CLTS Working Paper on how related are risk preferences and time preferences

The paper "How related are risk preferences and time preferences?" written by Stein T. Holden and Mesfin Tilahun is now published as a CLTS Working Paper.

CLTS Working Paper on how related are risk preferences and time preferences

CLTS WP 4/19

Photo
CLTS WP 4/19
The paper is published as a CLTS Working Paper and can be downloaded here.

Abstract
Risk and time preferences are fundamentally important for financial decisions. We study such preferences for business group members based on field experiments in Ethiopia. The relationship between risk preferences and time preferences has been subject to intensive research and debate among behavioral and experimental economists lately. We aim to contribute to this literature based on a Double Multiple Choice List approach used in an incentivized field experiment. First, we provide strong evidence of diminishing impatience in our data that cannot be explained by present bias.

Next, we assess whether measures of diminishing impatience can be associated with measures of risk aversion and probabilistic sensitivity. We also assess whether measurement error in the risk experiment could be the culprit and create spurious correlations between measures of risk aversion and discount rate elasticities with respect to time horizon. Using a random coefficient model, we find strong evidence of diminishing impatience and large and highly significant individual variation in discount rate elasticities with respect to time horizon. We find only weak support for the idea that diminishing impatience is explained by probabilistic sensitivity due to uncertainty about delayed payouts in the discount rate experiments. 

Risk aversion and optimism/pessimism were unrelated to model noise. More pessimistic and more risk averse respondents had more hyperbolic time preferences and these results were not sensitive to measurement error. Surprisingly, more consistent responses in the risk experiments (lower measurement error) were found for respondents with more hyperbolic time preferences and respondents with higher probabilistic insensitivity. 
Published 24. May 2019 - 13:58 - Updated 24. May 2019 - 13:58