FASB Research Fellow Phil Shane asked me a thought-provoking question last week:  what do we know about the differences between people’s stated beliefs about the value of information and their actual decisions?  I am hoping readers might help me flesh out a few preliminary thoughts.

First, a comment on the relevance of the question for standard setting.  Standard-setters often solicit investors’ opinions on whether some new disclosure would provide useful information, but to what extent are these opinions reliable?  Because the information hasn’t been available before, it is reasonable to wonder whether investors are even in a position to answer questions about how useful it would be.  Even granting that they can answer the question, there can still be many a slip twixt belief and decision.  Would they actually go through the effort needed to incorporate the new information into their analyses?  Would the information that seems valuable from 30,000 feet actually seem as useful when they are dealing with the messy details of a particular investment decision?  Under what conditions are generic conjectures about a disclosure’s value a reliable indicator of its true value in investment decisions?

One answer might come from the literature on contingent valuation of public goods–a staple of experimental and survey research in environmental economics.  These studies ask questions like “how much would someone have to pay you to live near a nuclear reactor?” or “how much would you pay to save the habitat of the spiny lap frog?”  (OK, I made that second one up).

A recent paper on biases in contingent valuation provides an interesting perspective on the difference between values elicited in hypothetical surveys and observations of willingness to pay or accept in the context of real decisions:

The vast majority of contingent valuation hypothetical bias experiments elicit subjective, homegrown values, rather than experimenter-controlled, pre-assigned induced values. Unlike induced values, the researcher cannot know these homegrown values with certainty. Because of this, when researchers observe that values in a hypothetical payment scenario are higher than the corresponding treatment with actual, consequential payments, it is impossible to know for sure whether hypothetical values are overstated or actual values are understated (or possibly a combination of the two). Researchers typically make the reasonable assumption that the responses in the real settings accurately represent the true economic value, yet it is entirely possible that the reverse is true. For example, when payments for public good provision are consequential, responses could be biased downward due to factors like free-riding or the desire to only pay one’s fair share. In the case of private goods, both Harrison et al. (2004) and Murphy and Stevens (2004) hypothesize that responses to actual payment questions for private goods may be censored by the market price. The linkage between value elicitation for public and private goods is particularly important considering that many experimental studies of non-market valuation.

The paper goes on to show that “homegrown” valuations can indeed be poorly measured by contingent valuation methods for hypothetical settings.

But what do we know about contingent valuations for accounting disclosures.  I am well aware that researchers (and peer reviewers!) often prefer to see beliefs as they are reflected in decisions, rather than in preferences over hypothetical scenarios.  But is there research showing systematic differences between these two forms of elicitation? If so, which should we believe more?  Also, how does the value of information differ from the value of environmental changes?  Is there reason to believe that contingent valuation would be more or less useful in financial reporting than in environmental policy?

Any thoughts would be much appreciated!