«Abstract We conduct a laboratory experiment which elicits subjects’ beliefs about the likelihood that they will redeem a mail-in form. By comparing ...»
Everyone Believes in Redemption: Overoptimism and Nudges
Robert Letzler Joshua Taso ú
Federal Trade Commission Claremont Graduate University
May 3, 2012
We conduct a laboratory experiment which elicits subjects’ beliefs about the likelihood that
they will redeem a mail-in form. By comparing subjects’ expected redemption rates to actual
redemption rates we ﬁnd that subjects are overoptimistic about their likelihood of redemption and thus “leave money on the table.” Moreover, we ﬁnd that overoptimism is increasing with the belief in redemption, suggesting that the consumers who are most likely to select an option requiring future action make the largest errors. We then test the impact of three “nudges” on overoptimism: (1) informing subjects about a previous cohort’s redemption rates, (2) remind- ing subjects about the redemption deadline, and (3) reducing transaction costs. Testing the interventions helps to both uncover the mechanisms of overoptimism and provides preliminary evidence for potential policy. The third nudge was the only treatment that had any detectable e ect, and it reduced overoptimism by approximately one half. It reduced overoptimism by in- creasing redemption rates but not by decreasing people’s mean belief. We ﬁnd that redemption is sensitive to the payo and cost of redemption but beliefs are almost constant. This suggests that weak cost-salience is the mechanism for overoptimism.
The views expressed in this paper are those of the authors and not necessarily those of the Federal Trade ú Commission or any individual commissioner. Letzler did not use FTC time or resources to help implement and run the experiment, analyze the data, or write up the results. We would like to thank: Peter Fishman, Matthew Rabin, Dan Acland, Botond K szegi, Colin Camerer, Stephanie Wang, Matthew R. Levy, Eric Helland, seminar participants at Berkeley’s Psychology and Economics Non-Lunch, Colin Camerer’s lab meeting, WEAI San Diego 2011, ESA International Conference Chicago 2011, ESA Tuscon Conference 2011, the UC Berkeley Goldman School of Public Policy, Claremont Graduate University Behavioral Economics and Institutions Seminar, the UC Riverside Theory Seminar, and the Southern California Conference in Applied Microeconomics. We thank Masyita Crystallin, Jason Henshall, Peiran Jiao, and Yanyan Yang for outstanding research assistance. We thank Oliver Ortlieb for outstanding programming and website administration. Taso gratefully acknowledges the ﬁnancial support of the Russell Sage Foundation through Grant No. 98-11-01. All errors are evidence of our overoptimism.
1 1 Introduction Consumers often need to accurately predict their future behavior in order to make reasonable decisions today. For example, the beneﬁt of purchasing a product with a mail-in rebate depends on the likelihood that one redeems the rebate. The beneﬁt of buying a good that qualiﬁes for a tax credit is conditional on whether the buyer saves the receipt and claims the credit on her tax returns, and the beneﬁt of putting debt on one’s credit card is dependent on the likelihood that one will pay o one’s balance at the end of the month.
However, there is growing evidence that people are overoptimistic about the likelihood that they will take a future costly action. DellaVigna and Malmendier (2006) ﬁnd that people oversubscribe to monthly gym membership, not exercising frequently enough to justify the monthly fees. In their study, 80% of consumers in the ﬁrst six months of their monthly membership would have saved money if they instead chose a pay-per-visit plan. Ericson (2011) demonstrates that people overestimate the likelihood that they will remember to send an email six months in the future. Giné et al. (2010) found that 66% of Filipino smokers who signed commitment contracts, in which smokers commit to give away money if they smoke, lost their money when they did not pass their nicotine test six months later. Furthermore there is quite robust evidence that people are overconﬁdent in their abilities and overoptimistic about important life events that will occur to them (Weinstein, 1980; Camerer and Lovallo, 1999).
In this paper we study the situation in which the net beneﬁt of an alternative is contingent on whether one engages in a future costly action. We hypothesize that consumers overestimate the likelihood of engaging in that future costly action and thereby choose the associated alternative too frequently. In the experiment, we use a mail-in form as the costly future task because it is relevant to important economic decisions including redeeming mail-in rebates, completing forms to claim a tax credit, and rolling over a balance onto a lower interest rate credit card.
Methodologically, a mail-in form has several advantages as a costly task. Subjects are familiar with this task and have likely had experience completing and mailing forms, so their beliefs should be well informed. The monetary payo is the only rewarding aspect of the task, which allows for identiﬁcation of overoptimism.
Mail-in rebates are a widespread marketing technique for consumer products in the United States. It is estimated that Americans buy goods that o er $4–10 billion of rebate opportunities every year and redeem $3 billion (Edwards, 2007). The economic environment is simple. The consumer, at the point of purchase, pays the pre-rebate price and has the opportunity to receive a rebate conditional on successful redemption. Redemption typically requires the consumer to include the original receipt, a cut out of a UPC barcode from the package, and to complete and mail in a form indicating the consumer’s name, address, and other contact information. A typical time for processing is six to eight weeks at which point the consumer receives a check or a debit card for the advertised amount. The structure of this marketing institution is a simple exemplar of the economic environment under question: the beneﬁt of a purchase with an associated mail-in rebate (the action-contingent alternative) is contingent on whether one successfully redeems the mail-in rebate (the costly future task). Redemption rates are quite low. Silk and Janiszewski (2009) interview promotion managers who report that, “rebates of $10 to $20 on a $100 software product range between 10% and 30%, and that redemption rates 2 on consumer electronics average approximately 40%.” The standard model implies consumers rationally predict these low redemption rates. An alternative hypothesis is that consumers are overoptimistic.
Claiming income tax credits has a very similar structure and is of obvious importance. The citizen can choose to purchase a good that has been granted a tax credit (the alternative). The beneﬁt of the tax credit is contingent on saving the receipt and claiming the credit in one’s tax returns (the costly future task). Filing one’s taxes can be vastly more complicated than ﬁlling out a mail-in rebate form. The stakes are higher than most mail-in rebates and taxpayers face signiﬁcant time or monetary costs. These factors may lead taxpayers to be overoptimistic that they will claim tax credits. In a related vein,Pitt and Slemrod (1989) estimate that 679,300 tax payers did not itemize in 1982 due to the costs of itemization. Whether taxpayers were overoptimistic about their itemization is unknown.
There are also situations in which the submission of a mail-in form or a similar future action is required to avoid an undesirable outcome. Many ﬁnancial institutions o er credit cards with introductory (“teaser”) rates. Introductory rates are lower than the standard rate but expire after an initial period of service. For example, the Citi Simplicity• Card o ers an introductory R rate of 0% on transfers and purchases for the ﬁrst eighteen months of service but the rate then reverts to 12.99–20.99% APR.1 Such o ers often allow for the low-cost transfer of a balance from another credit card. Understanding consumers’ ability to manage credit card interest rates is important since people who do not pay their entire bill carried an average balance of $7,200 in 2004 (Shui and Ausubel, 2005). Customers with a pre-existing balance on a credit card with a moderate interest rate have the incentive to transfer their balance to a credit card with a lower introductory rate and higher standard rate (the action-contingent alternative). When the introductory period ends, the consumer can transfer the existing balance to another card with a lower interest rate (the costly future task). The net beneﬁt of selecting the card with a low, introductory rate is contingent on switching to another card before the end of the introductory period.In a randomized ﬁeld experiment Shui and Ausubel (2005) ﬁnd that 60% of consumers do not transfer their balance after the introductory period has elapsed despite the fact that average balances were between $2,000 and $2,500.
The traditional economic explanation is that consumers do not submit forms because the cost of submission exceeds the beneﬁt. This hypothesis is calibrationally suspect because it implies that redemption is very costly. Consider mail-in rebates as an example. Our subjects report that redemption takes about ten minutes. Thus, failure to redeem rebates of even modest size would imply a huge shadow wage, a very low discount factor, or a very large psychic cost to submitting a form. As a simple example, consider a person does not redeem a rebate form worth $40 (after the price of a stamp and envelope) and redemption takes ten minutes. As is often done
in the literature to infer time preferences from choices over money we make three assumptions:
ﬁrst assume that money is fully consumed in the period it is received; second, consumption is approximately constant over time; and third, utility is approximately linear in utility for small sums. Then this implies that the person’s shadow wage is greater than $180/hour, the equivalent of a $480,000 annual salary. A typical American near the median who earns $40,000 annually 1 O er observed on the internet on March 6, 2012. https://creditcards.citi.com/credit-cards/citi-simplicity/ 3 and thus $20/hour should spend 10 minutes submitting a mail-in rebate as long as the payo exceeds $3.33 (after stamp and envelope costs).
A second hypothesis is that people are overoptimistic because they are naïve about their present-biased preferences (O’Donoghue and Rabin, 1999). Naïve present-biased decision makers discount all future periods relative to the present, but do not anticipate they will continue to do so in future periods. Thus, they discount the reward and the redemption cost at the point of purchase. But when they decide whether to redeem, they discount the future reward but not the immediate redemption cost. As a consequence, they procrastinate, putting redemption o into the future, until the deadline arrives at which point they redeem at a lower rate than they originally expected. This hypothesis is also calibrationally suspect because such behavior would imply severe present-bias. Continuing the earlier example, a person who does not redeem a rebate form worth $40 and discounts the future quasi-hyperbolically (Laibson, 1997) with — = 1 and ” = 1 implies that her annual income exceeds $240,000. In order for a consumer who 2 earns $40,000 annually to abstain from redemption on the day of the deadline, she would need 1 a — Æ 12 ¥ 0.0833. DellaVigna (2009) reports that estimates of — fall between 0.40 to 0.89.
A third hypothesis is that consumers overestimate their ability to remember to redeem. Ericson (2011) ﬁnds that people are overconﬁdent on a memory task. People are often overconﬁdent about their performance on easy tasks (Moore and Healy, 2008).
A fourth hypothesis is that consumers systematically underestimate the cost of redemption.
Despite full disclosure of the costs we hypothesize that subjects su er from weak cost-salience.
Subjects may be aware of the costs that they face but they do not account for them in their decision at the point of purchase.
All of these hypotheses, with the exception of rational expectation, are mutually consistent, and imply that the consumer overestimates the probability she will redeem.
It is important to identify and understand overoptimism because it can lead to market failure.Consider a simple example in which a consumer is risk neutral and perceives his probability of redemption as higher than his actual redemption rate by z œ (0, 1]. We simplify the notation by ignoring transaction costs. Deﬁne r as the value of redemption, and deﬁne p as the price less the expected beneﬁt of redemption at the decision maker’s actual redemption rate. The consumer’s misperception will shift his demand curve up from x(p) to x(p ≠ z ú r). Consequently the decision maker will overconsume the good, creating the deadweight loss triangle illustrated in Figure 1. Thus identifying overoptimism is valuable not only for understanding human decision making, but also for improving market e ciency. Moreover, overoptimism incentivizes marketers to structure transaction costs to prompt overoptimism and transfer surplus from consumers to ﬁrms.
Overoptimism may persist in the marketplace due to shrouding of the costly future action.
Gabaix and Laibson (2006) develop a model in which they assume there are accurate consumers who understand the total price of the good and myopic consumers who underestimate the price by neglecting add-on costs. One can conceptualize mail-in rebates and introductory-rate credit cards in a similar manner. Accurate consumers know the price they pay in expectation but myopic consumers neglect the redemption costs and risks, including procrastination and forgetting, and hence underestimate the price they pay. The authors demonstrate that shrouding
Figure 1: The gap between consumer-perceived price and actual price creates deadweight loss can theoretically exist in a competitive market because no one has an incentive to educate the myopes. Accurate consumers receive a cross-subsidy from the myopes. If education increases myopes’ redemption rates enough, any ﬁrm that successfully educates a myope will induce the myope to go to the ﬁrm o ering the rebate or the introductory rate in order to receive the cross-subsidy.