This paper examines experimentally the reputation building role of disclosure in an investment / trust game. It provides experimental evidence in support of sequential equilibrium behavior in a finitely repeated investment / trust game where information asymmetry raises the possibility of voluntary disclosure. I define two regimes, namely disclosure regime and no-disclosure regime and it is only in the disclosure regime that such disclosure of private information is a possibility. I compare investment levels across two regimes and find the startling result that investment is lower in disclosure regime. I find that this lower investment is attributable to the fact that the prior probability with which an investor in the disclosure regime believes that a manager is trustworthy is significantly lower than the prior probability with which an investor in the nodisclosure regime believes that a manager is trustworthy. I introduce a two-stage experimental design to homogenize prior beliefs about managers' trustworthiness and find that after such homogenization, investment is higher in disclosure.

This article analyses the role of information in building reputation in an investment/trust game. The model allows for information asymmetry in a finitely repeated sender -receiver game and solves for sequential equilibrium to show that if there are some trustworthy managers who always disclose their private information and choose to return a fair proportion of the firm's income as dividend to the investor, then a rational manager will mimic such behaviour in an attempt to earn a reputation for being trustworthy. The rational manager will mimic with probability 1 in the early periods of the game. The investor, too, will invest with probability 1 in these periods. However, in the later periods, the rational manager will mimic with a certain probability strictly less than 1. The probability will be such that it will make the investor indifferent between investing and not investing, and he, in turn, will invest with a probability (strictly less than 1) that will make the rational manager indifferent between mimicking and not mimicking; that is, the game will begin with pure-strategy play but will switch to mixed-strategy play. There is one exception, though: when the investor's ex ante beliefs about the manager's trustworthiness are exceptionally high, the game will continue in a pure strategy, and the switch to mixed-strategy play will never occur. Identical results obtain if the manager's choice of whether to share his private information with the investor is replaced by exogenously imposed information sharing.

Dating back at least to Adam Smith (1790), philosophers and researchers expect that people will behave differently when they know their actions are observable to others. We hypothesize that financial reporting reveals managers' actions and leads them to take different actions that are better aligned with investor interests. We posit that the reason why is the activation of our internal mental self-evaluation that Smith refers to as an "Impartial Spectator." We test this hypothesis with an experiment in which we manipulate the availability of a financial report that makes managerial actions transparent. Our evidence shows that financial reporting leads a manager to choose reinvestment and resource sharing actions that are better aligned with investor interests, even in a sparse experimental setting where the investor can impose no cost or confer no reward on the manager. This same effect holds in a setting where the investor can shut down the firm at any point and take a sizable portion of the assets. Our evidence is important because it suggests that part of financial reporting's economic value comes from its enabling moral evaluation by the manager in addition to its traditional contracting function

*This paper is based on my Ph.D. dissertation at University of Minnesota. I am indebted to my dissertation committee for the valuable suggestions and guidance: John Dickhaut (Chair), Beth Allen, Chandra Kanodia and Greg Waymire. I wish to thank Bill Neilsen (editor), an anonymous referee, Regina Anctil, Harry Evans, Christian Leuz, Jack Stecher and Jeroen Suijs for helpful comments and suggestions. I also wish to thank conference participants at

No abstract