Prudence and prevention: an economic laboratory experiment

ABSTRACT

In an economic laboratory experiment, we study the relationship between prudence and prevention in general decision situations. Previous theoretical research on this relationship posits a negative impact of prudence on the optimal level of prevention. Overall, we find both risk-averse and prudent behaviour among our subjects. Moreover, prudent subjects chose significantly less prevention than nonprudent subjects, confirming the theoretical results of one-period models in the literature. Our findings might have implications for health policy if prudence – rather than irrational decision behaviour, as previously assumed – is responsible for low levels of preventive effort.

KEYWORDS:

• Self-protection
• prevention
• prudence
• risk preferences
• laboratory experiment

JEL CLASSIFICATION:

• I10
• C91
• D81

Acknowledgements

We are thankful to Stefan Felder and Timo Heinrich for their valuable comments that improved the article and to Nadja Kairies for help with the experimental sessions.

Disclosure statement

No potential conflict of interest was reported by the authors.

Ethics

Subjects were recruited from registered students of the elfe experimental laboratory study subject pool at the University of Duisburg-Essen. There were no reasonably foreseeable risks or discomforts to participants. Experiences were similar to day-to-day decisions. Subject information, i.e. names, was substituted by study numbers. No names were used throughout the study, i.e. data cannot be linked to individual subjects.

Supplementary

Supplemental data for this article can be accessed here.

Notes

¹ Note that we assume a one-period model. Menegatti (2009) and Hofmann and Peter (2013) show that this discrepancy can vanish or even turn into two-period models.

² Since risk-neutral subjects are indifferent between the alternatives at a probability of exactly 50%, they may not start treatment until they reach the next probability of 52%.

³ To use a neutral setting, we used the term ‘investment’ instead of ‘prevention’/‘self-protection’ in the actual experiment.

⁴ The random payment technique is a well-established method to avoid wealth and averaging effects in repeated decisions. Several studies on this method show that selective payment of one decision does not weaken the monetary incentive in noncomplex choice tasks (e.g. Baltussen et al. 2012; Laury 2006; Cubitt, Starmer, and Sugden 1998; Starmer and Sugden 1991).

⁵ Note that Holt and Laury (2002) simply consider the number of safe choices subjects made rather than excluding multiple switchers from the sample. This is not feasible in our case, as we require a comparison of switching points across tasks one and two. However, including these subjects in our sample and replacing their missing switching points with the mid-point of their switching interval yield very similar results.

⁶ Note that only 8% of the subjects in the Ebert and Wiesen (2011) and 14% of the subjects in the Deck and Schlesinger (2010) study make all choices prudently.

⁷ We used the STATA user-package REOPROB (Frechette 2001a, 2001b) to calculate the random-effects ordered probit model. We also estimated alternative regressions (ordered probit with and without dummy variables for the individual decisions and clustered standard errors and OLS for the aggregated decisions – prevention score – at subject level; see Appendix C (available as supplemental data)), all of which yield very similar results.

Additional information

Funding

Financial support from the German Research Foundation (DFG) [FOR 655] is gratefully acknowledged.