nep-exp New Economics Papers
on Experimental Economics
Issue of 2015‒11‒01
twenty-six papers chosen by

  1. Endowment Origin, Demographic Effects and Individual Preferences in Contests By Simon, Curtis J.; Sheremeta, Roman
  2. Tug-of-War in the Laboratory By Deck, Cary; Sheremeta, Roman
  3. Risk taking and information aggregation in groups By Spiro Bougheas; Jeroen Nieboer; Martin Sefton
  4. Myopic behavior and overall utility maximization - A study of linked hawks and doves - By Siegfried Berninghaus; Stephan Schosser; Bodo Vogt
  5. Behavioral effects of tax withholding on tax compliance: Implications for information initiatives By Christian Vossler; Michael McKee
  6. Grit Trumps Talent? An experimental approach By Gerhards, Leonie; Gravert, Christina
  7. Experience and Gender Effects in an Acquiring-a-Company Experiment Allowing for Value Messages By Noemi Pace; Daniela Di Cagno; Arianna Galliera; Werner Güth; Luca Panaccione
  8. Compliance Behavior in Networks: Evidence from a Field Experiment By Francesco Drago; Friederike Mengel; Christian Traxler
  9. Competition Between and Within Universities: Theoretical and Experimental Investigation of Group Identity and the Desire to Win By Chen, Zhuoqiong; Ong, David; Sheremeta, Roman
  10. Accents, Race and Discrimination: Evidence from a Trust Game By Ece Yagman; Malcolm Keswell
  11. Market Efficiency, Trading Institutions and Information Mirages: evidence from an experimental asset market By Morone, Andrea; Nuzzo, Simone
  12. New Hampshire Effect: Behavior in Sequential and Simultaneous Election Contests By Irfanoglu, Zeynep; Mago, Shakun; Sheremeta, Roman
  13. Who cares about the balderdash I spouted yesterday?* – An experiment on the volatility of bargaining norms – By Stephan Schosser
  14. Does Gender Diversity Promote Non-Conformity? By Amini, Makan; Ekström, Mathias; Ellingsen, Tore; Johannesson, Magnus; Strömsten, Fredrik
  15. Exploding Offers with Experimental Consumer Goods By Alexander L. Brown; Ajalavat Viriyavipart; Xiaoyuan Wang
  16. The impact of study guides on “matric” performance: Evidence from a randomised experiment By Stephen Taylor; Patricia Watson
  17. Behavior in Group Contests: A Review of Experimental Research By Sheremeta, Roman
  18. The "tone effect" of news on investor beliefs: An experimental approach By Bosman, Ronald; Kräussl, Roman; Mirgorodskaya, Elizaveta
  19. The costs and benefits of symmetry in common-ownership allocation problems. By Alexander L. Brown; Rodrigo A. Velez
  20. Preferences-dependent learning in the Centipede game By Astrid, Gamba; Tobias, Regner
  21. Behavior in All-Pay Auctions with Ties By Gelder, Alan; Kovenock, Dan; Sheremeta, Roman
  22. Clinical trial design enabling e-optimal treatment rules By Charles F. Manski; Aleksey Tetenov
  23. Meaningful Learning in Weighted Voting Games: An Experiment By Eric Guerci; Nobuyuki Hanaki; Naoki Watanabe
  24. Labor market effects of reemployment programs: Experimental evidence during the great recession. By Marios Michaelides; Peter Mueser
  25. Confidence Biases and Learning among Intuitive Bayesians By Louis Lévy-Garboua; Muniza Askari; Marco Gazel
  26. Endogeneity and non-response bias in treatment evaluation - nonparametric identification of causal effects by instruments By Fricke, Hans; Frölich, Markus; Huber, Martin; Lechner, Michael

  1. By: Simon, Curtis J.; Sheremeta, Roman
    Abstract: In modern firms the use of contests as an incentive device is ubiquitous. Nonetheless, experimental research shows that in the laboratory subjects routinely make suboptimal decisions in contests even to the extent of making negative returns. The purpose of this study is to investigate how earning the endowment, demographic differences and individual preferences impact behavior in contests. To this end, we conduct a laboratory experiment in which subjects expend costly resources (bids) to attain an award (prize). In line with other laboratory studies of contests, our results show that subjects overbid relative to theoretical predictions and incur substantial losses as a result. Making subjects earn their initial resource endowments mitigates the amount of overbidding and thus increases efficiency. Overbidding is linked to gender, with women bidding higher than men and having lower average earnings. Other demographic information, such as religiosity, and individual preferences, such as preferences towards winning and risk, also influence behavior in contests.
    Keywords: contest, experiments, overbidding, endowment, gender, religiosity
    JEL: C72 C91 D61 D72 J16
    Date: 2015–03–25
  2. By: Deck, Cary; Sheremeta, Roman
    Abstract: Tug-of-war is a multi-battle contest often used to describe extended interactions in economics, management, political science, and other disciplines. While there has been some theoretical work, there is scant empirical evidence regarding behavior in a tug-of-war game. To the best of our knowledge, this paper provides the first experimental study of the tug-of-war. The results show notable deviations of behavior from theory. In the first battle of the tug-of-war, subjects exert fewer resources, while in the follow-up battles, they exert more resources than predicted. Also, contrary to the theoretical prediction, resource expenditures tend to increase in the duration of the tug-of-war. Finally, extending the margin necessary to win the tug-of-war causes more discouragement than either a reduction in the prize or greater impatience despite all three having the same expected effect. Potential behavioral explanations for these findings are also discussed.
    Keywords: tug-of-war, all-pay auction, multi-stage contest, laboratory experiment
    JEL: C91 D72 D74
    Date: 2015–09–10
  3. By: Spiro Bougheas; Jeroen Nieboer; Martin Sefton
    Abstract: We report a controlled laboratory experiment examining risk-taking and information aggregation in groups facing a common risk. The experiment allows us to examine how subjects respond to new information, in the form of both privately observed signals and signals reported from others. We find that a considerable number of subjects exhibit ‘reverse confirmation bias’: they place less weight on information from others that agrees with their private signal and more weight on conflicting information. We also find a striking degree of consensus when subjects make decisions on behalf of the group under a random dictatorship procedure. Reverse confirmation bias and the incidence of consensus are considerably reduced when group members can share signals but not communicate.
    Keywords: Group behaviour; teams; decision making; risk; experiment
    JEL: C91 C92 D71 D80
    Date: 2015
  4. By: Siegfried Berninghaus (Karlsruhe Institute of Technology); Stephan Schosser (Faculty of Economics and Management, Otto-von-Guericke University Magdeburg); Bodo Vogt (Faculty of Management, Economics and Social Sciences, University of Cologne)
    Abstract: At present, in the domain of simultaneous action selection and network formation games, game-theoretic behavior and experimental observations are not consistent. While theory typically predicts inefficient outcomes for (anti- )co-ordination games, experiments show that subjects tend to play efficient (non-Nash) strategy profiles. One reason for this discrepancy is the tendency to model corresponding games as one-shot and derive predictions. In this paper, we calculate the equilibria for a finitely repeated version of the Hawk-Dove game with endogenous network formation and show that the repetition leads to additional sub-game perfect equilibria; namely, the efficient strategy profiles played by human subjects. However, efficiency crucially depends on the design of the game. This paper theoretically demonstrates that, although technically feasible, the efficient profiles are not sub-game perfect equilibria if actions are fixed after an initial period. We confirm this result using an experimental study that demonstrates how payoffs are higher if actions are never fixed.
    Keywords: Network; Hawk-Dove; Game theory; Behavioral experiment; Finitely re- peated game
    JEL: D85 C72 C73 C92
    Date: 2015–07
  5. By: Christian Vossler; Michael McKee
    Abstract: Using a framed field experiment with working adults and deliberate tax framing, this study reports on the effects of tax withholding on subsequent individual tax reporting behavior. We find interesting behavioral asymmetries related to tax withholding position, in particular that tax underreporting is increasing in the level of expected as well as unanticipated tax underwithholding, but is largely invariant to the level of tax over-withholding and unexpected decreases in liability. Two information initiatives we explore – group compliance information and information related to fiscal exchange – serve to affect tax reporting in part through its influence on withholding. A third information initiative – a service that resolves uncertainty over tax liability – decreases the level of evasion by over twice as much for those who have underversus over-withheld. Using information from an extensive taxpayer questionnaire, we find several interesting associations between taxpayer characteristics and experimental tax reporting behavior. Key Words: tax withholding; tax information services; social norms; tax reporting and enforcement; experimental methods; framed field experiment
    JEL: H21 H26 C91 C92
    Date: 2015
  6. By: Gerhards, Leonie (Department of Economics and Business Economics, Aarhus University); Gravert, Christina (Department of Economics, School of Business, Economics and Law, Göteborg University)
    Abstract: Perseverance to accomplish long-term goals, also know as grit, is a crucial determinant for success in life. In the present study we introduce an innovative laboratory design to elicit grit in an incentivized and controlled way. Subjects work on a computerized task to solve anagrams. By observing their decision not to shirk, we measure their grittiness experimentally. We find that the original questionnaire measure of grit developed by Duckworth et al. (2007) is significantly correlated with our new experimental measure - even when controlling for ability and a questionnaire measure of self-control. Moreover, subjects' earnings increase in their experimentally elicited grit.
    Keywords: Grit; perseverance; laboratory experiment; real-effort task
    JEL: C91 D03 J24 M50
    Date: 2015–10
  7. By: Noemi Pace (Department of Economics, University Of Venice Cà Foscari); Daniela Di Cagno (Luiss Guido Carli, Rome); Arianna Galliera (Luiss Guido Carli, Rome); Werner Güth (Luiss Guido Carli, Rome, Frankfurt School of Finance and Management, Max Planck Institute, Bonn, Germany); Luca Panaccione (DEDI and CEIS, University of Rome Tor Vergata)
    Abstract: This paper focuses on a bargaining experiment in which the privately informed seller of a company sends a value message to the uninformed potential buyer who then proposes a price for acquiring the company. Participants are constantly in the role of either seller or buyer and interact over 30 rounds with randomly changing partners in the other role. We test how overstating the value of the company, underpricing the received value message and acceptance of price offers are affected by experience and gender (constellation). Like in our companion paper on single play (Di Cagno et al. 2015) we control via treatments for awareness of gender (constellation). One main hypothesis is that gender (constellation) matters but that the effects become weaker with more experience and that the main experience effects apply across gender (constellations).
    Keywords: bargaining, cheap-talk, experience effect, experiment, gender, winner’s curse.
    JEL: C78 C91
    Date: 2015
  8. By: Francesco Drago (Università di Napoli Federico II and CSEF); Friederike Mengel (University of Essex and Maastricht University); Christian Traxler (Hertie School of Governance, Berlin, Max Planck Institute for Research on Collective Goods and CESifo)
    Abstract: This paper studies the spread of compliance behavior in neighborhood networks involving over 500,000 households in Austria. We exploit random variation from a field experiment which varied the content of mailings sent to potential evaders of TV license fees. Our data reveal a strong treatment spillover: ‘untreated’ households, who were not part of the experimental sample, are more likely to switch from evasion to compliance in response to the mailings received by their network neighbors. We analyze the spillover within a model of communication in networks based on DeGroot (1974). Consistent with the model, we find that (i) the spillover increases with the treated households’ eigenvector centrality and that (ii) local concentration of equally treated households produces a lower spillover. These findings carry important implications for enforcement policies.
    Keywords: neighborhood networks; social learning; spillover; evasion; field experiment.
    JEL: D8 H26 Z13
    Date: 2015–10–26
  9. By: Chen, Zhuoqiong; Ong, David; Sheremeta, Roman
    Abstract: We study how salient group identity, created through competition between students from different universities, as well as differences in the value of winning impact competitive behavior. Our experiment employs a simple all-pay auction within and between two university subject pools. We find that when competing against their peers, students within the lower tier university bid more aggressively than students within the top-tier university. Also, students from the lower tier university, in particular women, bid more aggressively when competing against students from the top-tier university. These findings, interpreted through a theoretical model incorporating both group identity and differential value of winning, suggest that students at the lower tier university have a stronger group identity as well as higher desire to win.
    Keywords: experiments, all-pay auction, competitiveness, group identity
    JEL: C91 D03 J7 Z13
    Date: 2015–10–21
  10. By: Ece Yagman (SALDRU, the School of Economics, University of Cape Town); Malcolm Keswell (SALDRU, School of Economics, University of Cape Town)
    Abstract: We investigate discrimination according to accent and race on trust behaviour. Proposers were randomly paired with responders of the same/different race, and asked to play the trust game after looking at a photograph and hearing a 10 second audio clip of the responders reading a standardised script in English. This allows us to check for within and across-group favouritism in both race and accentedness. We find that accentedness is a statistically significant predictor of trust and is strongly non-linear in the race of the paired subjects for males but not for females. In the case of males, offers decrease by 11.3% if the responder has a mother-tongue English accent and does not share the same race as the proposer, but increases by about 6.6% if there is racial similarity. This effect is especially pronounced for Black males who are paired with other Black males: offers are 19.5% higher if responders have a mother-tongue English accent. By contrast, females in general seem less sensitive to the signal package. These large gender differences are not because men behave any more strategically than women.
    Keywords: Experiments, Trust, Accents, Discrimination, Race
    JEL: C91 D03 J15
    Date: 2015
  11. By: Morone, Andrea; Nuzzo, Simone
    Abstract: We investigate traders’ behaviour in an experimental asset market where uninformed agents cannot be sure about the presence of insiders. In this framework we compare two trading institutions: the continuous double auction and the call market. The purpose of this comparison is to test which of the two trading mechanisms performs better in disseminating the information and in promoting a convergence towards the efficient equilibrium price. Furthermore, we aim to determine which of the two trading institutions is more likely to promote a higher level of informational market efficiency. In a framework where the presence of insiders is neither certain nor common knowledge, inspired by Plott and Sunder (1982) and Camerer and Weigelt (1991), we first test whether a discrete time mechanism of trading, like the call market, might be able to prevent the occurrence of information mirages and promote a greater level of efficiency when no inside information is in the market. Second, we also compare the efficiency of the two trading institutions during periods when insiders are present in the market.
    Keywords: Experimental Markets; Market Efficiency; Information Mirages; Trading Institutions.
    JEL: C91 D02 G14
    Date: 2015–10
  12. By: Irfanoglu, Zeynep; Mago, Shakun; Sheremeta, Roman
    Abstract: Sequential contests are predicted to induce lower expenditure than simultaneous contests. This prediction is a result of a “New Hampshire Effect” – a strategic advantage created by the winner of the first battle. Contrary to this prediction, however, our laboratory study of the three-battle contests shows that sequential contests generate significantly higher expenditure than simultaneous contests. In case of sequential contests, we observe significant over-expenditure in all three battles and find no evidence of the “New Hampshire Effect.” Despite the strategic advantage, winners of the first battle make similar expenditures in the second battle as losers of the first battle. Moreover, instead of decreasing, subjects increase their expenditure in the second battle relative to the first battle. In case of simultaneous contests, subjects do not employ a uniform expenditure strategy and instead use a “guerilla warfare” strategy by focusing on only two of the three battles. We propose several explanations for these findings and discuss some implications.
    Keywords: election, sequential contests, simultaneous contests, experiments
    JEL: C72 C73 C91 D72
    Date: 2015–08–13
  13. By: Stephan Schosser (Faculty of Economics and Management, Otto-von-Guericke University Magdeburg)
    Abstract: When talking about possible bargaining results participants in the Nash bargaining game mainly use fairness norms to support their favored outcome. According to theory a variety of different, fair solutions exists from which the participants can choose. In this paper, we experimentally investigate Nash bargaining with a previous opportunity to chat about the bargaining outcome. We find that playing a dictator game prior to the Nash bargaining game establishes – without any additional communication – a fairness norm, the participants resort to. However, if nothing is played prior to the Nash bargaining game, participants discuss longer about what to play. In addition, we find that deviations in favor of one participant occur the longer preplay communication lasts.
    Keywords: bargaining game, dictator game, norms, experimental economics
    JEL: C7 C9
    Date: 2015–10
  14. By: Amini, Makan (Advent International); Ekström, Mathias (NHH Norwegian School of Economics); Ellingsen, Tore (Stockholm School of Economics); Johannesson, Magnus (Stockholm School of Economics); Strömsten, Fredrik (McKinsey & Company)
    Abstract: Failure to express minority views may distort the behavior of company boards, committees, juries, and other decision-making bodies. Devising a new experimental procedure to measure such conformity in a judgment task, we compare the degree of conformity in groups with varying gender composition. Overall, our experiments offer little evidence that gender composition affects expression of minority views. A robust finding is that a subject’s lack of ability predicts both a true propensity to accept others judgment (informational social influence) and a propensity to agree despite private doubt (normative social influence). Thus, as an antidote to conformity in our experiments, high individual ability seems more effective than group diversity.
    Keywords: Conformity; Gender Differences; Group Composition; Skill
    JEL: C90 D02 D71 D83 J16
    Date: 2015–10–20
  15. By: Alexander L. Brown (Texas A&M University, Department of Economics); Ajalavat Viriyavipart (Texas A&M University, Department of Economics); Xiaoyuan Wang (University of Electronic Science and Technology of China, School of Management and Economics)
    Abstract: Recent theoretical research indicates that search deterrence strategies are generally optimal for sellers in consumer goods markets. Yet search deterrence is not always employed in such markets. To understand this incongruity, we develop an experimental market where profit-maximizing strategy dictates sellers should exercise one form of search deterrence, exploding offers. We find that buyers over-reject exploding offers relative to optimal. Sellers underutilize exploding offers relative to optimal play, even conditional on buyer over-rejection. This tendency dissipates when sellers make offers to computerized buyers, suggesting their persistent behavior with human buyers may be due to a preference rather than a miscalculation.
    Keywords: exploding offer, search deterrence, experimental economics, quantal response equilibrium
    JEL: C91 D21 L10 M31
    Date: 2014–10–06
  16. By: Stephen Taylor (Department of Economics and Department of Basic Education); Patricia Watson (Department of Basic Education)
    Abstract: Most international literature on the impact of textbooks on educational achievement suggests that this is a relatively cost-effective intervention. However, recent experimental evidence from developing countries has called this into question, suggesting that resources alone are unlikely to impact on performance and that changes in school organisation, pedagogical methods or incentives facing teachers are more effective. South African studies, using observational data, typically show weak associations between achievement and additional resources, though in some studies textbooks emerge as an exception. Some argue that school management is a key mediating variable. This paper evaluates the impact of providing study guides to pupils shortly before their secondary school leaving examination (the “matric” exam). From a sampling frame of 318 schools in the Mpumalanga province, 79 schools were randomly selected to receive study guides, leaving 239 control schools. These study guides were developed by the National Department of Basic Education and distributed to treatment schools for four subjects – accounting, economics, geography and life sciences – resulting in four distinct treatments per school. The impact of the study guides was estimated using matric results from 2011 (baseline) and 2012 (endline). The accounting and economics guides did not have a significant impact on performance. However, the geography and life sciences guides improved scores in those subjects by approximately two percentage points. Treatment heterogeneity was apparent for geography where students in better-performing schools gained more from the guides than students in low-performing schools. This may relate to other studies suggesting that additional school resources matter conditionally upon overall school functionality, particularly management. A simulation indicated that distributing the geography and life science at scale could increase the overall matric pass rate by roughly one percentage point. A cost-benefit analysis calculating the standard deviations of impact on test scores per $100 spent indicates that this intervention is amongst the most cost-effective of educational interventions internationally that have been tested using randomised experiments. Possible reasons why the guides were effective in two subjects but not the other two are discussed.
    Keywords: South Africa, education, educational achievement, study guides, Randomised Controlled Trial
    JEL: I20 I21 I28
    Date: 2015
  17. By: Sheremeta, Roman
    Abstract: Group contests are ubiquitous. Some examples include warfare between countries, competition between political parties, team-incentives within firms, group sports, and rent-seeking. In order to succeed, members of the same group have incentives to cooperate with each other by expending individual efforts. However, since effort is costly, each member also has an incentive to abstain from expending any effort and instead free-ride on the efforts of other members. Contest theory shows that the intensity of competition between groups and the amount of free-riding within groups depend on the group size, sharing rule, group impact function, contest success function, and heterogeneity of players. We review experimental studies testing these theoretical predictions. Almost all studies of behavior in group contests find significant over-expenditure of effort relative to the theory. We discuss potential explanations for such over-expenditure, including the utility of winning, bounded rationality, relative payoff maximization, parochial altruism, and social identity. Despite over-expenditure, most studies find support for the comparative statics predictions of the theory (with the exception of the “group size paradox”). Finally, studies show that there are effective mechanisms that can promote within-group cooperation and conflict resolution mechanisms that can de-escalate and potentially eliminate between-group conflict.
    Keywords: groups, contests, experiments
    JEL: C7 C9 D7 H4 J4 K4 L2 M5
    Date: 2015–10–29
  18. By: Bosman, Ronald; Kräussl, Roman; Mirgorodskaya, Elizaveta
    Abstract: We investigate the effect of the tone of news on investor stock price expectations and beliefs. In an experimental study we ask subjects to estimate a future stock price for twelve real listed companies. As additional information we provide them with historical stock prices and extracts from real newspaper articles. We propose a way to manipulate the tone of news extracts without distorting its content. Subjects in different treatment groups read news items that are written either in positive or negative tone for each stock. We find that subjects tend to predict a significantly higher (lower) return for stocks after reading positive (negative) tone news. The effect is especially pronounced for stocks with poor past performance. Subjects are more likely to be optimistic (pessimistic) about the economy and to buy (sell) stocks after reading positive (negative) than negative (positive) tone news. Our results show that the news media might affect not only how investors perceive information, but also what they do in response to it.
    Keywords: Tone,News,Framing Effect,Price Expectations,Investor Sentiment,Investment Decisions,Experiment
    JEL: D83 G02 G11
    Date: 2015
  19. By: Alexander L. Brown (Texas A&M University, Department of Economics); Rodrigo A. Velez (Texas A&M University, Department of Economics)
    Abstract: In experimental partnership dissolution problems with complete information, the divide-and-choose mechanism is significantly superior to the winner's-bid-auction. The performance of divide-and-choose is mainly affected by reciprocity issues and not by bounded rationality. The performance of the winner's-bid-auction is significantly affected by bounded rationality. Contrary to theoretical predicitions divide-and-choose exhibits no first-mover bias.
    Keywords: experimental economics; no-envy; divide-and-choose; winner's-bid auction; behavioral mechanism design
    JEL: C91 D63 C72
    Date: 2014–09–18
  20. By: Astrid, Gamba; Tobias, Regner
    Abstract: We study experimentally whether heterogeneity of behavior in the Centipede game can be interpreted as the result of a learning process of individuals with different preference types (more and less pro-social) and coarse information regarding the opponent's past behavior. We manipulate the quality of information feedbacks provided after each play. If subjects rely only on their personal database, long run behavior resembles a Self-confirming equilibrium whereby less pro-social types take at earlier nodes due to prediction errors. Aggregate information release decreases heterogeneity of behavior by increasing the passing rates of pro-selfs and play moves towards Bayesian Nash equilibrium.
    Keywords: Social preferences, Learning, Self-confirming equilibrium, Experiment
    JEL: C71 C73 C91 D83
  21. By: Gelder, Alan; Kovenock, Dan; Sheremeta, Roman
    Abstract: Despite the wide occurrence of ties in a variety of contest settings, the strategic interaction that arises when ties are treated as viable outcomes has received little attention. Building on recent theoretical work, we experimentally examine an extension of the canonical two-player all-pay auction in which a tie occurs unless one player’s bid exceeds the other’s by some critical threshold. In the event of a tie, each player receives an identical fraction of the prize. For the case where players receive one-half of the prize when they tie, we find that players’ expenditures are non-monotonic in the threshold required for victory. Moreover, for certain positive thresholds, expenditures may even be higher than under the standard all-pay auction. We also find that decreasing the fraction of the prize players receive for tying may either increase or decrease total expenditures. In accordance with theory, the effect depends upon the threshold.
    Keywords: All-pay auction, contest, tie, draw, bid differential, experiment
    JEL: C72 C92 D44 D72 D74
    Date: 2015–10–29
  22. By: Charles F. Manski (Institute for Fiscal Studies and Northwestern University); Aleksey Tetenov (Institute for Fiscal Studies and Collegio Carlo Alberto)
    Abstract: Medical research has evolved conventions for choosing sample size in randomized clinical trials that rest on the theory of hypothesis testing. Bayesians have argued that trials should be designed to maximize subjective expected utility in settings of clinical interest. This perspective is compelling given a credible prior distribution on treatment response, but Bayesians have struggled to provide guidance on specification of priors. We use the frequentist statistical decision theory of Wald (1950) to study design of trials under ambiguity. We show that e-optimal rules exist when trials have large enough sample size. An e-optimal rule has expected welfare within e of the welfare of the best treatment in every state of nature. Equivalently, it has maximum regret no larger than e. We consider trials that draw predetermined numbers of subjects at random within groups stratified by covariates and treatments. The principal analytical findings are simple sufficient conditions on sample sizes that ensure existence of e-optimal treatment rules when outcomes are bounded. These conditions are obtained by application of Hoeffding (1963) large deviations inequalities to evaluate the performance of empirical success rules.
    Date: 2015–09
  23. By: Eric Guerci (GREDEG - Groupe de Recherche en Droit, Economie et Gestion - CNRS - UNS - Université Nice Sophia Antipolis); Nobuyuki Hanaki (GREDEG - Groupe de Recherche en Droit, Economie et Gestion - CNRS - UNS - Université Nice Sophia Antipolis); Naoki Watanabe (Faculty of Engineering, Information and Systems, University of Tsukuba - University of Tsukuba)
    Abstract: By employing binary committee choice problems, this paper investigates how varying or eliminating feedback about payoffs affects: (1) subjects' learning about the underlying relationship between their nominal voting weights and their expected payoffs in weighted voting games; and (2) the transfer of acquired learning from one committee choice problem to a similar but different problem. In the experiment, subjects choose to join one of two committees (weighted voting games) and obtain a payoff stochastically determined by a voting theory. We found that: (i) subjects learned to choose the committee that generates a higher expected payoff even without feedback about the payoffs they received; and (ii) there was statistically significant evidence of ``meaningful learning'' (transfer of learning) only for the treatment with no payoff-related feedback. This finding calls for re-thinking existing models of learning to incorporate some type of introspection.
    Keywords: two-armed bandit problem,learning, voting game, experiment
    Date: 2015–10–15
  24. By: Marios Michaelides; Peter Mueser
    Abstract: This paper examines four experimental design U.S. reemployment programs targeting Unemployment Insurance (UI) recipients during the Great Recession. Results show that, regardless of their specific requirements, reemployment programs led to shorter UI spells, UI savings that exceeded program costs, and improved employment rates and earnings. Program success was partly attributable to threat effects, occurring because some participants exited UI early in their claims to avoid program participation. Also, programs that conducted eligibility reviews were effective in reducing UI fraud, pushing job-ready or ineligible participants to exit UI and return to employment sooner. In addition to these initial effects, our results suggest that program services may have substantial direct value. The program that combined the eligibility review with mandatory job-counselling services produced greater effects than the other programs, and our analyses suggest that this is because services helped participants improve their job search. These findings provide important policy guidance on the kinds of programs that are effective and evidence that program benefits occur even during an economic downturn.
    Keywords: Great Recession, job search services, eligibility review, unemployment, Unemployment Insurance, program evaluation.
    Date: 2015–10
  25. By: Louis Lévy-Garboua (Centre d'Economie de la Sorbonne and CIRANO - Montreal); Muniza Askari (Centre d'Economie de la Sorbonne); Marco Gazel (Centre d'Economie de la Sorbonne - Paris School of Economics)
    Abstract: We design a double-or-quits game to compare the speed of learning one's specific ability with the speed of rising confidence as the task gets increasingly difficult. We find that people on average learn to be overconfident faster than they learn their true ability and we present a simple Bayesian model of confidence which integrates these facts. We show that limited discrimination of objective differences, myopia, and uncertainty about one's true ability to perform a task in isolation can be responsible for large and robust confidence biases, namely the hard-easy effect, the Dunning-Kruger effect, conservative learning from experience and the overprecision phenomenon (without underprecision) if subjects act as Bayesian learners. Moreover, these biases are likely to persist since the Bayesian aggregation of past information consolidates the accumulation of errors and the perception of contrarian illusory signals generates conservatism and under-reaction to events. Taken together, these two features may explain why intuitive Bayesians make systematically wrong predictions of their own performance
    Keywords: Confidence biases; Bayesian learning; double or quits experimental game; hard-easy effect; Dunning-Kruger effect; illusory signals
    JEL: C11 C91 D03
    Date: 2015–10
  26. By: Fricke, Hans; Frölich, Markus; Huber, Martin; Lechner, Michael
    Abstract: This paper proposes a nonparametric method for evaluating treatment effects in the presence of both treatment endogeneity and attrition/non-response bias, using two instrumental variables. Making use of a discrete instrument for the treatment and a continuous instrument for nonresponse/attrition, we identify the average treatment effect on compliers as well as the total population and suggest non- and semiparametric estimators. We apply the latter to a randomized experiment at a Swiss University in order to estimate the effect of gym training on students› selfassessed health. The treatment (gym training) and attrition are instrumented by randomized cash incentives paid out conditional on gym visits and by a cash lottery for participating in the follow-up survey, respectively.
    Keywords: local average treatment effect; attrition; endogeneity; weighting; instrument; experiment
    JEL: C14 C21 C23 C24 C26
    Date: 2015–10–22

General information on the NEP project can be found at For comments please write to the director of NEP, Marco Novarese at <>. Put “NEP” in the subject, otherwise your mail may be rejected.
NEP’s infrastructure is sponsored by the School of Economics and Finance of Massey University in New Zealand.