[The nature of information regulation depends on the informational efficiency of capital markets (see Beaver 1989, 152-71; Dyckman and Morse 1986, 82-91). Consequently, researchers in accounting and finance have spent considerable effort attempting to measure efficiency. Although this investigation has spanned many research designs and has been applied to many different information signals, empirical tests all suffer from the same basic problem: the benchmark of interest, an informationally efficient market, is unobservable. The asset price that would have prevailed in an efficient market must therefore be modeled, and the test of market efficiency is confounded with a test of the asset-pricing model. Because of this ambiguity, whenever a researcher claims to find an abnormal return based on some information signal another researcher invariably responds that risk was not adequately controlled. For instance, Bernard and Thomas (1989, 1990) present evidence that markets do not adequately adjust to quarterly earnings announcements (i.e., there is a significant post-announcement drift), while Ball et al. (1990) argue that the market adjustment may be correct if the level of risk during the announcement period is adequately controlled for. Unlike naturally occurring markets, the efficiency of a laboratory market can be measured directly by creating another "artificial" economy that is identical to the economy of interest, except that all information is fully disseminated. The price in the artificial economy is the efficient price by definition; it is determined endogenously and without reference to an asset-pricing model. Using this method of measuring a market's efficiency, this study investigates how efficiency is influenced by different information or market structures. Although such an investigation will not resolve the issue of whether naturally occurring markets are efficient, laboratory results can identify features of a market or information structure that aid or impede efficiency. The study compares two information structures that differ by whether there is aggregate certainty in the market; that is, whether the union of all traders' information signals perfectly identifies the value of the risky asset. Previous experimental research in market efficiency has used markets with aggregate certainty. However, many of the difficulties of decision making under uncertainty disappear when the information in the market collectively reveals the asset's payoff. On the other hand, for the experiments conducted here, there are relatively more signals to aggregate in the markets with aggregate certainty. The results show that in markets where different traders have different information signals, the presence of aggregate uncertainty significantly reduces efficiency relative to similar markets with aggregate certainty. However, the results also show that markets are very efficient when some traders have a common but imperfect information signal and other traders are uninformed. In these markets there is aggregate uncertainty but no diversity of information among informed traders. Thus, diversity of informed traders' information and aggregate uncertainty together lead to inefficient markets, but neither treatment by itself causes inefficiency. The study also manipulates the number of traders in the market. It is sometimes argued that markets are efficient because there are a large number of traders whose individual errors average out. However, there is no reason to believe that the asset-pricing relation applies equal weight to each trader's belief, so a central limit result may not hold. The results show that the number of traders has no significant impact on the efficiency of the final prices in a trading period. Within a trading period, however, markets with only a few traders converge to the efficient price much more quickly than do markets with many traders. The results also show that there is a greater diversity of behavior in the markets with many traders. It is possible that this increased diversity increases the number of "noisy" transactions, making it more difficult to infer information from market data. In any investigation of a market's efficiency, different traders must have different information at the time efficiency is being assessed; otherwise the market is efficient by definition. Although accounting disclosures are publicly available they can effectively generate different information signals to different traders. The markets presented here give two examples. In the aggregate certainty treatment, some traders received good news signals and other traders received bad news signals. An example of this type of information system is an economy where different traders having different earnings expectation models. In such an economy the same earnings report can be good news to some traders and bad news to other traders. As long as the "correct" earnings expectation model is unknown, each trader would find the other traders' signals-in this case their forecast errors-informative. In the number-of-traders treatment, some traders receive a signal while other traders do not. An example of this type of information system is an economy where some traders receive accounting disclosures very quickly by subscribing to a wire news while other traders receive the information via third-class mail. Here the uninformed traders would benefit by learning the informed traders' signal.]

Information Allocation, Private information, Information asymmetry, Disclosures

The article investigates how capital markets efficiency is influenced by different information or market structures in the United States. The nature of information regulation depends on the informational efficiency of capital markets. Researchers in accounting and finance have spent considerable effort attempting to measure efficiency. Although this investigation has spanned many research designs and has been applied to many different information signals, empirical tests all suffer from the same basic problem: the benchmark of interest, an informationally efficient market, is unobservable. The asset price that would have prevailed in an efficient market must therefore be modeled, and the test of market efficiency is confounded with a test of the asset-pricing model. Because of this ambiguity, whenever a researcher claims to find an abnormal return based on some information signal another researcher invariably responds that risk was not adequately controlled. The efficiency of a laboratory market can be measured directly by creating another artificial economy that is identical to the economy of interest, except that all information is fully disseminated.

Reviews the book "The Value of Communication in Agency Contracts: Theory and Experimental Evidence," by Joyce Berg, Lane Daley, Frank Gigler and Chandra Kanodia.

This paper provides experimental evidence on forward induction as a refinement criterion. In the basic extensive form, one of the two players chooses to play a battle-of-the-sexes game or to receive a certain payoff. According to forward induction, choosing to play the game is a signal about intended action. Though the presence of the outside option changes play, the authors find only limited support for the forward-induction hypothesis. The effects of the outside option also reflect the creation of a focal point through the asymmetry created by offering the outside option to one of the two players.