The Good, the Bad and the Ugly: Analyzing Forecasting Behavior within a Quantal Response Framework with Misclassification

Abstract:     
This paper presents a new approach of analyzing qualitative forecasting errors made by forecasters in tendency surveys. Based on a quantal response approach with misclassification we are able to define qualitative mispredictions of forecasters in terms of deviations from the qualitative rational expectation forecast and relate them to individual and macro factors driving these mispredictions.

Through the introduction of a dynamic, Markov type misclassification matrix our approach accounts for individual heterogeneity in forecasting behavior. It enables a detailed analysis of individual forecasting decisions allowing for the introduction of individual and economy wide determinants influencing the individual expectation formation process. The model can also be used to test for individual deviations from specific behavioral aspects of expectation formation (adaptive expectations, learning, focalism, etc.) at the macro level.

The model is estimated by maximum likelihood using a logistic generalized ARMA structure for the misclassification matrix based on the Financial Markets Survey of the Centre for European Economic Research (ZEW), a monthly qualitative survey of around 330 financial experts, giving six-month-ahead predictions of major macroeconomic aggregates and financial indicators.

Number of Pages in PDF File: 32

Keywords: Expectations, Tendency Survey, Forecasting Errors, Misclassification, GLARMA

JEL Classification: C23, C25, D84, E27

working papers series