An Experimental Market's Investigation of Discretionary Financial Disclosure

The extent to which market forces can induce full financial disclosure by managers has long been an issue of interest to regulators. Investigating this phenomenon with naturally occurring data produces a major obstacle: since managers' private information sets are unknown, it is necessary to make assumptions about them in order to interpret the nature (e.g., favourable or unfavourable, income increasing or income decreasing) of the information that is disclosed. The validity of the inferences relies critically on the validity of these assumptions. The present study uses a laboratory experiment to test three hypotheses derived from prior analytical and empirical research: (H1) When disclosure costs are zero, managers voluntarily disclose all (good and bad) news; (H2) When disclosure costs are positive. managers only disclose news which exceeds some threshold: and (H3) The mandatory disclosure of non-proprietary information induces an increase in the disclosure of correlated. proprietary information. One hundred and fifty-six subjects participated in markets with one firm manager and three investors. Over thirteen independent periods, the managers decided whether to truthfully disclose the liquidation value of the asset under their stewardship, and the investors submitted competing bids for the asset. With costless disclosure. investors price-protected themselves when managers withheld information, but the price penalty that they imposed was insufficient to induce full disclosure. With positive disclosure cost, investors reduced the price penalty that they imposed for non-disclosure, and managers disclosed proportionally fewer of the less extreme good news. Finally, mandatory disclosure of information had no significant impact on the voluntary disclosure of correlated proprietary information. Discussion centres on our failure to support the (equilibrium) prediction from analytical research that full disclosure should obtain when disclosures are costless. Several limitations of the study are examined. and it remains an open question whether additional trials (periods) in the present study might have provided full disclosure.