INVESTMENT CYCLES,STRATEGIC DELAY,AND SELF‐REVERSING CASCADES*

We study investment cycles in a social learning model, where investment returns fluctuate according to a Markov process. In our Waiting Game, agents observe the investment history and a private signal correlated with the current period’s investment return. Agents then decide whether to invest immediately or to delay their decision to later in the period. Cascades in which everyone invests or no one invests eventually reverse themselves. As compared to the No‐Waiting Game with no opportunity for delay, the Waiting Game has shorter investment cascades, longer recessions, shorter booms, more underinvestment, and less overinvestment.     

The effect of endogenous timing on coordination under asymmetric information: An experimental study

This paper investigates the role of endogenous timing of decisions on coordination under asymmetric information. In the equilibrium of a global coordination game, where players choose the timing of their decision, a player who has sufficiently high beliefs about the state of the economy undertakes an investment without delay. This decision (potentially) triggers an investment by the other player whose beliefs would have led to inaction otherwise. Endogenous timing has two distinct effects on coordination: a learning effect (early decisions reveal information) and a complementarity effect (early decisions eliminate strategic uncertainty for late movers). The experiments that we conduct to test these theoretical results show that the learning effect of timing has more impact on the subjects' behavior than the complementarity effect. We also observe that subjects' welfare improves significantly under endogenous timing.     

Catching-Up Investment without Regulatory Commitment

This paper analyses a regulated firms incentives to undertake catching-up investments when the firm has private information about the initial technology and the regulator is unable to commit himself to incentive contracts prior to the firms investment decision. In the absence of commitment power, the firm takes into account that the investment decision may serve as a signal to the regulator about the firms initial technology. Any pure strategy equilibrium of the signaling game is shown to be pooling in the sense that the efficient type mimics the inefficient type by investing. By not following this strategy, the efficient type reveals its efficiency to the regulator, who responds by inducing the firm to produce without rents. Restricting attention to undefeated pooling equilibria, the level of investment is shown to be lower than the first-best level.     

Dynamic costly state verification

Summary. I study a model of dynamic risk sharing with costly state verification (CSV). In the model, a risk neutral agent enters an infinitely repeated relationship with a risk averse agent. In each period, the risk averse agent receives a random income which is observed only by himself, unless the risk neutral agent engages in costly monitoring. I provide a set of characterizations for the optimal contract, and I show that CSV has interesting effects on the long run distribution of the agents expected utilities.Received: 19 February 2003, Revised: 11 February 2004, JEL Classification Numbers: D8.This research was initially joint work with Bruce Smith. I thank Bruce for the inspiration. I thank Fernando Alvarez, Narayana Kocherlakota, Steve Spear, and an anonymous referee for helpful comments. I also thank seminar participants at Carnegie Mellon University, Federal Reserve Banks of Minneapolis, Federal Reserve bank of Richmond, Duke University, SUNY Buffalo, University of Rochester, University of Pittsburgh, University of Western Ontario, the 1998 Econometric Society winter meeting, the 1999 SED meeting, and the 1999 SITE workshop at Stanford University for discussions.     

Influence and inefficiency in the internal capital market

I model inefficient resource allocations in M-form organizations due to influence activities by division managers that skew capital budgets in their favor. Corporate headquarters receives two types of signals about investment opportunities: private signals that can be distorted by managers, and public signals that are undistorted but noisy. Headquarters faces a tradeoff between the cost of attaining an accurate private signal and the value of the information the signal provides. In contrast to existing models of “socialism” in internal capital markets, I show that investment sensitivity to Tobin's Q is higher than first-best in firms where division managers hold equity (a result consistent with evidence presented in Scharfstein, 1998). When managers face high private costs from distorting information (equity holdings), headquarters may commit to investment contracts that place “too little” weight on private signals and “too much” weight on public signals (i.e. Q). This result has implications for managers in the design of capital budgeting processes and incentive compensation systems.     

Agency conflicts in the presence of random private benefits from project implementation

We study a contracting problem where a principal delegates the decision to implement a “project” to an agent who obtains private information about the value of the project before making the implementation decision. Moral hazard arises because the agent gets private random non-contractible benefits, or incurs private random non-contractible costs, if the project is implemented. This contracting problem is pervasive, when “project” and “benefits” are interpreted broadly.     

On the welfare effects of allowing unlimited renegotiation in agency relationships

We study renegotiation in an agency setting where the number of offers and accept/reject decisions parties can make is potentially unlimited. Thus any contract, either on or off the equilibrium path, may be subject to possible renegotiation. We first show that the principal will not be able to gain complete access to the agent’s private information with unlimited renegotiation, unlike when the potential number of renegotiations is finite. Rather the agent either employ a randomized reporting strategy or do not to report at all. We then identify conditions under which expected allocations are most efficient with the contract that induces no agent communication. More significantly, by doing so we also identify conditions under which the parties are made strictly worse off by committing to end renegotiation after a fixed number of rounds. We thank workshop participants at University of California, Irvine, University of Chicago, Duke University, Hong Kong University of Science and Technology, University of Houston, London School of Economics, University of Minnesota, Ohio State University, the Carnegie Mellon Accounting Research Conference, the editor, and an anonymous referee for their helpful comments and suggestions.     

Collusion through mediated communication in repeated games with imperfect private monitoring

Summary. This paper studies repeated games with imperfect private monitoring when there exists a third-party mediator who coordinates play by giving non-binding instructions to players on which action to take and by collecting their private information. The paper presents a Nash-threat folk theorem for a communication equilibrium based on such mediation when monitoring is jointly -perfect in the sense that every player is almost perfectly monitored collectively by other players.JEL Classification Numbers: C72, D82.I am very grateful to Mark Armstrong, V. Bhaskar, and Michihiro Kandori for helpful comments. Part of this research was conducted while I was visiting the University College London. Their hospitality is gratefully acknowledged.     

Semiparametric estimation and testing in a model of environmental regulation with adverse selection

We consider a principal–agent model of environmental regulation with adverse selection, where firms are regulated through contracts. We show how the model allows to recover information on structural cost parameters. We use a semiparametric method to estimate consistently such parameters without specifying the distribution of the agents private information. We also show how to check for the specification of the econometric model, as well as auxiliary parametric assumptions, by means of specification tests based on nonparametric estimation. Results are used to discuss a selection of economic issues related to environmental regulation.First version received: August 2002/Final version received: November 2003Previous versions of this work were presented at Seminaire INRA-IDEI, Toulouse, the conference on Applications of semiparametric methods for micro-data, Tilburg, and seminars at UC Berkeley, University of Wisconsin–Madison, CREST, and INRA-LEA. We are indebted to participants, and especially Michael Visser, for their comments. We thank two referees for helpful comments. Financial support from the Conseil Régional Midi-Pyrénées and from INRA is gratefully acknowledged.     

Professional Investors as Naturalistic Decision Makers: Evidence and Market Implications

This paper is the first to examine investment decision-making from a naturalistic decision perspective. Naturalistic decision procedures tend to be used by experts making decisions in complex, ill-structured, and indeterminate situations. Survey results indicate that investment professionals, like other naturalistic decision-makers, rely heavily on mental imagery, reasoning by analogy, and decision procedures that become more intuitive as complexity increases. Also, they are "satisficers," not optimizers. In other words, their primary aim is to make an acceptable choice; finding the best choice is not necessary. Clinical training is recommended to improve the performance of investment professionals. In particular, investment professionals may benefit from training similar to that given to medical professionals wherein emphasis is placed to extensive repetitive exposure to "real world" decisions complete with plenty of immediate and unambiguous feedback.     

Information Cascades in the Labor Market

A model of herding behavior in the labor market is presented where employers receive signals with limited precision about the workers types, and can observe previous employers decisions. Both the employer and the worker can influence the signal probabilities. In particular, the employer tries to increase the precision of the signal about the workers type whereas the worker wants to get a good signal, independent of her type. In a two-period model, we derive conditions for an equilibrium in which only down-cascades occur, i.e., the second employer does not hire a worker with a bad history even if he receives a favorable private signal about the workers type, but he follows his own signal if the workers history is good.     

Dynamic coordination games

Summary Gains from coordination provide incentives for delay. In this paper, the extent of delay is studied in a dynamic,N-person, coordination game. There is no social gain from delay, so an equilibrium with delay is always inefficient. For fixedN, there is no coordination failure when the period length is short: all equilibrium outcomes converge to the Pareto efficient outcome as the period length converges to zero. On the other hand, holding period length fixed, there exist equilibria in which delay is proportional toN, for arbitrarily large values ofN. In addition, it can be shown that the possibility of delay depends on the timing of strategic complementarities. However, under certain conditions, delay is shown to be a robust phenomenon, in the sense that well-behaved equilibria exhibit infinite delay forN sufficiently large.This paper grew out of discussions with Christophe Chamley. While writing it I benefited from discussions with Ken Binmore, Russell Cooper, Bob Rosenthal and Michael Manove. Joe Farrell, Drew Fudenberg, Martin Hellwig and Sawoong Kang made very useful comments on an earlier version that led to substantial improvements. Helpful comments were also made by seminar participants at the London School of Economics, the SUNY at Stoney Brook, the NBER Summer Institute, Northwestern University, and the University of Chicago. I would like to thank Nick Yannelis and an anonymous referee for their editorial advice. Financial support for this research was provided by the National Science Foundation under Grant No. SES 9196061.     

The value of information in a principal-agent model with moral hazard: The ex post contracting case

In a principal-agent model with moral hazard, a signal about the principal?s technology — the stochastic mapping from the agent?s action to the outcome — is observed before the contract is offered. The signal is either uninformative (null information), informative and observed only by the principal (private information), or also observed by the agent (public information). We show that, from an ex ante standpoint (before the signal is observed): (i) the agent prefers private to both null and public information; (ii) the principal sometimes prefers null to both private and public information; and (iii) when the principal prefers public to null information, she prefers public to private information, whereas the agent prefers private to public information. In this last situation, we also show that (iv) for any separating equilibrium with private information, there exists a contract with public information that both strictly prefer.     

Changing One's Mind when the Facts Change: Incentives of Experts and the Design of Reporting Protocols

Experts often collect and report information over time. What reporting protocol elicits the most information? Here, a principal receives reports sequentially from an agent with privately known ability, who observes two signals about the state of the world. The signals differ in initial quality and, unlike previous work, differ in quality improvement . The paper finds that "mind changes" (inconsistent reports) can signal talent if a smart agent improves faster. Also, sequential reports dominate when the principal's decision is very sensitive to information; a single report dominates if the mediocre agent's signals improve faster or the agent is likely mediocre.
When the facts change, I change my mind. What do you do, sir? — John M. Keynes ¹     

The Right Choice at the Right Time: A Herding Experiment in Endogenous Time

Herding describes the phenomenon in decision-making where an economic agent disregards his own private information to follow the actions of his predecessors as in Banerjee (1992). With later decision-makers simply copying earlier decisions their private information cannot be inferred by other decision-makers and will be forever lost. There is some experimental evidence on simple sequential herding of this type in the literature, notably Anderson and Holt (1997). This paper differs by allowing subjects to delay their decision-making in order to benefit from observing others' actions as in more recent herding models such as Chamley and Gale (1994). The results in this paper suggest that subjects will indeed delay when their private information is not sufficiently strong. Despite this ability to wait, as predicted in the theoretical literature, cascades remained ubiquitous and more worrying still, reverse-cascades occurred in which incorrect decisions made by early decision-makers produced informational cascades on the wrong action. In an alternative design, informing subjects that they had made incorrect choices only made matters worse as subjects moved further away from rational behavior.     

Equilibria in a multi-object uniform price sealed bid auction with multi-unit demands

Summary In many existing markets demanders wish to buy more than one unit from a group of identical units of a commodity. Often, the units are sold simultaneously by auction. The vast majority of literature pertaining to the economics of auctions, however, considers environments in which demanders buy at most one object. In this paper we derive necessary and sufficient conditions for a set of bidding strategies to be a symmetric monotone Bayes-Nash equilibrium to a uniform price sealed bid auction using the first rejected bid pricing rule in an independent private values environment with two-unit demands. In any symmetric monotone Bayes-Nash equilibrium, all bidders submit one bid equal to their higher valuation and one bid lower than their lower valuation. We characterize the equilibrium and derive the exact amount of underrevelation in the lower bid.This article is based on chapter 2 of the author's Ph.D. thesis. I would like to acknowledge the financial support of the Clarence Hicks Memorial Fellowship, the California Institute of Technology and the Jet Propulsion Laboratory. I also thank D. Roderick Kiewiet, Richard McKelvey, Charles Plott, David Porter, Annemieke Tromp, participants in the Tinbergen Institute Seminar Series, an anonymous referee and especially John Ledyard for countless enlightening comments.     

Income Distribution, Redistributive Politics, and Economic Growth

This article studies the political economy of inequality and growth by combining the political economy approach with an imperfect capital market assumption. In the present model, there emerges a class of individuals whose members do not invest privately beyond the state-financed schooling, due to their initial wealth constraint. We show that inequality affects private investment not only through the political effect, which relates inequality to private investment negatively, but also through what we call the threshold effect, which associates inequality to private investment positively. In general, private investment and inequality do not show a monotone negative relationship.     

Expropriation risk,investment decisions and economic sectors

We build a Real Options model to assess the importance of private provision and the impact of expropriation risk on investment timing, business values, governmental costs and social welfare. We consider two types of businesses (essential and non-essential) and two stages (operating businesses and investment opportunities) and answer questions regarding three main topics: the firm's reaction to expropriation risk, the government drivers to expropriate, and the welfare costs of expropriation. Our results show that responding to expropriation risk the private investor is driven to suboptimal investment decisions. When we endogenize the reputational costs of expropriation, our results show that the decision of the government to expropriate largely depends on the type of business being targeted. In terms of welfare, our results show that expropriation is always associated with a loss.     

Consumption externalities, rental markets and purchase clubs

Summary. A premise of general equilibrium theory is that private goods are rival. Nevertheless, many private goods are shared, e.g., through borrowing, through co-ownership, or simply because one persons consumption affects another persons wellbeing. I analyze consumption externalities from the perspective of club theory, and argue that, provided consumption externalities are limited in scope, they can be internalized through membership fees to groups. Two important applications are to rental markets and purchase clubs, in which members share the goods that they have individually purchased.Received: 2 June 2003, Revised: 8 March 2004, JEL Classification Numbers: D11, D62.This paper was supported by the U.C., Berkeley Committee on Research, and the Institute of Economics, University of Copenhagen. I am grateful to Birgit Grodal for her collaboration on the theory that underlies this paper, and for her helpful and motivating comments about these particular extensions. I also thank Hal Varian, Doug Lichtman, Steve Goldman, Karl Vind, anonymous referees, and members of the Berkeley Microeconomics Seminar for discussion.     

Asymmetric information and survival in financial markets

Summary. In the evolutionary setting for a financial market developed by Blume and Easley (1992), we consider an infinitely repeated version of a model á la Grossman and Stiglitz (1980) with asymmetrically informed traders. Informed traders observe the realisation of a payoff relevant signal before making their portfolio decisions. Uninformed traders do not have direct access to this kind of information, but can partially infer it from market prices. As a counterpart for their privileged information, informed traders pay a per period cost. As a result, information acquisition triggers a trade-off in our setting. We prove that, so long as information is costly, uninformed traders survive.JEL Classification Numbers: D50, D82, G14.I am deeply indebted to Luca Anderlini for his helpful guidance. I also benefited from discussion with Larry Blume, David Easley, Jayasri Dutta, Thorsten Hens, Hamid Sabourian, Klaus Reiner Schenk-Hoppé and Hyun Song Shin. Useful comments came from an anonymous referee and participants to seminars in Barcelona, Bielefeld, Cambridge, Manchester, Oxford, Rotterdam, Venice, Zurich, to the PhD Awards Italian tour in Rome, Naples, Padova and Milan, and to ESEM99 and EEA99 in Santiago de Compostela.