Discussion of “Risk Preference Types, Limited Consideration, and Welfare” by Levon Barseghyan and Francesca Molinari

Abstract

This article is part of an impressive research agenda by the authors which develops tools to identify models of risk preferences (Barseghyan, Prince, and Teitelbaum 2011; Barseghyan et al. 2013, 2018, 2021; Barseghyan, Molinari, and Teitelbaum 2016; Barseghyan, Teitelbaum, and Xu 2018; Barseghyan, Molinari, and Thirkettle 2021). Such work is prominent in industrial organization, development, health, labor, finance, and public economics because it is pivotal to studying incentives and assessing the welfare impact of policy interventions in insurance markets. In this article, the authors provide a novel method to identify a static model of decision-making under risk, where agents choose insurance bundles over multiple lines of property coverage, belong to different preference types, display unobserved heterogeneity in attitudes toward risk, and may consider a limited amount of bundles when making their choices. This rich framework is critical for rationalizing data patterns but introduces substantial econometric challenges. The crucial insight consists of exploiting the single crossing property (SCP) that the model features within each coverage context and an exclusion restriction to characterize the response to changes in the covariates of the choice probability of the cheapest bundle. From these elasticities, we can identify the type shares and the distribution of unobserved heterogeneity and consideration sets for each type. I devote the first part of the discussion to summarizing the identification strategy and giving context to the novelty of the arguments. In doing so, I applaud the authors for expertly and smoothly guiding us throughout their overarching research agenda to learn econometric tools that prove extremely useful for the specific setting at hand and, more generally, for employment by structural economists and other applied researchers. In the second part of the discussion, I suggest additional aspects that could play an important empirical role in the functioning of property insurance markets, namely private information about risk and supply-side issues, and pave the way for possible approaches to introduce them into the authors’ framework.