Herd behavior in consumer inflation expectations - Evidence from the French household survey.

This article investigates whether the formation of individual inflation expectations is biased towards a consensus and is thus subject to some kind of herding behavior. Basing on the traditional Carlson-Parkin approach to quantify qualitative survey expectations and its extension by Kaiser and Spitz (2002) in an ordered probit context, a method to gain individual level inflation expectations is proposed using a Markov chain Monte Carlo Hierarchical Bayesian estimation method. This method is applied to micro survey data about inflation expectations of households from the monthly French household survey “Enquête mensuelle de conjoncture auprès des ménages - ECAMME ” (January 2004 to December 2012). Finally a modified version of the non-parametric test for herding behavior by Bernardt et al. (2006) is conducted on the cohort-level expectation estimates, showing that the expectation formation is not subject to a bias towards the consensus expectation. In contrast, it exhibits a strong anti-herding tendency which is consistent with the findings of other studies (Rülke and Tillmann, 2011).