This award funds the development of a new method for inferring the choices people will make in possible situations from a variety of 'non-choice' data. Many practical problems in microeconomics call for the researcher to predict the distribution of households' choices in not-yet-observed situations. Economists prefer to draw such inferences from actual choices in a closely related domain. However, this approach is sometimes problematic in practice because of the practical limitations of choice data. For example, we may not have data about closely related choices, or we may have serious concerns about uncontrolled factors, selections, and the endogeneity of opportunity sets.
The new method uses non-choice data. This method goes beyond current 'non-choice' methods such as the use of survey responses to include possible measures of passive physiological and neurological responses that have been widely studied in behavioral science. The new approach has several features beyond the use of non-choice data. The researchers focus on non-choice responses that are likely to have context-independent meanings over large collections of choice problems. The researchers treat the decision problem as the unit of observation, and they treat the problem as one of optimal statistical prediction. Finally, they want to determine whether and when these new methods are better than the standard method using imperfect choice data.
Broader impacts are substantial. The research could result in a new valuation method that could be used in addition to or instead of such methods as contingent valuation. The result would be improvements in how microeconomics is used to analyze a broad array of practical questions.
