Andy Schotter
NYU Department of Economics

New, Forthcoming and Revised Papers

1) "Sampling for Information or Sampling for Imitation: An experiment on social learning with information on ranks" (With Theo Offerman) , To appear in Games and Economic Behavior, 2008

 2) "Talking Ourselves to Efficiency: Coordination in Inter-Generational Minimum Effort Games with Private, Almost Common and Common knowledge of Advice", (With Barry Sopher and Ananish Chaudhuri) To appear in Economic Journal, 2008.

3) "Trust and Trustworthiness in Games: An Experimental Study of Intergenerational Advice" (With Barry Sopher) ( Experimental Economics, 2007)

4) "Advice and Behavior in Intergenerational Ultimatum Games: An Experimental Approach",  (With Barry Sopher)  ( Games and Economic Behaviour, 2007)

5) "Creating Competition Out of Thin Air: An Experimental Study of Right-to-Choose Auctions" (With Kfir Eliaz and Theo Offerman,  Games and Economic Behavior, March 2008

6) "Learning Under Supervision: An Experimental Study", (With Raghuram Iyengar) Forthcoming, Management Science, 2008

7) "Convergence: An Experimental Study " (With Kyle Hyndman, Erkut Ozbay, and Wolly Ehrblatt) [New]

8) "Hyperbolic Discounting: An Experimental Analysis", (with Jess Benhabib and Alberto Bisin) [New]

9) On the Severity of Bank Runs: An Experimental Study , (With Tanju Yorulmazer)  [Revised].Appendix,

10) "Words Speak Louder than Actions and Improve Welfare:
An Experimental Test of Advice and Social Learning", (With Boğaçhan Çelen and Shachar Kariv) [Revised].

11) "On the Informational Content of Advice: A Theoretical and Experimental Study",  (With Yaw Nyarkoand Barry Sopher) [Economic Theory, 2005].

12) "Paying for Confidence: An Experimental Study of Preferences Over Beliefs" (With Kfir Eliaz) (New)
 

My Current Experimental Economics Projects At The Center for Experimental Social Science

Project on Learning

This series of papers investigates the process by which economic agents go about the task of learning. The papers are divided into two groups. Papers 1-3 investigate the question of which types of economic environments are most conducive to learning. They use a technique called a "surprise quiz" to test what exactly it is that subjects learn when they repeatedly partake in a single or multi-person decision problem. What is discovered is that the process by which information is generated and the method used to compensate subjects as they learn is crucial both to how they go about the task of learning, what they choose to learn about, and how well they learn. Papers 4-6 investigate belief-learning models by eliciting the beliefs of subjects using a proper scoring rule. Since we are able to observe the beliefs of subjects directly, we are in a position to give the belief-learning theory its best test yet. What we find is that our elicited beliefs do a better job of explaining behavior than do the types of historical beliefs so often used in theoretical work (i.e. fictitious play). Paper 6 will compare this best belief learning model to the best available reinforcement learning model to see which explains the data best.

Papers

1. "A Surprise Quiz View of Learning" (Andrew Schotter and Antonio Merlo), Games and Economic Behavior, 1999.

2. "Learning by Not Doing "(Andrew Schotter and Antonio Merlo), Forthcoming,  Games and Economic Behavior, 2003.

3. "Learning Under Supervision: An Experimental Study", (With Raghuram Iyengar)

4. "An Experimental Investigation of Belief Learning With Real Beliefs" (Andrew Schotter and Yaw Nyarko),   , Econometrica.

5. "Comparing Learning Models Using Ideal Micro-data Sets”, (Andrew Schotter and Yaw Nyarko), Under Revision, Games and Economic Behavior.

6. "Convergence: An Experimental Study " (With Kyle Hyndman, Erkut Ozbay, and Wolly Ehrblatt) [New]

 Project on Intergenerational Games and Advice

While most game theory investigates the behavior of economic agents in games that are played either once and only once or repeatedly between the same set of players, in the real world most games are played very differently. More precisely, in the real world, while games may have infinite lives (i.e., there may always be a Ford and G.M. competing in the car market, or an I.B.M. and Compac competing in the computer market) the agents who play in these games live for only a finite length of time and are replaced periodically as they either retire or die. When the transition between players takes place, the older generation informs their successors of all of the norms and conventions of behavior that have been created either during their lifetime or in the lifetimes of their predecessors. It is these norms and conventions that are the focus of papers 1-3 below. Papers 4-5 use this intergenerational game set-up and investigates how different learning in games becomes when instead of taking place as one player repeatedly plays the same game over and over, a succession of players do so each coming into the game with a different prior about the behavior of his or her opponent. These papers have convinced us that games played with advice are played quite differently than games played without advice. Advice leads to behavior that is more like that predicted by economic theory, and makes behavior more “conventional” in that the variability in behavior between generations is dampened. In social learning settings, we find that there are more herds and cascades when advice exists. 

Papers

1. "Social Learning and Coordination Conventions in Inter-Generational Games: An Experimental Study", Andrew Schotter and Barry Sopher (Journal of Political Economy, June 2003).

2.  "Advice and Behavior in Intergenerational Ultimatum Games: An Experimental Approach",  (With Barry Sopher)  (2007, Games and Economic Behavior)

3.  "Trust and Trustworthiness in Games: An Experimental Study of Intergenerational Advice" (With Barry Sopher) (2007, Experimental Economics)

4. "Talking Ourselves to Efficiency: Coordination in Inter-Generational Minimum Effort Games with Private, Almost Common and Common knowledge of Advice", (With Barry Sopher and Ananish Chaudhuri) [Revised]

5. "Words Speak Louder than Actions and Improve Welfare:
   An Experimental Test of Advice and Social Learning", (With Boğaçhan Çelen and Shachar Kariv) [Revised]

6. "Learning in Intergenerational Games" (Andrew Schotter, Barry Sopher, and Ananish Chaudhuri), Economic Bulletin, 2007

Project on Group Incentives and Tournaments

These papers investigate the principal-agent problem involved in trying to create work incentives among a group of economic agents whose efforts can not be observed but whose joint (or individual) output can. Paper 1 looks at the group incentive problem and compares the performance of a number of group incentive formuli used widely in the business world (i.e. profit sharing, partnerships, gains-sharing, and team based tournaments). What is found is that the tournament- based formula outperforms all the rest. Papers 2-3 look at tournaments, or relative-compensation mechanisms, and tests their performance capabilities in both symmetric and asymmetric situations. On the whole they perform quite well. Finally paper 4 looks at the use of long term bonus plans often used in industry and demonstrates that unless they are properly designed they might easily fail to stimulate the performance of workers.

Papers

1. "Productivity Under Group Incentives: An Experimental Study" (Andrew Schotter and Haig Nalbantian), American Economic Review, Vol. 87, No.3, pp. 314-340, June 1997.

2. "Tournaments and Piece rates: An Experimental Study", (Andrew Schotter, Clive Bull and Keith Weigelt), Journal of Political Economy, Vol. 95, No.1, 1987, pp. 1-33.

3. "Multiperson Tournaments: An Experimental Examination", (With Alannah Orrison, and Keith Weigelt)
   "Behavioral Consequences of Corporate Incentive and Long Term Bonuses: An Experimental Study", Management Science, September 1992.

4. “Workaholics and Drops in Optimal Organizations” with Wieland Mueller, , N.Y.U., 2003

.

Project on Advice Giving and Following 

My work on inter-generation games described above indicated that advice passed on between generations of players is of  fundamental importance to behavior. People act differently in games where advice has been offered even if that advice is not offered by an expert but simply someone who has marginally more experience and information than they do. Further,  behavior in games with advice  is more consistent with economic theory and efficiency increasing. The question then is why is advice efficiency increasing, why do people follow advice, and what model explains advice giving? Also This research project investigates the informational content of advice and by setting up a market for advice, looks into what factors makes advice valuable. Finally, the paper number 3 looks at the influence of advice on the ability of a decision maker to learn. Work on this project is funded by the National Science Foundation.

Papers

1. "On the Informational Content of Advice: A Theoretical and Experimental Study",  (With Yaw Nyarko and Barry Sopher) [Economic Theory, 2005].

2. "Decision Making With Naive Advice", American Economic Review, May 2003.

3. "Learning Under Supervision: An Experimental Study", (With Raghuram Iyengar), forthcoming Management Science
    

Project on Bank Runs 
(with Tanju Yorulmazer)

 In this project we investigate financial crises such as bank runs by designing an experiment where we create a bank run in the lab. We are interested in examining what factors make some bank runs more severe than other and what policies are successful in dampening them.

Papers

1. On the Severity of Bank Runs: An Experimental Study , (With Tanju Yorulmazer)  [Revised].

Current Graduate Student Projects at the Center for Experimental Social Science (CESS)

The Center for Experimental Social Science

The Center for Experimental Social Since at New York University is an interdisciplinary center started in the Fall of 2001 dedicated to doing experimental research in the social sciences. It is comprised members of the Economics, Psychology and Politics departments. At present I am its director and administer the center in conjunction with a set of Co-Directors including Andrew Caplin of the Economics Department, Tom Tyler of Psychology, Yaw Nyarko of Economics and George Downs of Politics. More recently we have started a close affiliation with the newly created Center for Neuro-Economics  and Decision Making which is headed by Paul Glimcher of the Neuro Science department. The Center has a large number of Research Associates and Faculty Affiliates (a list of which is attached at the end of this section of the proposal.

The Center runs a weekly seminar in Psychology and Economics, a bi-weekly seminar in experimental economics, and has sponsored a set of conferences most recently on Econometrics and Experimental Economics.  More importantly for this proposal we also run a weekly experimental design seminar where faculty and graduate students present their proposed research and have it discussed as it evolves. This has proven to be an immense aid to research generating at least eight graduate student research projects a year. The Center has recently hired two young faculty Eric Dickson of Politics and Guillaume Frechette of Economics and will be expanding in the future on both the senior and junior level.

Recently Published work by Graduate Students:

Former student members of the Center have been extremely successful in publishing their work in top journals. Below is a list of these accomplishments.

Distinguishing Informational Cascades from Herd Behavior in the Laboratory  

 Boğaçhan Çelen (now at Columbia Business School) and Kariv, Shachar (now at Berkeley Economics) American Economic Review, June 2004, 94(3), pp. 484-497.

Abstract. This paper reports an experimental test of how individuals learn from the behavior of others. By using techniques only available in the laboratory, we elicit subjects' beliefs. This allows us to distinguish informational cascades (convergence of beliefs) from herd behavior (convergence of actions). By adding a setup with continuous signal and discrete action, we enrich the ball-and-urn observational learning experiments paradigm of Anderson and Holt (1997). We test a model that explains subjects' behavior as a form of generalized Bayesian behavior that incorporates limits on the rationality of others. We find strong evidence that, in Bayesian terms, subjects put too much weight on their own information and too little weight on the public information. Put differently, subjects are overconfident in the precision of their private information. To put the observed behavior into perspective, we use a simple modification of the Bayesian model, which provides a framework that enables us to understand individual behavior in the laboratory. 

An Experimental Test of Observational Learning under Imperfect Information

Boğaçhan Çelen (now at Columbia Business School) and Kariv, Shachar (now at Berkeley Economics)  Forthcoming, Economic Theory. 

 Abstract. To explore the difference between social learning under perfect and imperfect information, this paper takes an experimental look at a situation in which individuals learn by observing the behavior of their immediate predecessors. Our experimental design is based on the theory of Çelen and Kariv (Observational Learning Under Imperfect Information) and uses the procedures of Çelen and Kariv (Distinguishing Informational Cascades from herd Behavior in the Laboratory) with the exception that the history of actions observed by subjects is different. We find is that imitation is much less frequent when subjects have imperfect information, even less frequent than the theory predicts. Further, while we find strong evidence that under perfect information a form of generalized Bayesian behavior adequately explains behavior in the laboratory, under imperfect information behavior is not even consistent with this generalization of Bayesian behavior. To reconcile this with the conclusions under perfect information, we undertake a modification of the model that abandons the assumption of common knowledge of rationality.  

Observational Learning Under Imperfect Information

Boğaçhan Çelen (now at Columbia Business School) and Kariv, Shachar (now at Berkeley Economics) Games and Economic Behavior, March 2004, 47(1), pp. 72-86.

Abstract. This paper explores Bayes-rational sequential decision making in a game with pure information externalities, where each decision maker observes only her predecessor's binary action. Under perfect information, the martingale property of the stochastic learning process is used to establish convergence of beliefs and actions. Under imperfect information, in contrast, beliefs and actions cycle forever. However, despite the instability, over time the private information is ignored and decision makers become increasingly likely to imitate their predecessors. Consequently, we observe longer and longer periods of uniform behavior, punctuated by increasingly rare switches. These results suggest that the kind of episodic instability that is characteristic of social behavior in the real world makes more sense in the imperfect-information model, and that the imperfect information premise provides a better theoretical description of fads and fashions. 

Behavioral Aspects of Learning in Social Networks: An Experimental Study, Syngjoo Choi (with Douglas Gale and Shachar Kariv), forthcoming in Advances in Behavioral and Experimental economics, edited by John Morgan, JAI press.

 Abstract. Networks are natural tools for understanding social and economic phenomena. For example, all markets are characterized by agents connected by complex, multilateral information networks, and the network structure influences economic outcomes. In an earlier study, we undertook an experimental investigation of learning in various three-person networks, each of which gives rise to its own learning patterns. In the laboratory, learning in networks is challenging and the difficulty of solving the decision problem is sometimes massive even in the case of three persons. We found that the theory can account surprisingly well for the behavior observed in the laboratory. The aim of the present paper is to investigate important and interesting questions about individual and group behavior, including comparisons across networks and information treatments. We find that in order to explain subjects’ behavior, it is necessary to take into account the details of the network architecture as well as the information structure. We also identify some black spots where the theory does least well in interpreting the data.

Herding and Price Convergence in a Laboratory Financial Market Marco Cipriani and Antonio Guarino, (Forthcoming,  American Economic Review)

Abstract:We study whether herding can arise in a laboratory financial market in which agents trade sequentially. Agents trade an asset whose value is unknown and whose price is efficiently set by a market maker. We show that the presence of a price mechanism destroys the possibility

of herding. Most agents follow their private information and prices converge to the fundamental value. This result contrasts with the case of a fixed price, where herding and cascades arise. When the price moves, however, agents may behave as contrarian, i.e., they may trade against the market, something not accounted for by the theory. Finally, we study wheteher informational cascades arise when trade is costly (e.g, because of a Tobin tax). With trade costs, most subjects rationally decided not to trade and the price was unable to aggregate private information efficiently.

Erkut Ozbay

Auctions with regretful agents, American Economic Review, forthcoming.

 Abstract: The discrepancy between ex-ante and ex-post decisions of individuals is considered as the source of regret.  According to regret theory, if individuals anticipate that they are going to feel regret, they will take this future feeling into account in their decisions.    In this paper, using a  first-price sealed bid auction,  we investigate whether individuals anticipate regret and reflect this feeling in their bidding strategies.

  Experimental evidence shows that if the players are informed that the highest price will be told at the end of an auction, the players anticipate regret and overbid.  However, if they are informed that no information will be given or the second highest bid will be told, they do not anticipate regret and overbidding is not observed.  We also provide a theoretical model explaining these experimental findings. 

Graduate Student Work Under Submission

 Learning in Networks: An Experimental Study, Syngjoo Choi (with Douglas Gale and Shachar Kariv), submitted at the American Economic Review

 Abstract. Individuals living in society are bound together by a social network, the complex of relationships that brings them into contact with other agents. In many social and economic situations, individuals learn by observing the behavior of others in their local environment. This process is called social learning. Learning in incomplete networks, where different agents have different information sets, is especially challenging: because of the lack of common knowledge individuals must draw inferences about the actions others have observed as well as about their private information. Whether individuals can rationally process the information available in a network is ultimately an empirical question. This paper reports an experimental investigation of learning in three-person networks and uses the theoretical framework, Gale and Kariv(2003), to interpret the data generated by the experiments. The family of three-person networks includes several non-trivial architecture, each of which gives rise to its own distinctive learning patterns. We find that the theory can account for the behavior observed in the laboratory in variety of networks and informational settings. To account for errors in subjects behavior, we adapt the model of Quantal Response Equilibrium of McKelvey and Plafrey (1995, 1998) an find that its restrictions are also confirmed. The ‘goodness of fit’ is better for the QRE model than for the game-theory model. This provides an important support for the use of QRE to interpret experimental data.

The Advice Puzzle: An Experimental Study of Social Learning where Words Speak Louder than Actions

Boğaçhan Çelen (now at Columbia Business School) and Kariv, Shachar (now at Berkeley Economics) and Andrew Schotter, NYU. Version: May 27, 2004. Submitted for publication. 

Abstract. This experimental paper studies how individuals learn by observing the behavior of predecessors (Social Learning) as well as from their advice (Word-of-Mouth Learning). What we find is a truly puzzling result that we call the advice paradox. This paradox can be stated as follows: subjects in a laboratory social learning situation played with and without advice appear to be more willing to follow the advice given to them by their predecessor than to copy their action, despite the fact that both pieces of information should, in equilibrium, be equally informative (in fact, identical). The consequence of this advice paradox is that in experiments with advice subjects tend to herd more than they do in experiments where they can only view their predecessor's action. Remarkably, these herds tend to select the correct action and, hence, advice tends to be efficiency increasing when compared to experiments where subjects can only observe their predecessor’s action.

"Convergence: An Experimental Study", Wolf Ze'ev Ehrblatt, Kyle Hyndman and Erkut Y. Ozbay
Abstract
One way to define a Nash equilibrium is by positing a set of beliefs (or conjectures) for
each player over (about) the actions of their opponents that has the property that, given these
beliefs, when each player best responds, the actions taken confirm the initial beliefs. This
rational expectations definition leaves open the question of how beliefs and actions get into this
self-confirming state. For example, do beliefs converge to their equilibrium state ¯rst and drag
actions into alignment or is the process action driven with them converging before beliefs. What
we find is that the process of convergence is one where actions converge before beliefs. However,
after reaching equilibrium in actions, the beliefs of subjects converge to the degenerate beliefs
that place all the weight on the partner's equilibrium action extremely rapidly (within 2 periods
on average). We also identify differences between the early converger and the late converger in
a group | often it is the case that the early converger plays his part of the Nash action profile
long enough to convince his opponent to adhere. Finally, we investigate the process of belief
formation and argue that, unlike all of the most common learning models, the belief formation
process is one that takes into account not only the payoff of the learner but also those of his
opponents.
Graduate Student Projects in Progress:

 Neslihan Uler and Yusufcan Masatlioglu

 Behavioral Differences between Direct and Indirect Mechanisms: Evidence from First Price Auctions

 Abstract: The Revelation Principle depends on a seemingly innocuous assumption that theoretically outcome-equivalent (TOE) mechanisms are behaviorally equivalent as well. However, this strong assertion has not yet been tested in previous experimental studies. In this paper, we aim to fill this gap.

We settled on the first-price sealed-bid auction as our indirect mechanism and then constructed TOE direct mechanisms. In contrast with what theory proposes, the subjects behaved significantly different under direct and indirect mechanisms. We established the following conclusions: (i) The revenue equivalence did not hold - the indirect mechanism generated higher revenue than the direct mechanisms, (ii) the subjects behaved as if they were less risk averse in the direct mechanisms, (iii) moreover, we observed behavioral differences across direct mechanisms. The main implication of these findings is that the Revelation Principle may not be applicable and therefore, it may not be sufficient to focus on the direct mechanisms alone.

 A Test for Reference Dependent Preferences

 Abstract: A sizable amount of experimental data has revealed the following regularity: Individuals are likely to value an object in their possession more than one that is not. This is called the status quo bias in behavioral economics (Samuelson and Zeckhauser (1988)). Most of the experimental studies that find gaps between buying and selling prices provide support for the status quo bias as we understand the term here; see, for instance, Knetsch (1989) and Kahneman, Knetsch and Thaler (1990, 1991). Moreover, Hartman, Doane and Woo (1991) and Johnson et al. (1993) find status quo bias in the field settings. While these results have been treated as anomalies in previous studies, Masatlioglu and Ok (2004) develop the first rational choice theory with status quo bias, that can consistently explain such findings. Our model, given that an agent satisfies a set of axioms, predicts a discrepancy between buying and selling prices of goods, and offers a new explanation for the (in)famous preference reversal phenomenon. In this paper we test both the implications and the prediction power of our model compared to other existing models on reference dependent preferences.

Syngjoo Choi

 Cognitive Hierarchy of Learning in Networks: Theory and Experimental Evidence (Job Market paper)

 Abstract: Motivated by two earlier papers (Choi, Gale and Kariv (2004 a,b)), I propose a model and an experimental evidence of how the complexity of decision problems interact with the degree of bounded rationality in learning in various social networks. In CGKs, we tested experimentally the Bayesian learning in networks and found out that the large-scale behavior in the lab accords well with the theoretical predictions with introducing trembling in actions. While the results are interesting, we have also identified situations where the theory does less well in explaining subjects’ behavior. Specifically, there are instances of reduced rationality and efficiency in subjects’ behavior in the star network and in the treatments where some subjects have no private information. These observations lead us to the tentative conclusion that, although the Bayesian learning model accounts adequately for the large-scale features of the data, there are clear departures from the theory of Bayesian learning when we look at the behavior of individual subjects.

            As an alternative, I propose a model where individuals are assumed to exhibit different levels of cognitive ability. Specifically, individuals are assumed to be all Bayesian but have heterogeneous cost functions of cognitive reasoning. They differ in the number of levels of the hierarchy of beliefs they can use in decision-making. In particular, the model will generate a class of decision rules that are characterized by the stopping time of information processing. This model of cognitive hierarchy yields a parsimonious econometric specification which is identifiable under some mild setting. It will allow me to undertake the structural estimation of the type distribution within the subject pool and the choice probabilities associated with each decision rule at each decision period. . Preliminary work indicates that there is a significant proportion of sophisticated individuals and a smaller proportion of less sophisticated individuals. In fact, simple decision rules corresponding to the types of the cognitive hierarchy can be identified in the raw data. The estimated type distributions reflect these features and it is clear that they depend on the structure of the social network.

 Network Formation and Social Learning: An Experimental Study, (with Boğaçhan Çelen and Kyle Hyndman)

 Abstract: Networks are good tools to describe how complex and what kinds of the structure of economic and social relationships emerge through the interactions among individuals. For example, some agents and institutions work as a center for multilateral trades. Çelen, Choi, and Kariv (2004) study the process of formation of information networks in the standard environment of sequential social learning. Specifically, unlike the perfect information structure of social learning such as Bikchandhani, Hershleifer and Welch (1992), we endogenize the information structure of observing others by individuals’ link decisions. When individuals show up in the sequential decision queue, they can observe some of preceding agents’ actions if they form a costly link to them. Under mild conditions of the parameters of cost and information, we found that such network formation enhances the aggregation of information through the process of social learning compared to the regime of perfect information.

            This paper studies an experimental investigation of network formation through social learning and utilizes the theoretical framework by ÇCK. Especially, we focus on four-person games of network formation and test the theoretical predictions with various variations of information structure. Numerous questions are interestingly proposed. First, how subjects behave according to Bayesian rationality? For example, do they perceive rationally the information that can be inferred from the structure of networks? Secondly, a huge network may contain conflicted signals. Thus, individuals may decide to not experiment the information in such a huge network. We can call this phenomenon the paradox of the network size. Can we replicate such phenomenon in the experimental laboratory?

 Monotone Games in Voluntary Contribution of Public Goods: An Experimental Study, (with Douglas Gale and Shachar Kariv)

 Abstract: We study experimentally one class of monotone games where a set of actions decrease monotonically as the game is played, which were theoretically studied by Gale (2000), in the environment of dynamic voluntary contribution of public goods. Specifically, a group of subjects are initially endowed with one unit of a private good. At the first period, they decide simultaneously whether they keep that unit for private consumption or contribute it to public goods. At the next period, subjects who still keep their unit for private consumption can make the same decisions simultaneously. The game is played at the finite time horizon. Due to the irreversibility of the contribution to public goods, subjects who already contributed their unit do not have any more decisions at later periods. The theory predicts that, under some mild set of parameters such as the payoff structure and the length of time period, there is always positive probability of the provision of public goods.

            We focus on the three-player game where a group of three players interacts for finite time horizon to provide a public good. From the initial set of experiments, we found that the provision of public goods is quite high across most of the treatments, as the theory predicts. Especially, in the treatments where the public good is provided if at least two players contribute their units to the public good, the game is quite strategic and most of players wait till the end of the game to see whether their opponents contribute. In order to explain subjects’ behavior, we estimate subjects’ choice probabilities under the equilibrium restriction imposed by Quantal Response Equilibrium. Somewhat surprisingly, the estimated choice probabilities explain the transition probabilities in the game from the experimental data. Further, we conducted an analysis of individual behaviors. Most of subjects are quite strategic, who may be classified either as ‘conditional cooperator’ or ‘free rider’, while a small proportion of subjects behave as if they initially contribute their unit to the public good to encourage others to contribute at later period as well. This may suggest that an equilibrium analysis based on individual heterogeneity explains the data better, as we will further seek.

 Kyle Hyndman

 Biases in Bargaining Over The Business Cycle

 Abstract:Consider the following puzzle. Suppose that two agents are interacting repeatedly in an economy subject to the fluctuations of the business cycle. Is it more likely that the agents will agree to some collective action such as dividing a pie or restricting output with collusive behaviour when the economy is in a boom rather than in a recession? Rather surprisingly, we see many examples which indicate a special relationship between disagreements and the business cycle. For example, there is a well-documented empirical finding that strikes are less frequent but longer lasting during recessions. Furthermore, in studying production announcements by the Organisation of Petroleum Exporting Countries (OPEC), Hyndman [2004] argued that agreements to tighten quotas are less

likely than are agreements to loosen them.

It is our belief that reference points and entitlement effects are at play in the above situation. For example, if OPEC members see their quota as a property right to which they are entitled, they may be unwilling to see property rights be removed, even though market conditions call for tightening. Of course, reference points have been studied before. Indeed, Zajac [1995] argues that the status quo is a strong reference point in certain situations and that feelings of entitlement to the status quo are often more important than notions of fairness. Bazerman [1985] has argued that reference points play an important role in wage arbitration. We wish to explore these issues further on a number of grounds.

1. Are agreements more likely when the size of the pie is expanding rather than contracting? Even if not, do agreements come about faster when the pie is expanding?

2. At the conclusion of the sequence of bargaining rounds, does the final allocation look much the same as or significantly different from the initial bargaining outcome?

3. How has the initial allocation of entitlements affected the outcome? Do property rights wear out?

4. Are agreements more likely in later bargaining rounds? Does the sequence of pies make a difference?

5. Does the degree of inequality in initial entitlements matter?

6. What is the degree of inefficiency in bargaining?

 An Experimental Study of Belief Elicitation (with  Wolf Ze’ev Erhblatt and Erkut Ozbay)

 Abstract: We use belief elicitation procedures in the laboratory in order to answer a number of important economic and behavioral questions. Subjects play a normal form game for a fixed number of periods with an opponent. In addition to action choices, we elicit their beliefs regarding the likely actions of their opponents. We then ask the following questions:

1. Does play converge to a Nash equilibrium?

• If so, do beliefs converge to equilibrium beliefs?

• Do beliefs converge faster or slower than actions?

2. Do players play a best response to their beliefs or are their action choices governed

    by some other motive such as fairness, reciprocity, spite, etc?

3. How does the complexity of the game affect beliefs and convergence to equilibrium?

• What are the beliefs associated with dominated strategies?

• Can we see the iterated elimination of dominated strategies through subjects’

stated beliefs?

4. Does a group of observers have a similar process of belief formation as the subjects

who both choose actions and report beliefs?

Preliminary results addressing questions 1 and 2 suggest that there is some tendency to fairness and efficiency—even if doing so is at odds with Nash equilibrium. Moreover, subjects do not best respond to their stated beliefs. This seems especially so in games for which there is a non-equilibrium strategy pair which gives payoffs higher than the Nash equilibrium

Matthew Embrey

 Reputation Formation in Bargaining: A Proposed Experiment

      Abstract: The aim of this project is to test experimentally the bargaining model of Abreu and Gul 2000. They add a role for reputation in bargaining by adding behavioural types: players who always demand a fix amount and will never accept an offer that would give them less. This has stark consequences for the equilibrium behaviour of rational agents, which give the model a much richer set of equilibrium predictions.

    The proposal is to run the following stylised bargaining game (from Abreu and Gul 2000) in the lab: In stage one players simultaneously announce the share of the pie they would like. If the sum of the demands is larger than the pie, the game moves to stage 2, where players engage in a war of attrition over whose division to adopt. Equilibrium play predicts that rational agents replicate the behavioural types. More specifically, every "aggressive" behavioural type (one that demands strictly more than (1/2) of the pie) is replicated with strictly positive probability. The main objective will be to test whether subjects recognise this and act accordingly. The feature of the experimental design that will facilitate this is the addition of computer players to the subject pool. The computers will follow a commonly known behavioural rule. Irrespective of the possible types in the population of subjects, if the computer players are "aggressive" then rational subjects should replicate each of the computers with positive probability. By altering the number of computer players and their corresponding rules, a comprehensive analysis of whether subjects are acting strategically can be undertaken.

 Anna Bassi (Department of Politics)

 An Experimental Study on Fair Division

 Abstract: We consider the problem of allocating a bundle of perfectlydivisible goods through two different mechanisms proposed byCrawford. Assuming selfish, rational players, both the procedures are able to produce Pareto-efficient and envy-free outcomes, but the latter, by auctioning off the role of divider instead of allocating it randomly, produces outcomes that satisfy also an equitability property. In the experiments presented in this paper we examine two treatments, each one testing one mechanism. The model incorporating the auction produces efficient, envy free, and equitable outcomes, even if players are not completely selfish and rational. This happens because of two reasons: the role of divider is not chosen by nature, but it is earned through a fair competition, and dividers pay the auction by transferring part of his payoffs to choosers.

Voting Systems and Strategic Manipulation

Abstract:  We present an experiment on manipulability of Voting Systems. We compareThe manipulability of four voting systems: two ranked procedures (Borda count\cite{Borda} and Hare system) and two unranked rules (Plurality and Approval voting). In the experiments presented in this paper we examine three treatments: incomplete information (players know polls), complete information, and communication game (players may coordinate their actions). Any preferences aggregation method is vulnerable to strategic manipulation by voters. But, we expect that, among the voting procedures, Approval Voting shows one of the lowest degree of manipulation. Supporters of minority candidates would not be torn away simply because there was another candidate who, though less appealing to them, was generally considered a stronger contender. Approval voting would allow these supporters to vote for both candidates, then they would not be tempted to desert the one who is weak in the polls, as under plurality voting. In this scenario, Approval voting should encourage voters to be more sincere.

Selected Papers

1. Bogaçhan Çelen and Shachar Kariv, "Distinguishing Informational Cascades from Herd Behavior in the Laboratory," Accepted, AER.

2. Bogaçhan Çelen and Shachar Kariv, "An Experimental Test of Observational Learning under Imperfect Information," Accepted, ET. 

3. Marco Cipriani and Antonio Guarino, “Information Cascades in a Market With an Efficient Specialist” , submitted American Economic Review.

4. Tanju Yorumazer, "On the Severity of Bank Runs: AN Experimental Study", (with Andrew Schotter), Submitted American Economic Review.

5. Neslihan Uler and Yusufcan Masatlioglu,

   Behavioral Differences between Direct and
   Indirect Mechanisms: Evidence from First Price Auctions".
    

[Home] [Contact.html] [Vitae.html] [Books] [Research] [Teaching] [C.E.S.S.] [Experiments] [Links] [Directions]