Schedule - Parallel Session 3 - Risk Attitudes 3

WMG IDL Boardroom - 15:40 - 17:10

Income Inequality and Risk Taking

Ulrich Schmidt; Levent Neyse; Milda Aleknonyte


Standard economic theory assumes that individual risk taking decisions are independent from the social context and therefore also independent from the income distribution. Recent experimental evidence however shows that the income of peers has a systematic impact on observed degrees of risk aversion. In particular, subjects strive for balance in the sense that they take higher risks if this gives them the chance to break even with their peers. The present paper is, to the best of our knowledge, the first systematic analysis of income inequality and risk taking. We perform a real-effort field experiment where inequality is introduced to different wage rates. After the effort phase subjects can invest (part of) their salary into a risky asset. Besides the above mentioned possibility of higher risk taking of low-wage individuals to break even with high-wage individuals, risk taking can be influenced by an income effect consistent with e.g. decreasing absolute risk aversion and a house money effect of high-wage individuals. Our results show that the dominant impact of inequality on risk taking is what can be termed a social house money effect: high-wage individuals take higher risks than low-wage individuals only if they are aware of the inequality in wages.

Ulrich Schmidt

Professor, Kiel Institute for the World Economy

Lottery- and Survey-Based Risk Attitudes Linked Through a Multichoice Elicitation Task

Giuseppe Attanasi; Nikolaos Georgantzìs; Valentina Rotondi; Daria Vigani


In this paper we compare two mutually uncorrelated risk-attitude elicitation tasks. In particular, we test for correlation of the elicited degrees of monetary risk aversion at a within-subject level.
We show that sufficiently similar incentivized mechanisms elicit correlated decisions in terms of monetary risk aversion only if other risk-related attitudes are accounted for.
Furthermore, we ask subjects to self-report their general willingness to take risks. We find evidence of some external validity of the two tasks as predictors of self-reported risk attitudes in general human domains.

Giuseppe Attanasi

Assistant Professor, Strasbourg

Measuring Multivariate Risk Preferences

Gijs van de Kuilen; S. Ebert


It has long been recognized that risk attitudes play a crucial role in economic behavior. While Pratt (1964) has demonstrated the ubiquitous importance of risk aversion, Leland (1968) and Kimball (1990) showed that the higher-order risk attitudes prudence and temperance complement risk aversion in important ways. For example, in the realm of saving behavior, risk aversion drives a preference to smooth consumption over time (“consumption smoothing”), prudence governs how saving behavior changes as future income becomes riskier (“precautionary saving”), and temperance determines how saving behavior is affected by changes in macroeconomic risks such as interest rate risk or unemployment risk. The existing literature on higher-order risk attitudes has only considered risk preferences for a single, monetary, attribute so far. In many important settings, decision makers face risky alternatives with multiple, potentially non-monetary, attributes (Keeney and Raiffa 1993). According to theory, the cross-risk attitudes correlation aversion, cross-prudence and cross-temperance determine how risk preferences over multiple attributes co-vary and interact. This paper reports the results of an experiment designed to measure multivariate risk preferences for different attributes, not only for the second order (risk aversion), but also for higher orders (prudence and temperance), in three important domains (social preferences, time preferences, and preferences for leisure time). This first systematic empirical exploration of multivariate risk preferences provides evidence for assumptions made in economic models on inequality, labor, time preferences, saving, and insurance. We observe correlation seeking and cross-intemperance in a condition involving social preferences, which is in line with models predicting inequality aversion. Results from a condition involving time preferences cast doubt on the separability of utility across time, an assumption often invoked by models of intertemporal decision making.

Gijs van de Kuilen

Associate Professor, Tilburg University

Heterogeneity in Risk Attitudes across Domains: A Bivariate Random Preference Approach

Anna Conte; Werner Guth; Paul Pezanis-Christou


In a series of field experiments, we elicit risk preferences for financial, life-duration, and environmental domains using sequential multiple price-list auctions. We intentionally oversample subjects who frequently engage in activities that increase their mortality risk. Under the assumption that subjects are Rank Dependent Utility maximizers, we estimate the joint distribution of the CRRA and probability weighting coefficients. The model we propose to estimate the data is the Bivariate Random Preference (BRP) model. This is a new estimation approach that enables us to incorporate all available information derived from subjects’ switch points in the lists. We find that the experienced risk takers are less likely than the student control group to overweight small probability, extreme events in their decision making. This is true in all three domains. We find that the tendency of women to be more risk averse in the financial domain than men arises from probability weighting rather than differences in the utility function. Finally, we show that a significant minority of subjects deviate from the type of s-inverse probability weighting typically observed in experiments.

Anna Conte

Senior Lecturer, University of Westminster