Refereed journal articles
Abstract: We develop a model of strategic communication between an uninformed receiver and a partially informed sender who is guilt-averse toward the receiver. The sender's cost of sending a particular message is endogenous, depending on the receiver's beliefs induced by this message rather than on its exogenous formulation. Such preferences lead to the endogenous emergence of evasive communication in that the sender types who prefer not to reveal their information to the receiver pool with uninformed types rather than with types observing different information. As a result, the receiver may prefer an equilibrium with a smaller amount of messages used on the equilibrium path. Besides, dealing with an ex ante less informed sender can be beneficial to the receiver, while the sender himself may want to commit to a smaller ex ante likelihood of being informed.
Disguising Lies - Image Concerns and Partial Lying in Cheating Games (with Dirk Sliwka), American Economic Journal: Microeconomics, Volume 11 (4), November 2019, 79-110.
Abstract: We study equilibrium reporting behavior in Fischbacher and Föllmi-Heusi (2013) type cheating games when agents have a fixed cost of lying and image concerns not to be perceived as a liar. We show that equilibria naturally arise in which agents with low costs of lying randomize among a set of the highest potential reports. Such equilibria induce a distribution of reports in line with observed experimental patterns. We also find that higher image concerns lead to an increase in the range of reported lies while the effect of the fixed cost of lying is the opposite.
Sharing Guilt: How Better Access to Information May Backfire (with Roman Inderst and Axel Ockenfels), Management Science, Volume 65, July 2019, 2947-3448.
Abstract: We study strategic communication between a customer and an advisor who is privately informed about the most suitable choice for the customer but whose preferences are misaligned with the customer’s preferences. The advisor sends a message to the customer who, in turn, can secure herself from bad advice by acquiring costly information on her own. In our experiments, we find that making the customer’s information acquisition less costly leads to less prosocial behavior of the advisor. This can be explained by a model of shared guilt, which predicts a shift in causal attribution of guilt from the advisor to the customer if the latter could have avoided her ex post disappointment. We conclude that providing better access to information through, e.g., consumer protection regulation or digital information aggregation and dissemination, may have unintended negative consequences on peoples’ willingness to take responsibility for each other.
Evasive Lying in Strategic Communication (with Bettina Rockenbach and Peter Werner), Journal of Public Economics, Volume 156, December 2017, 59-72.
Abstract: Information asymmetries in economic transactions are omnipresent and a regular source of fraudulent behavior. In a theoretical and an experimental analysis of a sender-receiver game we investigate whether sanctions for lying induce more truth-telling. The novel aspect in our model is that senders may not only choose between truth-telling and (explicit) lying, but may also engage in evasive lying by credibly pretending not to know. While we find that sanctions promote truth-telling when senders cannot engage in evasive lying, this is no longer true when evasive lying is possible. Then, explicit lying is largely substituted by evasive lying, which completely eliminates the otherwise positive effect of sanctions on the rate of truth-telling. As outlined in our model, the necessary prerequisite for such an ‘erosion’ effect is that evasive lying is perceived as sufficiently less psychologically costly than direct lying. Evidence from our experimental data and a survey conducted with additional participants indicate that the shift towards evasion can indeed be attributed to lower psychological costs. Overall, our results clearly demonstrate the limitations of sanctioning lying to counteract the exploitation of informational asymmetries and may explain the empirical evidence from the finance industry that sanctions for financial misconduct eventually appear to be not very effective.
Testing Guilt Aversion with an Exogenous Shift in Beliefs, Games and Economic Behavior, Volume 97, May 2016, 110-119.
Abstract: We conduct a laboratory experiment to test whether subjects tend to meet the expectations of others (the guilt aversion hypothesis). The specificity of our approach is that second-order beliefs are manipulated exogenously just by changing the parameters of the experimental game. In particular, we consider a simple communication game where the sender is perfectly informed about his material benefit from lying to the receiver. At the same time, the receiver knows only the ex-ante distribution of the sender's material incentives. By changing this distribution between the experimental treatments, we achieve an exogenous variation in the receiver's payoff expectations (and hence in the corresponding sender's second-order beliefs) while keeping the sender's actual material incentives fixed. The results show that the rate of lying is significantly lower when the receiver is supposed to have higher payoff expectations, however only in the case when the monetary incentives for lying are fixed at a moderate level.
Surprising Gifts - Theory and Laboratory Evidence (with Axel Ockenfels and Peter Werner), Journal of Economic Theory, Volume 159 (A), September 2015, 163-208.
Abstract: People do not only feel guilt from not living up to others' expectations (Battigalli and Dufwenberg (2007)), but may also like to exceed them. We propose a model that generalizes the guilt aversion model to capture the possibility of positive surprises when making gifts. A model extension allows decision makers to care about others' attribution of intentions behind surprises. We test the model in a series of dictator game experiments. We find a strong causal effect of recipients' expectations on dictators' transfers. Moreover, in line with our model, the correlation between transfers and expectations can be both positive and negative, obscuring the effect in the aggregate. Finally, we provide evidence that dictators care about what recipients know about the intentions behind surprises.
Chapters in books
Exploitative Strategies: Consequences for Individual Behavior, Social Structure, and Design of Institutions (with Ben Greiner, Andrew J. King, Michael Kosfeld, Sasha R. X. Dall, Tatsuya Kameda, Wolfgang Leininger, Claus Wedekind, and Bruce Winterhalder). In: L-A. Giraldeau, P. Heeb, and M. Kosfeld (eds.): Investors and Exploiters in Ecology and Economics: Principles and Applications. Strüngmann Forum Reports, vol. 21 (J. Lupp, series editor), MIT Press, 2017, 205-214.
Abstract: Abundant experimental and field evidence suggests that people tend to dislike open disagreement. We propose a formalization of perceived disagreement and study the implications of perceived disagreement aversion in disclosure games involving agents with different priors. Across a variety of settings, the ideal conditions for disclosure involve identical prior variances and differing prior means. When equilibrium disclosure is partial, it is biased towards evidence that is congruent with the most confident agent's prior bias. Perceived disagreement aversion leads to assortative matching in prior beliefs that provides a theoretical basis for echo chambers. Equilibria may feature higher average perceived or actual disagreement than a hypothetical full disclosure scenario. Perceived disagreement aversion arises endogenously within simple games of delegation and competitive authority assignment.
Abstract: It has been argued that guilt aversion (the desire to meet others' expectations) and the social norm compliance (the desire to act similarly to other individuals in the same situation) are important drivers of human behavior. But as we show in a theoretical model, these two motives are empirically indistinguishable when only one signal (either the expectation of a person affected by the choice or a signal about the descriptive norm) is revealed as each of these signals transmit information on the other benchmark. We then address this problem running an experiment in which signals for both benchmarks are revealed simultaneously. We find that both types of information affect transfers in the dictator game but the descriptive social norm has a stronger effect than the recipient's expectation, and the effect of the recipient's expectation is non-monotonic and decreasing for very high expectations. We provide further evidence for the importance of both benchmarks in a second experiment where we display the recipient's expectation and the expectation of a randomly picked recipient of another dictator.