomic variables, particularly inflation, pose the question which model should be treated as the most adequate. To answer this question we run in-sample simulations, calculating inflation forecasting errors of all the models under consideration. We conclude that the model that assumes rational inflation expectations displays the lowest forecasting accuracy, while the model using inflation expectations of enterprises is the best-performing model. It suggests that the assumption of rational inflation expectations does not match the actual data well. Inflation expectations of Polish enterprises seem the most relevant from the macroeconomic point of view – more relevant than inflation expectations of consumers and financial sector analysts.