Gender differences in the stability of trust and risk-taking

Abstract

The individual willingness to trust is compared to the inclination to take lottery risk in six distinct scenarios, controlling the return distributions. Trust responds to changes in the admissible return levels, but exhibits significantly smaller responsiveness to return expectations compared to parallel risk-taking. Paired comparisons suggest that the investors sacrifice 5% of the expected payoff to trust anonymous responders. Trust is more calculated and volatile for males, while appearing relatively stable for females. The results connect with evidence regarding physiological differences between trust and risk-taking, and in addition suggest that trust is more of a distinctive trait for females compared to males. The paper broadly discusses the results and their implications, connecting to diverse streams in the trust literature.

KEYWORDS:

• Trust
• risk
• gender

Acknowledgements

The experiment discussed in the paper was presented at the 2015 Social and Biological Roots of Cooperation and Risk-Taking (SBRCR) workshop in Kiel, Germany; the 2015 European meetings of the Economic Science Association in Heidelberg, Germany; and the 2016 International Economic Science Association meetings in Jerusalem, Israel. We have benefited from comments and conversations with Jason Aimone, Steven Bosworth, Eyal Ert, Amir Levkowitz, Tommaso Reggiani and Eyal Weinstock. We thank two anonymous reviewers and the editor Guido Möllering for constructive feedbacks and suggestions. We thank the research authority at COMAS for funding the experiment.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 As explained in Section 3, belief elicitation is irrelevant in the two conditions with fixed return distributions.

2 In Guiso et al. (2008) the trust coefficient loses significance in the most comprehensive estimations, but it is robustly significant when the estimations are restricted to the subjects with less than high school education. Balloch et al. (2015) also control for financial literacy, showing the literacy and trust mutually affect participation and allocations to stocks.

3 The list of chapters and the full translated questionnaire are provided in the appendix. In brief, chapter 1 presented the binary-return trust games; chapter 2 presented the WS14 gain-domain risk attitude task; chapter 3 elicited the trustor’s return expectations, for the binary-return trust games of chapter 1; chapter 4 presented the WS14 loss-domain risk attitude task; chapter 5 presented the lottery allocation tasks; chapter 6 collected socio-demographic information; and chapter 7 introduced four additional trust game, as explained in 3.5 below.

4 Social distance may also reduce the effect of betrayal aversion (Bohnet & Zeckhauser, 2004) on trust. Indeed (similarly to FD12), our results do not support the hypothesis that betrayal aversion decreases the willingness to trust compared to parallel willingness to take risk.

5 The probabilities to be copied referred to LRL (HRL) in 2 (3) conditions.

6 Interestingly, the subjects in C1 still show close to average investments (mean INV of 50 in the 0.9 games and 28 in the 0.15 games), similarly to the ‘inconsistent trusters’ of Chaudhuri et al. (2003).

7 The difference is smaller when the comparison is run on all six games (105.1 vs. 109; p < 0.01). When the trust game payoffs are calculated assuming the mean return rates R (e.g. (100-INV)+INV·1.11 at 0.9–1.35), the mean trust game payoff falls further to 100.4 (103.6 for all six games) due to the lower-than-expected return at 0.9–1.8.

8 The hypothesis that the LOT differences are equal to the INV differences is rejected at p < 0.01 for the males and for the females. The hypothesis that the differences are equal for males and females is rejected at p = 0.04 for LOT but could not be rejected for INV (p = 0.10).

9 RA_G did not correlate with the E(R) of males or females. Web supplement F contrasts the INV, LOT and E(R) of males and females in each game, controlling for RA_G. Note also that in terms of the price paid for trusting (as defined in Section 5.6), the means are 5.1 for the females compare to 6.0 for the males (p = 0.54).

10 The Purdue university ORSEE recruitment platform, for a random example, announces payments between $15 and $30 for participating in an economic experiment (https://orsee.krannert.purdue.edu/public/faq.php) and similar rates are advertised in diverse other recruitment sites.

Additional information

Notes on contributors

Doron Sonsino

Doron Sonsino holds a PhD in Business Economics from Stanford University and an MSc in Operations Research from Tel-Aviv University. He has joined Cyprus International Institute of Management (CIIM) in January 2022. Doron is an associate editor at the Journal of Behavioral and Experimental Economics. His main current research topic is the financial decision of private investors.

Max Shifrin

Max Shifrin took part in the trust experiments as an MBA with thesis student at the College of Management Academic Studies (COMAS), while pursuing a high-tech career as a software developer and team leader in Varonis.

Eyal Lahav

Eyal Lahav holds a PhD in Economics from Ben-Gurion University and is currently a faculty at the Open University of Israel. His research covers topics in behavioural economics such as time preferences, risk preferences, and nudges.