On the accuracy of the probability method for quantifying beliefs about inflation

This paper assesses the probability method for quantifying EU consumer survey data on perceived and expected inflation. Based on household level data from the Swedish consumer survey that asks for both qualitative and quantitative responses, it is found that the theoretical assumptions of the method do not hold. In particular, estimates of unrestricted response schemes indicate that response intervals are asymmetric and that qualitative inflation expectations are formed relative to perceptions of current inflation. Nevertheless, the probability method generates series that are highly correlated with the mean of actual quantitative beliefs. For quantifying the cross-sectional dispersion of beliefs, however, an index of qualitative variation is most accurate.