indicate that most consumers prefer tender beef and that many are willing to pay a premium for a
guarantee of tenderness. However, none of these studies explored the effect of inconsistent tenderness
on consumer WTP for beef. If, and as is hypothesized, consumers are risk averse, part of the premium
paid for guaranteed tender beef may be due to the difference in tenderness variability between generic
and guaranteed tender beef.
Market experiments with carefully controlled products, probabilities, and payoffs can be used
to investigate consumers’ WTP for various products and the importance of risk aversion in choices
among products. There are several experimental methods for eliciting individual risk preferences; see
Wakker and Deneffe for a discussion of risk preference elicitation methods. A common method is to
elicit the WTP for risky prospects. The lower the WTP is relative to the expected value of the
prospect, the higher the level of risk aversion (e.g., Harrison, 1986; Kachelmeier and Shehata, 1992;
Pennings and Garcia, 2001).
We investigate how inconsistency in beef tenderness affects consumers’ valuations by using
an experimental auction to elicit consumers’ valuation of beef in three tenderness categories as well as
uncategorized generic beef. Given the known tenderness distribution of the generic beef, we calculate
the ratio between the WTP for the generic beef and the expected value of the generic beef for each
auction participant. The lower this risk ratio, the more risk averse the participant. The contributions of
this paper are as follows. First, we show how lottery theory and experimental auctions can be used to
study the effect of quality inconsistency on consumers’ WTP for food products. Second, the effects of
inconsistent beef tenderness on consumers’ WTP for beef have not been studied in the past and we fill
this empirical gap in the literature.
Experiment and Auction Design
In November 2003, we established a market experiment at the Norwegian Food Research Institute. We
conducted four sessions with each session lasting approximately 90 minutes. Fifty-one participants
were recruited through local organizations, including choirs and soccer teams, in southeastern
Norway.1 In each organization, the contact person was instructed to provide a sample of regular
consumers of beef, between 25 and 60 years old, with an approximately equal division of sexes. Each
participant was paid NOK 3002 to participate. In addition, NOK 200 was paid to the recruiting
organization for each participant who completed the experiment.
Table 1 presents the descriptive statistics for the participants. The participants’ age ranged
from 23 to 59 years, with an average of 38 years. Fifty-one percent were female. The average
household income was NOK 531,000 and 55% had at least some university level education. The
sample chosen is representative for the age group in the region.
We obtained 120 kilograms of beef sirloin from Norsk Kj0tt, Norway’s largest meat
processing company. The quality of the beef reflected the variation in quality found in local stores.
The beef was produced on two consecutive days, vacuum-packed as whole loins, and stored for
tenderizing for 14 days. The loins were then numbered from one to 48, cut into 1.5 cm slices,
partitioned into portions weighing approximately 400 grams, and packed in consumer packages in a
modified atmosphere similar to that found in Norwegian meatpacking.
Neither consumers nor experts can determine the tenderness of beef by visual inspection, and
two cuts with the same marbling classification, for example U.S. select strip loin, can differ
significantly in tenderness (Miller et al., 2001). We used the Warner-Bratzler Shear Force (WB) test to
measure tenderness (Boleman et al., 1997). The WB test measures the amount of force required to
penetrate a cut of meat: the lower the force required the more tender the meat. Consumer preferences
regarding beef with various WB scores have been investigated in several papers (Boleman et al., 1997;
Sivertsen, Kubber0d, and Hildrum, 2002; and Huffman et al., 1996). Sivertsen, Kubber0d, and
Hildrum (2002), for example, concluded that WB test results are highly correlated with consumer
ratings of tenderness with a sample correlation coefficient of -0.87.
The WB test was used to categorize the beef into three tenderness categories: very tender,
tender, and less tender. This categorization was based on the measured distribution of the WB scores
in the sample, i.e., the relative tenderness of the beef. Accordingly, the lowest 25% of WB scores in
the sample were categorized as very tender, the next 50% as tender, and the remaining 25% as less
tender.