Experimental Methods, Game theory, the Economics of Social Psychology, and Network

‘ Secure Survey Design in Organizations: Theory and Experiments,’ Chassang, S., 2019 (with C. Zehnder).

We study the impact of secure survey designs ensuring plausible deniability on information transmission in organizations. We are interested in settings in which fear of retaliation makes potential informants reluctant to reveal the truth. Theory predicts that: (i) popular randomizedresponse designs fail to induce informative reports, because they are strategically equivalent to non-secure direct-elicitation designs; (ii) hard-garbling designs that exogenously distort survey responses improve information transmission; and (iii) unbiased estimates of the impact of survey design on information transmission can be obtained in equilibrium. Laboratory experiments qualify these predictions. While hard-garbling does improve information transmission over direct-elicitation, other predictions fail: randomized response performs much better than expected; and false accusations lead to a small but persistent bias in treatment effect estimates. We show that these deviations from equilibrium can be accounted for in an off-the-shelf model of boundedly rational play, and that this model of play makes specific predictions over the bias of treatment effect estimators. Additional experiments reveal that play converges to equilibrium if players can (socially) learn from cross-sectional data. These results suggest that randomized response cannot be used systematically in organizational settings, whereas hard garbling improves survey quality even under long-run equilibrium conditions.

‘ Long-Run effects of lottery wealth on Psychological well-being,’ Cesarini, D., 2018 (with E. Lindqvist, and R. Ostling).

We surveyed a large sample of Swedish lottery players about their psychological well-being and analyzed the data following pre-registered procedures. Relative to matched controls, large-prize winners experience sustained increases in overall life satisfaction that persist for over a decade and show no evidence of dissipating with time. The estimated treatment effects on happiness and mental health are significantly smaller, suggesting that wealth has greater long-run effects on evaluative measures of well-being than on affective ones. Follow-up analyses of domain-specific aspects of life satisfaction clearly implicate ˝financial life satisfaction as an important mediator for the long-run increase in overall life satisfaction.

‘Backward Discounting,’ Ray, D., 2018 (with N. Vellodi, and R. Wang).

We study a model in which lifetime individual utility is derived from both present and past consumption streams. Each of these streams is discounted, the former forward in the usual way, the latter backward. We presume that an individual at date t
evaluates consumption programs according to some weighted average of her own felicity (as perceived at date t) and that of “future selves” at dates greater than t. This simple formulation allows agents to partially anticipate future regret in current decisions, and generates a set of novel testable implications in line with empirical evidence. The model is used to capture the notion of parental influence and investigate its impact on equilibrium savings. The paper also examines other applications of “backward discounting.”.

‘Strategic Network Formation with Many Agents,’ Menzel, K., 2017.

We consider a random utility model of strategic network formation, where we derive a tractable approximation to the distribution of network links using many-player asymptotics. Our framework assumes that agents have heterogeneous tastes over links, and allows for anonymous and non-anonymous interaction effects among links. The observed network is assumed to be pairwise stable, and we impose no restrictions regarding selection among multiple stable outcomes. Our main results concern convergence of the link frequency distribution from finite pairwise stable networks to the (many-player) limiting distribution. The set of possible limiting distributions is shown to have a fairly simple form and is characterized through aggregate equilibrium conditions, which may permit multiple solutions. We analyze identification of link preferences and propose a method for estimation of preference parameters. We also derive an analytical expression for agents’ welfare (expected surplus) from the structure of the network.

‘Reputation and information Design,’ Mathevet, L., D. Pearce, and E. Staccheti, 2019.

Can the commitment assumption underlying information design be replaced by reputational enforcement? A long-run sender periodically makes cheap talk announcements to the public, anticipating how it may affect his reputation as a trustworthy type. As he becomes perfectly patient, his payoff converges to his information-design value in all equilibria. By contrast, in the standard repeated game, he typically underperforms compared to information design. In a specialized environment, we show that convergence also happens in behavior: players’ equilibrium behavior coincides asymptotically with the information-design solution. We also examine welfare properties numerically by adapting strategic dynamic programming to reputational games..

‘Simplifying Bayesian Persuasion,’ Mathevet, L. (with E. Lipnowski), 2017.

In Bayesian Persuasion (Kamenica and Gentzkow (2011)), the sender’s optimal value is characterized by a concave envelope. Since concavification of a general function is notoriously difficult, we propose a method to reduce the problem, using the underlying economic structure of the indirect expected utility. The key observation is that one can find, using the receiver’s preferences alone, a small set of posterior beliefs on which some optimal information policy must be supported. This simplifies, sometimes dramatically, the search for optimal information.

‘Rational Inattention and Psychometrics,’ Caplin, A. (with D. Csaba, J Leahy), 2018.

We incorporate rational inattention (RI) theory and psychometric methods into an interdisciplinary approach to learning behavior and choice. Our “Weber curve”, which summarizes data from a novel incentivized experiment, allows costs of attention and consumer welfare to be recovered. The recovery result exploits a precise parallel between RI theory and the theory of competitive supply. This yields an elementary microeconomic toolbox for understanding attention costs. We also develop revealed preference methods that allow us to make rich inference about costs from limited psychometric data.

‘Present-Bias, Procrastination and Deadlines in a Field Experiment,’ Bisin, A. (with K. Hyndman), 2018.

We study procrastination in the context of a field experiment involving students who must exert costly effort to complete certain tasks by a fixed deadline. Descriptively, we document a strong demand for commitment, in the form of self-imposed deadlines, which appear to be associated with students’ self-reported psychological characteristics and cost of time. We structurally estimate students’ present-bias and cost of time by fitting the experimental data to a stylized stopping time choice model. We find that present-bias is relatively widespread but that having multiple repeated tasks appears to activate effective internal self-control mechanisms. Finally, we also document an important form of partial naivete on the part of students in anticipating their ability to self-control when setting deadlines.

‘A Theory of Experiments,’ Chassang, S. (with A. Banerjee, S. Montero, and E. Snowberg), 2018.

We try to better understand why, and when experimenters randomize. Formulating experiment-design as a decision problem lets us gauge the pros and cons of other experimental practices, such as rerandomization.

‘Rolodex Game in Networks,’ G. Menzio (with B. Brugermann, P. Gautier ), 2017.

The paper studies the Rolodex bargaining game in networks. We first revisit the Rolodex game originally proposed in the context of intra firm bargaining, in which a central player bargains sequentially with multiple peripheral players. We show that the unique no-delay SPE of this game yields the Myerson-Shapley value for the star graph in which the central player is linked to all peripheral players. Second, we propose a Rolodex game for a general graph. Links in the graph negotiate sequentially, with one of the linked players making an offer to the other. If the respondent rejects, the link moves to the end of the line and the direction of the offer is reversed for the next negotiation of this link. As in the original Rolodex game, all agreements are renegotiated in the event of a breakdown. We show that the unique no-delay SPE of this game yields the Myerson-Shapley value for the corresponding graph.

‘An Experimental Study of Imperfect Public Monitoring: Efficiency versus Renegotiation-Proofness,’ Fréchette, G.R. and E. Stacchetti (with M. Embrey), 2013.

We study experimentally behavior in a repeated partnership game with public imperfect monitoring, and focus on whether subjects are affected by renegotiation concerns. The signal in our design is rather simple: it indicates only a success or a failure in each period. In some treatments, the equilibrium with the highest payoffs is renegotiation-proof, while in others it is not. Results indicate subjects’ play is affected by the inclusion of a choice that permits some cooperation with more forgiving punishments, but that they do not play the renegotiation-proof equilibrium. However, when the renegotiation hypothesis predicts forgiving (short) punishments, subjects using cooperative strategies are indeed more likely to be forgiving. The experiment also reveals the use of strategies that have not been documented before, highlighting the importance of exploring different monitoring structures. Finally, our design includes communication, which we observe to be used to reduce strategic uncertainty.

‘Predictive Repeated Game Theory: Measures and Experiments,’ Mathevet, L. (with J. Romero), 2014.

One of the fundamental results in repeated games is the Folk theorem, which predicts a plethora of equilibrium outcomes. Many have argued that this extreme number of equilibria is a virtue, as it can explain a variety of different behaviors. However, this result leaves us with almost no predictive power. This paper provides measures for evaluating the predictive power of a theory given experimental data. After running experiments with human subjects in the experimental laboratory, we use these measures to compare a variety of different theories including Mathevet (2014)’s axiomatic approach, and Ioannou and Romero (2014)’s learning model.

‘The People’s Perspective on Libertarian-Paternalistic Policies,’ Rubinstein, A. (with A. Arad), 2017.

Online experiment is carried out in three countries to examine the non-material welfare implications of libertarian-paternalistic (soft) government interventions. We investigate people’s attitudes towards such interventions and their choices in several hypothetical scenarios of government involvement. We identify a significant proportion of people who (1) think negatively of soft government interventions, (2) forgo the encouraged action presumably in protest against such government interventions even though they would have chosen it otherwise, or (3) prefer the government to only provide information to the public in order to influence their choices rather than an intervention with a more effective choice architecture. In two of the countries, a majority of non-objectors to the soft intervention don’t object to harder interventions either, such as the imposition of taxes. The above findings illuminate the potential welfare loss of a non-negligible portion of the population caused by soft government interventions.

‘Trust Me: Communication and Competition in Psychological Games,’ Schotter, A. (with M. Agranov, U. Dasgupta), 2018.

In this paper we study a communication game with and without competition between senders and embed it in the framework of a psychological game where players experience a wide range of emotions such as guilt, lying, and disappointment aversion. We theoretically characterize the set of equilibria that can be sustained in the game with competition and the game without it and test these predictions in a controlled laboratory experiment. Here we induce material payoffs as well as psychological payoffs. Theoretically, an introduction of the psychological payoffs into a communication game is unequivocally beneficial since without them the no-trade equilibrium is the only equilibrium possible. Moreover, the effect of competition on market outcomes in such a communication game is theoretically ambiguous as it depends on which equilibrium subjects choose to play. Empirically, we find that competition is welfare decreasing for both buyers and sellers. This result is driven by an increase in the amount of lying that competing sellers engage in order to attract business from the buyers, who fail to recognize such an effect and tend to believe these messages more than they should.

‘Social Learning and The Winner’s Curse,’ Schotter, A. (with A. McClellan), 2018.

This paper adds to the existing experimental (and theoretical) common value auction literature in two ways. First we introduce social learning into an auction and demonstrate that such learning can be dysfunctional, i.e., lead to a greater incidence of the winners curse. Second, on a methodological level, we introduce some methods not previously used in the laboratory. More precisely, we elicit the entire bid function of bidders each period during the auction rather than eliciting a single bid associated with one particular value (or signal). We demonstrate that the winners curse is not uniform over the domain of the bid function – – it is less pronounced for high as opposed to low signals (values).

‘Learning and Mechanism Design: An Experimental Test of School Matching Mechanisms with Intergenerational Advice,’ Schotter, A. (with T. Ding), 2018.

While the mechanisms that economists design are typically in the form of static one-shot games, in the real world mechanisms are used repeatedly by generations of agents who engage in the mechanism for a short period of time and then pass on advice to their successors. Hence, behavior evolves via social learning and may diverge dramatically from that envisioned by the designer. We demonstrate that this is true of school matching mechanisms, even those, like the Gale-Shapley deferred acceptance mechanism in which truth-telling is a dominant strategy. Put differently, our results indicate that experience with an incentive compatible mechanism may not foster truthful revelation if that experience is achieved via social learning.

‘Complementary Institutions and Economic Development,’ Schotter, A. (with A. Kloosterman), 2016.

This paper considers the problem of why societies develop differently, a question most recently articulated by Acemoglu and Robinson (2012). We follow North (1990) in defining institutions as the “rules of the game in society.” The question then becomes why do some societies develop functional institutions while others do not? To investigate this question, we develop and examine a specific type of dynamic game (which we call an Institutional Game). Our point is that complementarities among the choices that all societies make as they develop can help to answer this question.

‘Matching and Chatting: An Experimental Study of the Impact of Network Communication on School-Matching Mechanisms,’ Schotter, A. (with T. Ding), 2016.

While, in theory, the school matching problem is a static non-cooperative one shot game, in reality the “matching game” is played by parents who choose their strategies after consulting or chatting with other parents in their social networks. In this paper we compare the performance of the Boston and the Gale-Shapley mechanisms in the presence of chatting through social networks. Our results indicate that allowing subjects to chat has an important impact on the likelihood that subjects change their strategies and also on the welfare and stability of the outcomes determined by the mechanism.

‘Rules and Commitment in Communication: An Experimental Analysis,’ Fréchette, G.R. and A. Lizzeri (with J. Perego), 2018.

We investigate models of cheap talk, information disclosure, and Bayesian persuasion, in a unified experimental framework. Our umbrella design permits the analysis of models that share the same structure regarding preferences and information, but differ in two dimensions: the rules governing communication, which determine whether information is verifiable; and the sender’s commitment power, which determines the extent to which she can commit to her communication strategy. Commitment is predicted to have opposite effects on information transmission, depending on whether information is verifiable. Our design exploits these variations to explicitly test for the role of rules and commitment in communication. Our experiments provide general support for the strategic rationale behind the role of commitment and, more specifically, for the Bayesian persuasion model of Kamenica and Gentzkow (2011). At the same time, we document significant quantitative deviations. Most notably, we find that rules matter in ways that are entirely unpredicted by the theory, suggesting a novel policy role for information verifiability.

‘Dynamic Linear Economies with Social Interactions,’ Bisin, A. (with O. Ozgur, Y. Bramoulle), 2018.

Social interactions are arguably at the root of several important socio-economic phenomena, from smoking and other risky behavioural patterns in teens to peer effects in school performance. We study social interactions in linear dynamic economies. For these economies, we are able to (i) obtain several desirable theoretical properties, such as existence, uniqueness, ergodicity; to (ii) develop simple recursive methods to rapidly compute equilibria; and to (iii) characterize several general properties of dynamic equilibria. Furthermore, we show that dynamic forward-looking behaviour at equilibrium plays an instrumental role in allowing us to (iv) prove a positive identification result both in stationary and non-stationary economies. Finally, we study and sign the bias associated to disregarding dynamic equilibrium, e.g., postulating a sequence of static (myopic) one-period economies, a common practice in empirical work.

‘On Blame and Reciprocity: An Experimental Study,’ Schotter, A. (with B. Celen and M. Blanco), 2016.

The theory of reciprocity is predicated on the assumption that people are willing to reward kind acts and to punish unkind ones. This assumption raises the question as to how to define “kindness.” In this paper we offer a novel definition of kindness based on a notion of blame. This notion states that in judging whether player i is kind or unkind to player j, player j has to put himself in the position of player I and ask if he would act in a manner that is worse than i does. If player j would act in a worse manner than player i acted, then we say that player j does not blame player i. If, however, player j would be nicer than player i was, then we say that player j blames player i. We consider this notion a natural, intuitive and empirically functional way to explain the motives of people engaged in reciprocal behavior. After developing the conceptual framework, we test this concept by using data from two laboratory experiments and find significant support for the theory.

‘Is Response Time Predictive of Choice? An Experimental Study of Threshold Strategies,’ Schotter, A. and I. Treviño, 2014.

This paper investigates the usefulness of non-choice data, namely response times, as a predictor of threshold behavior in a simple global game experiment. Our results indicate that the signal associated to the highest or second highest response time at the beginning of the experiment are both unbiased estimates of the threshold employed by subjects at the end of the experiment. This predictive ability is lost when we move to the third or higher response times. Moreover, the response time predictions are better predictors of observed behavior than the equilibrium predictions of the game. They are also robust, in the sense that they characterize behavior in an “out-of-treatment” exercise where we use the strategy method to elicit thresholds. This paper is the first to point out the predictive power of response times in a strategic situation

‘Attention in Games: An Experimental Study,’ Schotter, A. and A. Avoyan, 2018.

A common assumption in game theory is that players concentrate on one game at a time. However, in everyday life, we play many games and make many decisions at the same time and have to decide how best to divide our limited attention across these settings. In this paper we ask how players solve this attention-allocation problem. We find that the attention of players is attracted to particular features of the games they play and how much attention a subject gives to a particular game depends on the other game that he or she is simultaneously attending to.

‘Bringing “Suspense and Surprise” to Data,’ Schotter, A. and A. McClellan (with J.B. Kessler), 2015.

In “Suspense and Surprise”, Ely, Frankel and Kamenica (2015, henceforth EFK) derive suspense-optimal information policies for entertainment when agents derive utility from suspense and surprise. One difficulty with bringing EFK to data, however, is that their measure of suspense requires information about agents’ beliefs about their future beliefs, which is non-trivial to observe or elicit. Using the EFK model, we develop a measure of how far entertainment is from the optimal suspense path that relies only on contemporaneous beliefs. We therefore show that EFK is easier to use empirically than expected. We bring this insight to data with an experiment in which we compare beliefs of subjects watching NFL football games to those watching highly popular scripted television dramas. Following EFK, we hypothesize that popular scripted television shows should have a closer-to-optimal suspense path than football games since writers can more finely control viewers’ beliefs than can sports contests. Surprisingly, we find that the NFL games more closely match the optimal suspense structure without sacrificing on surprise. We discuss the implication of this finding for the model in EFK.

‘Mood and Economic Decision Making: Experimental Evidence,’ Schotter, A. and A. McClellan (with J.B. Kessler), 2015.

In this paper we investigate the impact of short-term fluctuations in happiness (i.e. a positive mood) on decision-making. The emotional fluctuations we analyze are orthogonal to the fundamentals of the decision maker’s wellbeing and standard theory suggests they should have no effect. We perform an experiment in a sports bar on the Upper East Side of Manhattan while subjects watch an NFL football game. During numerous commercial breaks, we monitor subjects’ happiness and ask them to perform a set of standard experimental tasks involving charitable giving, risk taking, trust, and willingness to pay. We find that the events of the game not only affect the mood of our subjects but that these changes in mood lead to predictable changes in decisions. Our experiment introduces a new method for inducing a dynamic mood sequence and provides a model that can explain why changes in short-term emotions influence economic decisions.