Choosing for others

ABSTRACT

Experiments conducted in the US and France were used to study how individuals make trade-offs between health and taste for themselves and others. When someone receives a choice made for them that differs from their preference, they experience a welfare loss; at least in the short-term. We measure the empirical magnitude of this loss, and suggest it play a role in assessing the desirability of paternalistic policies motivated by behavioural economics. We show that the welfare loss changes with the provision of new information and the impact of this information differs for the two countries.

KEYWORDS:

• Behavioural economics
• experiment
• health
• paternalism
• value of information

JEL CLASSIFICATION:

• C91
• I10
• Q18

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ A number of field experiments have been conducted on paternalistic policies in which adults or regulators make choices on behalf of (or utilize nudges to influence) children (e.g., Belot and James 2011; List and Samek 2015; Maimaran and Fishbach 2014), but few studies have looked at adults choosing on behalf of other adults.

² In general, any person who has assumed the power to make choices on behalf others is a paternalist regardless of what type of choice is made. In our context, we, refer to someone who chooses healthier options for another person than they do for themselves a ‘paternalist’. We choose this moniker because a ‘real world’ paternalist who has taken on the role of making food choices for others (as opposed to our participants who were randomly assigned the role) is likely to do so under the premise that they can improve the other person’s long-term well-being by making healthier choices than the person would make for themselves. Of course, it could be the case that the chooser truly believes the other person will be happier (either in the short or long run) with healthier options than the individual personally chooses, so that the decision is a benevolent one. Implicit in all paternalistic decisions is some belief about other people’s preferences. That said, we admit that the initial decisions made in our experiment before information may not have been made thinking of the long-term health consequences of the recipient of the choices, as would likely be the case for a ‘real world’ paternalist.

³ The data utilized in this article are part of a larger experimental design discussed by Lusk, Marette, and Norwood (2014). This article focuses specifically on the choices of the people assigned to decide for others, and draws out the welfare implications of those choices.

⁴ We were unable to find a cookie brand that was well known in both locations, and as such, we used different brands that were nonetheless similar in familiarity and nutritional content in each locations. In 2011, Oreos were not carried in many grocery stores, had been reportedly bought by only 10% of French households, and only had a market share of 1.1% (Lentschner 2011). If we wanted to use a cookie that was comparable in terms of popularity and brand awareness, it would not have been appropriate to use Oreo in both countries (because, for example, we would have confounded comparisons with novelty of the unhealthy item). That we used different brands across locations might cause a difference in the mean number of cookies preferred within a location; however, any brand effect should cancel out when taking differences in the choices for oneself versus choices for others, which is the focus of this analysis.

⁵ There is an equally likely chance any of the 15 choices is selected as binding, and as such the number of cookies participant i would expect to receive is equal to $E[Cookies{]}_i=(1/15){\sum }_{j=0}^{14}j\times y_{ij}$. Therefore, $Prob[Cookies{]}_i=(15/105)E[Cookies{]}_i$.

⁶ Weight locus-of-control, measuring beliefs about their ability to control weight, was elicited using the scale developed by Stotland and Zuroff (1990). A higher score implies higher perceived control over one’s weight.

⁷ For a same product, Roosen, Marette, and Blanchemanche (2010) compared the WTP elicited with a Becker-DeGroot-Marschak (BDM) mechanism and the WTP defined in Equation 3 with the observed market price ${\bar{p}}_A$ of the good used as the endowment. They showed that differences in WTP disappear when considering only engaged bidders, namely the ones with nonzero bids in the BDM or the ones choosing once the products like cookies in Fig. 1 with ${\tilde{q}}_{C,i}^T\le 14.$

⁸ Appendix 1 shows the absence of statistical difference between the choosers’ WTP for themselves (defined by $WT{P}_{C,i}^1$) and the receivers’ WTP elicited prior to information.

⁹ Appendix 2 shows that there is a symmetric way of defining the welfare loss of paternalism by starting from the recipient surplus, which leads to similar results.

Additional information

Funding

This work was supported by the French National Institute for Agricultural Research, Oklahoma Agricultural Experiment Station and Willard Sparks Endowed Chair.