Sam Hirshman Contact Coffee CV Research

Published and Working Papers:

Ownership, Learning, and Beliefs
(with Samuel M. Hartzmark and Alex Imas)
revise and resubmit Quarterly Journal of Economics

Abstract: We study how ownership affects learning and beliefs. Using an experimental asset market, we find that owning a good leads people to over-extrapolate from signals about its underlying value: after seeing positive signals, people become too optimistic, after seeing negative signals, they become too pessimistic. This result holds relative to a Bayesian benchmark and compared to learning about goods they do not own. In fact, learning is less biased and more "correct" about goods that are not owned. We replicate these results in field data, showing that asset owners over-extrapolate nearly twice as much as non-owners from the same signals.

Minimum Payments Alter Debt Repayment Strategies across Multiple Cards
(with Abigail B. Sussman)
revise and resubmit Journal of Marketing

Abstract: US Households currently hold $808 billion of credit card debt and manage repayments across multiple cards. We investigate how minimum payments (i.e., the requirement to allocate some amount of money to each account with a balance) alter consumers’ allocation strategies. We first document in field data that consumers spread debt repayment allocations across more accounts than would be predicted by previously documented strategies. Next, we use a debt management task with an incentive compatible design to investigate how minimum payments alter debt repayment strategies. We find that minimum payments cause consumers to allocate more than the minimum to a greater number of accounts, which we term the dispersion effect of minimum payments. This dispersion effect interferes with the implementation of the optimal strategy. Finally, we provide evidence of an underlying mechanism in which minimum payment requirements lead to increased dispersion by altering the size of the consideration set.

As Wages Increase, Do People Work More or Less? A Wage Frame Effect
(with Luxi Shen)
revise and resubmit Management Science

Abstract: In jobs in which workers have the flexibility to decide how much work to supply, such as in the gig economy, the effect of a wage change on work supply can be hard to predict. A wage increase, for example, offers workers the opportunity to make more money, so they may want to work more, but at the same time, it allows them to enjoy more leisure, so they do not need to work so much. Economic theory alone does not predict which outcome is more likely to occur, and empirical evidence on the short-term effect of wage change on work supply is also mixed. This research provides some psychological insights into this economic problem by showing that the effect of wage change on work supply depends on how the change is framed. Specifically, for a worker who used to work A hours to earn $X, if the wage change is presented as a payment change (“work the same A hours and earn $Y”), then work supply is expected to change in the same direction as the wage change. By contrast, if the wage change is presented as a workload change (“work B hours and earn the same $X”), then work supply is expected to change in the opposite direction of the wage change. This wage frame effect occurs because in multi-attribute decisions, decision makers assign greater weight to attributes that change than to those that remain constant. A series of experiments (total N = 2,599) demonstrates the wage frame effect on both expressed willingness-to-work and actual work performance, and tests the proposed account as well as alternative explanations. Since any wage change has to be communicated with some specific frame, the choice of the frame can have powerful effects. In fact, it is even possible for a wage decrease to elicit the same increase in work supply as a wage increase. This research (a) offers psychological insights into a classic economic problem, (b) documents a novel framing effect for the judgment and decision-making literature, and (c) suggests a nudge idea in incentive designs to managers and policy makers.

Mental Budgeting vs. Relative Thinking
(with Devin G. Pope and Jihong Song)
AEA Papers and Proceedings, May 2018

Abstract: A growing literature uses economic behaviors in field settings to test predictions generated by various psychological models. In some cases, psychological theories make conflicting predictions for the same consumer context. In this paper, we attempt to reconcile two conflicting predictions about upgrading behavior, one made by category budgeting (e.g.,Heath and Soll, 1996, Thaler, 1985)—which suggests people will upgrade less as prices go up—and one made by relative thinking (e.g., Kahneman and Tversky, 1981)—which suggests people will upgrade more as prices go up.

Works in Progress:

Ownership Alters Learning and Beliefs about Products
(with Abigail B. Sussman, Samuel M. Hartzmark and Alex Imas)

Abstract: Consumers often make repeated purchase decisions. We investigate how owning a product alters learning about the owned goods as well as other products in the same category or made by the same brand. Across both experimentally controlled and naturally occurring purchase behavior, we show that owners (vs. non-owners) update their beliefs more extremely in response to new information. This exacerbates the effects of motivated reasoning in the positive domain, but attenuates it in the negative domain where it can even become more pessimistic than non-owners. We propose that differential attention to information drives these patterns of updating. Owners also report higher likelihood than non-owners of purchasing another product from a brand with a better product in our assessment task, but are no more likely to do so for for a brand with a worse product. Our results suggest that ownership alters the way consumers incorporate new information about related products and brands into their beliefs.

When is Now?: Empirical Tests of Quasi-Hyperbolic Discounting
(with Daniel M. Bartels)

Abstract: Quasi-hyperbolic discounting (Laibson, 1997) is often used to model excessive impatience. Our project aims to better understand (i) how quasi-hyperbolic people’s preferences are and (ii) if they are, when does “now” end and the future start? We find “now" periods including tomorrow and a week outperform a "now" period only including today, though all outperform an exponential model. In addition, participants’ best fitting "now” periods are sensitive to magnitudes. Even participants’ responses to questions with a year delay are best fit by a two-parameter model, a pattern inconsistent with the predictions of the quasi-hyperbolic model.

Metrics for detecting the ‘hot hand’ in basketball using waiting time properties of binomial distributions (with Connor Dowd and Nick Polson)