An increasing product differentiation coupled with an increasing availability of electronic data has boosted the number of hedonic price analyses applied to food and agricultural products. Most of these studies estimate the first stage of a complete two-stage model as proposed by ROSEN. However, there are also a few studies that estimate the second stage, i.e. supply and demand functions for characteristics. The present paper reviews both the theoretical and applied literature on Rosen’s two-stage model in the context of food and agricultural economics. Based on these findings, a theoretical model for specialty coffee auction data is proposed and tested empirically. The empirical model comprises non-linear hedonic bid functions at stage one and an inverse demand function for one characteristic, the sensory quality score (SQS), at stage two. The first-stage results indicate a high variability of the marginal price of the SQS across different auctions, i.e. across time and space. The second-stage results suggest that the marginal prices of the SQS increased in the analysed period 2003-2009 and that country-of-origin and buyer effects are important. The highest marginal prices are paid for Rwandan and Honduran coffee. At first glance, this is surprising, since at the first stage Honduran coffees are almost always sold at discounted prices compared to coffees of other origins. However, it seems that the SQS is a much more important quality cue for a coffee origin with a low reputation than for a coffee origin with a well-established reputation in the marketplace.
