Experiments to Behaviorally Inform Policy

Paper Session

Saturday, Jan. 7, 2017 2:30 PM – 4:30 PM

Sheraton Grand Chicago, Mississippi
Hosted By: Society for the Advancement of Behavioral Economics
  • Chair: Angela C.M. de Oliveira, University of Massachusetts-Amherst

Attribute Overload, Consumer Finance and Welfare

Jeffry P. Carpenter
,
Middlebury College
Emiliano Huet-Vaughn
,
Middlebury College
Julian Jamison
,
World Bank
Peter Hans Matthews
,
Middlebury College
Andrea Robbett
,
Middlebury College

Abstract

While economists have generally considered it welfare improving to offer an individual more options to choose among, a large literature has documented the phenomenon of “choice overload,” where provision of additional choices to individuals can actually lead to suboptimal choices and welfare loss. We investigate a related but distinct phenomenon termed “attribute overload” where increasing the number of characteristics, or attributes, per choice (while keeping the size of the choice set constant) may in and of itself lead consumers to make
welfare diminishing choices. In a two-stage experiment (the stages separated by seven weeks), we first gather time and risk preference information that is used to structurally estimate the individual-specific parameters of a CRRA utility function. For each participant the observed utility function is then used to inform a choice experiment in stage two over typical consumer finance payment card products. Based on their stage two choices we can measure simple dominance violations and measure the welfare implications of choosing the “wrong” card. Participants are randomized into two information architecture treatments that vary the size of the attribute set in otherwise similar choices in order to determine whether dominance violations can be decreased, and, therefore, welfare increased by adopting a simple disclosure policy. Assignment to the few-attribute treatment significantly reduces the chance of choosing dominated credit products, suggesting that indeed consumers experience attribute overload when facing multi-attribute choices.

Promises Renegotiated

Stephen Leider
,
University of Michigan
Erin Krupka
,
University of Michigan

Abstract

Motivated by the subprime mortgage crisis of 2008, we use a trust game to test the impact of promises on behavior in the presence of uncertainty about future states (where the source of the uncertainty stems from acts of nature rather than strategic individuals). Second, we test whether promise-keeping norms affect subsequent decisions about whether to renegotiate the terms of the initial agreement once the uncertainty is resolved. Anecdotal evidence from interviews with homeowners of underwater mortgages, suggests that a norm to honor the promise of repayment and an unease with requesting to renegotiate terms with the bank may have the surprising and unintended effect of inhibiting renegotiation utilization even when government or bank policies would enable such renegotiations. We find that second movers frequently make promises, and that these promises affect their subsequent decisions. We demonstrate that a utility model in which actors have preferences over outcomes as well as social norms is able to account for the pronounced effect of promises on behavior, explain why a second mover would make the promise in the first place and why a first mover would be willing to enter the game. For policies written against the backdrop of strong norms, we address implications and guidelines.

(Im)patience by Proxy: Making Intertemporal Decisions for Others

Angela C.M. de Oliveira
,
University of Massachusetts-Amherst
Sarah Jacobson
,
Williams College

Abstract

Decisions with consequences that play out over time are ubiquitous in business, policy, and family relations, and frequently the agent making these decisions is distinct from those who bear the consequences. We use a lab experiment to examine whether individuals make different intertemporal decisions for others of varying social distance than for themselves. Subjects make a series of intertemporal work time allocation decisions for themselves and for another individual, either a friend or a stranger. We find that people who receive no information about their recipient’s preferences choose more impatiently (moving a disutility cost into the future) for others than for themselves. In other words, a decision made for you by proxy is more impatient than a decision you would make for yourself and thus is probably suboptimal. This result contrasts with the literature, and this may be because, as we find from a separate survey, people perceive procrastination of tasks as qualitatively different from other discounting decisions. Survey evidence suggests that individuals believe that they are more patient than the other subjects are, suggesting that these too-impatient decisions are made for others out of benevolence with a mistaken belief. Further, when the decision-maker is given information about the recipient’s own-stated preferences, this bias of excessively impatient decisions, particularly when the recipient is a friend. Taken together, our results imply that given limited information, proxy decision-makers choose more impatiently (pushing more costs into the future) than agents would prefer, but information mitigates this error.

Dynamic Inconsistency in Food Choice: Experimental Evidence from a Food Desert

Sally Sadoff
,
University of California-San Diego
Anya Samek
,
University of Southern California
Charles Sprenger
,
University of California-San Diego

Abstract

Despite the great deal of research on dynamic inconsistency in time preferences, few studies have ventured into investigating the question in a natural context. To address this gap, we conduct a natural field experiment with over 200 customers at a grocery store to investigate dynamic inconsistency and the demand for commitment in food choice. Over a 3 week time period, subjects are invited to allocate and re-allocate food items received as part of a grocery delivery program. We observe substantial dynamic inconsistency in our experiment, as well as a demand for commitment among a non-negligible number of subjects. Interestingly, individuals who demand commitment are more likely to be dynamically consistent in their prior behavior. For academics, our work provides direct evidence of dynamic inconsistency in consumption choices in the field and points towards potential extensions to models of temptation. For policy-makers, our findings provide insights on innovations to alter food choices.

Using Preference Estimates to Customize Incentives: An Application to Polio Vaccination Drives in Pakistan

James Andreoni
,
University of California-San Diego
Michael Callen
,
Harvard University
Karrar Hussain
,
University of Southern California
Muhammad Yasir Khan
,
University of California-Berkeley
Charles Sprenger
,
University of California-San Diego

Abstract

We use structural estimates of time preferences to customize incentives for a sample of polio vaccinators during a series of door-to-door immunization drives in Pakistan. Our investigation proceeds in three stages. First, we measure time preferences using intertemporal allocations of vaccinations. Second, we derive the mapping between these structural estimates and individually optimal incentives given a specific policy objective. Third, we experimentally evaluate the effect of matching contract terms to individual discounting patterns in a subsequent experiment with the same vaccinators. This exercise provides a test of the specific point predictions given by structural estimates of time preference. We document present bias among vaccinators and find that tailored contracts achieve the intended policy objective of smoothing intertemporal allocations of effort. The benefits of customized incentives, in terms of achieving the policy objective, are largest for vaccinators allocating when present bias is relevant to the decision.
JEL Classifications
  • C9 - Design of Experiments
  • I3 - Welfare, Well-Being, and Poverty