Published online by Cambridge University Press: 26 January 2015
In-person interviews were carried out with Mexican millers who were administered a conjoint-type survey designed to incorporate uncertainty in attribute levels. Two methods were used to model millers' risk preferences: a modified mean-variance approach and an explicit expected utility approach. Controlling for variability, Mexican millers are willing to pay premiums for increases in quality factors such as test weight, protein content, falling number, and dough strength/extensibility. We find millers are not particularly sensitive to changes in the variability of quality characteristics. Out-of-sample forecasts suggest the mean-variance model provides an accurate depiction of actual Mexican imports.