Mental Macro-Finance Models: Evidence and Theory

Abstract

How do people understand business-cycle comovements of macro-finance variables? I develop an empirical and theoretical framework to investigate people’s mental models, extending the univariate Coibion and Gorodnichenko (2015) regression to address multivariate relations. Forecast revisions of related variables predict forecast errors, with coefficients sharing the same sign as the cross-variable relations, for consensus and individual forecasts. I present a new moment of cross predictability across variable pairs, and propose a theory with misperception of cross-variable relations, whose calibrated value is less than one half. Taking into account variable comovements and misperception thereof is also crucial for interpreting previous univariate findings.

Keywords: Mental models, macro-finance, belief formation, misperception, misreaction

JEL Classification: C3, C53, D83, D84, E17, E31, E32, E43, E44, G12