Global Evidence on Misperceptions and Preferences for Redistribution

Abstract

Individuals often hold erroneous beliefs about their socio-economic status relative to others. We develop a new machine learning technique to measure these misperceptions and use large-scale international survey data to compute status misperception for 241,757 households from 97 countries (24 OECD, 73 non-OECD). We show that status misperception is a widespread phenomenon across the globe. Upward-biased perceptions are associated with lower preferences for redistribution and have direct consequences for welfare provision via the tax and transfer system. The effect accounts for approximately 9% of the variation in redistribution preferences, is independent of socio-demographic characteristics, robust to measurement errors in social surveys, and occurs similarly when we change the underlying micro data or examine party preferences.

Keywords: misperceptions, machine learning, socio-economic status, preferences, redistribution, welfare provision, taxes and transfers

JEL Classification: D310, H530, I300, C430