In-Store Valuation of Steak Tenderness

Jayson L. Lusk, John A. Fox, Ted C. Schroeder,

James Mintert, and Mohammad Koohmaraie

Experimental methods were used to examine consumer willingness-to-pay for steak tenderness
in a grocery store setting. When relying on a taste test alone to determine product quality, the
participants paid an average premium of $1.23/lb for a tender versus tough steak. Fifty-one percent
of the participants were willing to pay an average of $1.84/lb when they had completed a taste
test and were also provided information about the steak’s tenderness. Results indicate that most
consumers prefer more tender steaks and that many are willing to pay a premium for tender steaks.
Key words: beef, consumer demand, experimental economics, field experiment, tenderness, willing-
ness to pay.

The beef sector has faced weakening demand et al.; Lamb and Beshear; Moon and Ward;
for two decades, with demand declining every Purcell; Unnevehr and Bard). Factors cited
year from 1979 to 1998 (Purcell). U.S. Depart- by these authors as contributing to the
ment of Agriculture (USDA) data indicate demand decline include changes in relative
per capita beef consumption dropped 20% prices, consumer health concerns, food safety
from 1970 to 1998 while inflation-adjusted concerns, product convenience and offering,
retail beef price declined 25% (LMIC). From product quality and consistency, changing
1990 to 1998 alone beef demand declined demographics, and evolving consumer pref-
by 18%.1 The dramatic demand decline has erences. Notable among these factors are
contributed to considerable restructuring and product quality issues. Although changes in
downsizing of the beef industry. For example, consumer “tastes and preferences” are exoge-
from January 1975 (the U.S. beef cow inven- nous to the beef industry, product quality
tory peak) to January 1999, the U.S. beef cow and consistency are attributes that can be
herd declined by 27% (LMIC).2 controlled during production and process-
Numerous studies have provided insights ing, albeit at some cost. Improvements in
into possible causes of the precipitous decline beef quality and consistency, however, have
in beef demand (e.g., Brester and Schroeder; been slow.
Brester and Wohlgenant; Capps, Moen, and Surveys of beef packers, purveyors, restau-
Branson; Chavas; Eales and Unnevehr 1988 ranteurs, and retailers indicate product uni-
and 1993; Flake and Patterson; Kinnucan formity, consistency, and tenderness are
among the highest ranked beef quality con-
The authors are, respectively, assistant professor, Mississippi
cerns (Smith et al., 1995). Among these
State University, associate professor, professor, and professor, quality traits, tenderness has been demon-
Kansas State University and research leader, Meat Research strated as the most important palatability
Unit, USDA-ARS U.S. Meat Animal Research Center.
The authors acknowledge the Research Institute on Livestock attribute of beef (Dikeman; Huffman et
Pricing for funding. The authors thank Jason Shogren and two al.; Miller et al.). However, current USDA
anonymous reviewers for their helpful comments and insights. quality grading standards are ineffective at
The authors are grateful to Jenny Graff, Matt Hoobler, and
Christy Lusk for assistance with data collection. Mention of a identifying meat tenderness (Savell et al.).
trade name, proprietary product or specific equipment is nec- The current grading system uses intramuscu-
essary to report factually on the available data; however, the
USDA neither guarantees nor warrants the standard of the prod-
lar fat or marbling as a measure of quality.
uct, and the use of the name by the USDA implies no approval of Marbling, however, is poorly correlated with
the product to the exclusion of others that may also be suitable.
1
tenderness. Wheeler, Cundiff, and Koch con-
Demand decline estimate is the percentage decline in real
retail price (from 1990 to 1998) holding per capita quantity con-
cluded that marbling explained at most only
stant at the 1990 level and assuming an own-price demand elas- 5% of the variation in beef tenderness. This
ticity of −0
67.
2
inconsistency in quality identification has
Although the beef cow herd has declined over this time
period, beef production has changed little due to technological led many industry participants to the same
change, increases in feeding efficiency, and lower feed prices. conclusions as Wheeler, Cundiff, and Koch
Amer. J. Agr. Econ. 83(3) (August 2001): 539–550
Copyright 2001 American Agricultural Economics Association
540 August 2001 Amer. J. Agr. Econ.

(p. 3150) that, “USDA quality grade does A Tenderness Based Grading System
not sufficiently segregate carcasses for palata-
bility differences, and thus a direct measure One direct method for measuring tender-
of meat tenderness is needed to supplement ness is the Warner–Bratzler (WB) shear force
USDA quality grade.” test. The WB test measures the amount of
Recent improvements in technology have force required to penetrate a cooked cut of
enabled researchers to more effectively seg- meat and assigns a numerical value to a cut
regate carcasses into tenderness categories, of steak indicating its tenderness level. The
thus providing a means of increasing prod- WB method explains more of the variation
uct quality and consistency to the consumer. in meat tenderness than any other tender-
Additionally, more appropriate price signals ness testing system (Shackelford, Wheeler,
may be passed from retail to farm levels and Koohmaraie). Recently, Koohmaraie
by improving quality measurement. Improve- et al. developed a beef processing system
ments of this nature provide the potential that incorporates the WB test and can
to counteract declining demand by helping be used in a commercial processing plant.
the beef industry provide consumers a prod- Based on findings from professional taste
uct that possesses the quality attributes they panels, Koohmaraie et al. proposed that
desire. the tenderness grading system be used to
New beef grading systems will only be segregate carcasses into three tenderness
based on meat tenderness if consumers value classifications–guaranteed tender, intermedi-
this attribute. In this study, we use an exper- ate tender, and probably tough.4 The sys-
iment to estimate consumer willingness-to- tem segregates carcasses with 90% accuracy,
pay for a higher level of steak tenderness, higher than any system previously devised
and estimate the influence of economic and (Shackelford, Wheeler, and Koohmaraie). In
demographic factors on willingness-to-pay addition to providing a more consistent pre-
values.3 We also show how revealing informa- diction of eating quality, this tenderness clas-
sification system results in more carcasses
tion about tenderness affects both preference
being segregated into the highest quality
and willingness-to-pay for tender steaks.
grade than under current USDA quality
Data were obtained using a novel exper-
grading standards. This occurs because some
imental method with consumers shopping carcasses that are graded as Choice or Select
in retail grocery stores. Following a taste in the current grading system would be cate-
test, steaks of varying tenderness were auc- gorized as guaranteed tender under this alter-
tioned using a variant on the true-value native grading system.
revealing auction mechanism proposed by
Becker, DeGroot, and Marschak. Results
indicate that: (a) consumers readily distin- Methods
guish between tenderness levels in a blind
taste test, (b) many, but not all, consumers
are willing to pay a premium for more tender We used an experimental market procedure
to elicit consumer willingness-to-pay values
relative to less tender steaks, and (c) when
for a higher level of tenderness in beef steaks.
information regarding tenderness is revealed
Valuation in non-hypothetical experiments
to consumers in addition to taste tests, they
has unique advantages (Fox et al.) and the
are significantly more likely to prefer and experimental valuation of food attributes is
pay more for tender steak. These results have well established in the literature (for exam-
important implications for beef producers ples see Shogren et al., 1994; Hayes et al.;
and processors, especially firms considering Melton et al.; Roosen et al.). To date, most,
investing in tenderness measurement tech- if not all, such valuation experiments have
nology to market steaks identified by tender- been conducted in laboratory settings with
ness levels. groups of participants randomly selected

3 4
In the only other study to report on consumer valuation of One important change that would occur with such a
tenderness, Boleman et al. provided steaks of varying tender- tenderness-based grading system is that a larger percentage of
ness to 42 families. Participants readily distinguished between the beef carcasses would receive the highest quality grade than under
tough and tender steaks. In a second phase of that study, 19 pur- the current system. Shackelford et al. predict that 29% of grain-
chased steaks. A majority (55%) of the steaks purchased were in fed carcasses would grade guaranteed tender while the current
the most-tender category but the study did not report the num- system grades carcasses as 1% Prime, 47% Choice, 47% Select,
ber of families purchasing different steak types. and 5% Standard.
Lusk et al. In-Store Valuation of Steak Tenderness 541

from the general population. Values are typ- incentives for subjects to accurately represent
ically elicited using a Vickrey second price their preferences. The incentive structures in
auction where participants submit a sealed the Vickrey and BDM mechanisms are sim-
bid, and the auction winner pays the second ilar in that bids are separated from market
highest bid price (Vickrey). In this study, the price. In the Vickrey auction the high bidder
valuation setting is moved to more familiar pays a price equal to the second highest bid;
territory for the consumer by conducting the in BDM, participants receive the item if their
experiment in a grocery store with individual bid exceeds a subsequently drawn random
shoppers rather than in a lab with randomly number, where the randomly drawn number
selected groups of participants. becomes the market price. The BDM pro-
In-store valuation has demonstrable and cedure is commonly used to elicit certainty
potential advantages for the experimenter equivalents for lotteries in risk preference
compared to a lab setting. First, in selecting studies (e.g., Kachelmeier and Shehata) and
subjects, it allows one to more closely tar- has also been used to value private goods
get the population of interest, in this case (Boyce et al.).5
meat buyers, by conducting the experiment To facilitate our experiment in the retail
at the meat case where the relevant pur- store, we modified BDM by using a prede-
chase decisions are normally made. Sample termined price for the good being auctioned,
selection bias may still arise since not every i.e., an upgrade from a tough steak (pro-
shopper will participate, but that bias will vided free for participating in the experi-
likely be smaller than in laboratory exper- ment) to a tender steak. This predetermined
iments because participation involves less price was unknown to the subjects and thus
inconvenience for the subject. Compensating independent of their stated bid. The price was
for participant inconvenience in laboratory revealed as the purchase price only to sub-
sessions requires financial inducements, typ- jects whose bid exceeded the price.6 When
ically around $20 to $30. Recruitment with- the price is higher than a participant’s bid,
out compensation likely increases selection they are pleased not to pay for the upgrade;
bias, but, since opportunity costs vary across when the price is lower, they are pleased to
individuals, it is possible that uniform com- obtain the upgrade for an amount below their
pensation may differentially impact subjects’ maximum willingness to pay.
revealed values. Buzby et al. (1998) reported
a significant positive effect on revealed values
for reductions in Salmonella risk when a $3 Procedures
participation payment was made to student
subjects whose opportunity costs were likely
Data were collected from shoppers at three
near 0.
urban retail grocery stores, owned by a large
A clear advantage to the in-store experi-
regional chain, in the midwestern United
ment is the reduced cost when $20–$30 par-
States. The experiment involved five steps in
ticipation payments are eliminated. Although
each of two treatments. Consumers partici-
some form of inducement for subjects to par-
pated one at a time without any knowledge
ticipate is still required, the total cost per
of the previous respondent’s outcomes. After
observation is likely to be much lower. In
completing the experiment, participants were
this study the inducement we provided was
asked not to discuss the results of the exper-
integral to the valuation exercise. Therefore,
iment with other shoppers in the store.
recruitment and participation expenses were
Step 1. Shoppers approaching the meat
effectively zero. Facilitating larger sample
counter were asked to participate in a short
sizes for a given research budget addresses an
issue, i.e., small samples, that has long been a
concern in economic experiments. 5
Recent work by Shogren et al. (2000) emphasizes the provi-
The switch to valuation in an individual sion of a market environment with feedback in valuation studies.
rather than a group setting also required There seems to be a trade-off here between (a) the ability to
an alternative valuation mechanism since the provide feedback from the market, and (b) the ability to create
an auction environment in-store that is relatively quick and not
Vickrey auction only works for a group. We totally disruptive to the store environment. We feared that using
elicited individual values using a variation repeated trials with market feedback would make the experi-
on the Becker–DeGroot–Marschak (BDM) ment a spectacle (or circus) that might lead bidders to feel like
they were on display and consequently affect bidding behavior
mechanism (Becker, DeGroot, and Marshak), in uncontrollable ways.
which, like the Vickrey auction, provides 6
Price for the upgrade was typically between $0.50 and $1.00.
542 August 2001 Amer. J. Agr. Econ.

experiment for which they would receive a force to give an actual value of steak tender-
free 12 oz ribeye steak. ness. Steaks are separated into different cat-
Step 2. Participants completed a short egories according to shear force values. The
written survey that required disclosure of three categories are: Guaranteed Tender,
basic demographic information including age, Intermediate Tender, and Probably Tough.
gender, household size, household income,
education level, and preference for steak The steaks used for sampling were deemed
doneness and USDA quality grade. tender or tough according to a slice shear test
Step 3. Participants then sampled two dif- using the procedures outlined in Shackelford,
ferent types of steaks labelled Red or Blue: Wheeler, and Koohmaraie. Meat scientists at
Red was “Guaranteed tender” (based on the the USDA’s Meat Animal Research Center
slice shear force test) and Blue was “Prob- at Clay Center, Nebraska conducted tender-
ably tough.”7 In experimental treatment 1, ness measurement and categorization of the
consumers were not told that the samples dif- steaks.
fered in tenderness—they had to make this A flow diagram of the experiment is shown
assessment independently. in figure 1. Consumer choices in the experi-
Step 4. Participants next responded to ment are based on their ability to “taste” ten-
questions about which steak they preferred derness, their preference for tenderness, their
overall and for the individual attributes of attitudes toward beef in general, and previous
taste, tenderness, texture, and juiciness. experience with beef. These factors may be
Step 5. Participants were given, free of captured by various consumer demographics,
charge, a 12 oz. Blue (probably tough) rib- consumption habits, and economic character-
eye steak. If they preferred the Blue (prob- istics. Factors such as age, education, income,
ably tough) steak, the experiment ended. gender, and household size are hypothesized
If they indicated a preference for the Red to play a role not only in consumer prefer-
(guaranteed tender) steak, they were asked ence for steak tenderness, but also in deter-
(both verbally and on the survey) to indi- mining willingness-to-pay. Additionally, the
cate the most they would be willing-to-pay to amount of beef consumed, knowledge of the
exchange their Blue (probably tough) steak current USDA quality grading system, and
for the 12 oz. Red (guaranteed tender) ribeye preference for steak doneness may affect
steak. Respondents were told that if their bid decisions.
exceeded a predetermined price (unknown
and exogenous to them), they would make
the exchange at that predetermined price. If
their bid was less than the predetermined Data and Results
price, then they kept their Blue (probably
tough) steak. Demographics and Consumption Habits
The second treatment was identical to the
A total of 313 consumers participated in the
first except that the words Red and Blue
study, 227 in the first treatment and 86 in
were replaced with “guaranteed tender” and
the second.8 Table 1 provides summary statis-
“probably tough”, respectively. That is, the
consumers were provided information about tics for the two groups. A majority of par-
the steaks in addition to their taste sam- ticipants were female and the average age
pling. Koohmaraie et al. used the descriptions in both groups was about 47 years. Partici-
“guaranteed tender” and “probably tough” pants on average had at least some college
when segregating steaks into different tender- education and a household income between
ness categories. To provide participants with a $40,000 and $50,000 in the first treatment
consistent explanation of the tenderness clas- and between $50,000 and $60,000 in the sec-
sification system, we provided the following ond treatment. Although the study includes
statement: only midwestern consumers, the participants

The USDA has developed a technology to

categorize steaks according to tenderness. 8
A fixed number of “in-store days” for conducting the experi-
The classification system uses slice shear ments were allocated by grocery store administration. The num-
ber of people that would enter the store each day was not known
prior to the experiment, only the number of days in each store
was known. More store days were allocated to the first treatment
7
“Guaranteed tender” and “Probably tough” steaks have slice of the experiment because the primary goal was to determine
shear force value of 15 kg or less and higher than 35 kg, consumers’ ability to identify tenderness and their willingness-
respectively. to-pay for tenderness based on a blind taste test.
Lusk et al. In-Store Valuation of Steak Tenderness 543

Figure 1. Consumer decision tree in steak tenderness valuation experiment

are representative of a wide range of demo- little or no beef took part in the study. Eigh-
graphics. For example, ages ranged from 19 teen consumers indicated that they did not
to 82; education ranged from less than high typically purchase beef at all and 60 con-
school to Ph.D.; income ranged from less than sumers indicated that they did not typically
$20,000/year to more than $120,000/year; and purchase steak. Regarding the quality of beef
total beef consumption ranged from 0 times typically purchased, 38% indicated they typ-
per week to over 12 times per week.9 ically purchased USDA Choice, 19% indi-
Consumers on average ate ground beef cated USDA Select, 19% indicated that they
about 2.2 times per week at home and slightly typically purchased store brand beef, and the
over 1 time per week away from home. Par- remaining 22% did not know the grade of
ticipants indicated that steak was consumed beef they usually purchased. Respondents, on
1.1 times per week at home and 0.6 times per average, preferred their steak cooked to a
week away from home. Although the exper- medium doneness.
iment was conducted near the meat counter,
several participants that routinely consumed
Consumer Preferences for Sampled Steak
9
Participants in the experiments were more educated and
Table 2 shows the proportion of participants
had slightly higher incomes as compared to the general popu- indicating preference for Red (guaranteed
lation. According to the U.S. Census Bureau (USCB), in 1998 tender) or Blue (probably tough) steak in
the median household income was $38,885/year and 22.2% of
the population had graduated from college. In our sample, the
both treatments. In the first treatment, where
median yearly household income was between $40,000 and consumers were not informed of the differ-
$49,999 and 41% of the respondents held at least a B.S. or B.A. ence in the two steaks, the Red (guaranteed
degree. However, participant demographics were similar to the
general population with respect to age and number of people tender) steak was preferred overall by 69%
living in the household. In 1998, the average age of the popula- of participants. Interestingly, of the four qual-
tion (over 18) was about 46 and the average number of people
living in the household was 2.63 (USCB). Experimental partici-
ity attributes, more people indicated they pre-
pants were on average 47.5 years of age and an average of 2.69 ferred the Red steak for tenderness (72%)
people lived in their households. A greater percentage of women than for any of the other attributes. This
participated in the in-store experiment (65%) than in the gen-
eral population (51%); however, this difference is likely due to suggests consumers were able to determine
the fact that women traditionally do most of the shopping. independently that the primary distinguishing
544 August 2001 Amer. J. Agr. Econ.

Table 1. Variable Definitions And Summary Statistics

Treatment Averages
Variable Definition 1 2
Gender 1 if female; 0 if male 0
69 0
58
(0
47) (0
49)
Age Age of respondent in years 47
76 46
87
(16
06) (12
91)
Education Education level of respondent 4
50 4
29
1 = less than 12th grade; 2 = high school graduate; (1
88) (1
60)
3 = some technical, trade; business school;
4 = some college; 5 = B.S. B.A., complete;
6 = some graduate work; 7 = M.S., M.A. complete;
8 = Ph.D., D.D.S., M.D., J.D., etc.
Income Household income level 4
39 5
78
1 = less than $20,000; 2 = $20,000 to $29,000; (2
71) (2
77)
3 = $30,000 to $39,999; 4 = $40,000 to $49,999;
5 = $50,000 to $59,999; 6 = $60,000 to $69,999;
7 = $70,000 to $79,999; 8 = $80,000 to $89,999;
9 = $90,000 to $99,999; 10 = $100,000 to $109,999;
11 = $110,000 to $119,999; 12 = more than $120,000
Adults Number of adults in household 1
96 2
02
(0
64) (0
70)
Children Number of children in household 0
71 0
74
(1
13) (1
09)
Beef home Number of times per week respondent consumes 2
28 2
21
ground beef (hamburger) at home (1
57) (1
46)
Beef away Number of times per week respondent consumes 1
32 1
17
ground beef (hamburger) away from home (1
58) (1
21)
Steak home Number of times per week respondent 1
12 1
05
consumes steak at home (0
85) (0
82)
Steak away Number of times per week respondent 0
54 0
63
consumes steak away from home (0
87) (0
71)
Doneness Preference of steak doneness; 1 = rare, 2 = medium rare, 3
22 3
05
3 = medium, 4 = medium well, 5 = well (1
07) (0
97)

Note: Number in parentheses are standard deviations. There were 227 participants in treatment 1 and 86 in treatment 2.

characteristic between the two steaks was Although a majority of respondents pre-
tenderness. ferred the tender steak, many indicated they
In the second treatment, consumers were were not willing to pay more to exchange
informed that one steak was guaranteed ten- their tough steak for a tender one. In treat-
der and the other was probably tough. When ment 1, 69% preferred, but only 36% of
compared with results from the first treat- the consumers were willing to pay extra
ment, consumers were more likely (84% to obtain a Red (guaranteed tender) steak.
When tenderness was revealed in treatment
compared to 69%) to indicate a preference
2, 84% preferred, but only 51% of the con-
for the Red or guaranteed tender steak. In sumers were willing to pay for the guaranteed
fact, for all four quality attributes, the pro- tender steak. The proportion of zero bids is
portion indicating preference for the tender surprisingly high, but there are several possi-
steak is higher when the differences in the ble explanations. In particular, zero bids may
steak samples were revealed. This suggests be symptomatic of participants’ prior expe-
that labelling the steaks with relevant infor- rience with USDA grades in the sense that
mation influences revealed preferences. they lacked confidence that a second steak
Lusk et al. In-Store Valuation of Steak Tenderness 545

Table 2. Respondent Steak Preferences

Treatment 1 Treatment 2
Attribute Red/Blue Tender/Tough
Taste % preferring Red (Tender) 65
04 77
91
% preferring Blue (Tough) 23
45a 10
47
Tenderness % preferring Red (Tender) 72
12 89
53
% preferring Blue (Tough) 18
58 4
65
Texture % preferring Red (Tender) 65
48 69
77
% preferring Blue (Tough) 22
12 13
95
Juiciness % preferring Red (Tender) 59
73 75
58
% preferring Blue (Tough) 27
88 10
47
Overall % preferring Red (Tender) 69
16 83
72
% preferring Blue (Tough) 22
02 9
30
% willing to pay more for Red (Tender) 36
12 51
16
Willingness to pay in $ per lbsb 1
23 1
84
Number of Participants 227 86
a Percentages do not sum to one in each category because some respondents indicated that they were “indifferent” between the two steaks with respect to
the given attribute.
b Average willingness to pay included only those individuals who offered a non-zero bid.

from the same category would have similar Econometric Models

qualities to the first (a consequence of the
very problem that objective tenderness scor- Figure 1 shows the three choices that con-
ing is designed to address). That the propor- sumers made in the experiment. First, respon-
tion bidding zero was lower when steaks were dents indicated which steak they preferred:
identified as “probably tough” and “guaran- Red (guaranteed tender) steak, Blue (prob-
teed tender” appears to lend support to this ably tough) steak, or neither. A multinomial
hypothesis. logit model was used to examine the impacts
Alternatively, zero bidding in this exper- of various factors on the probability of steak
iment may reflect the necessarily lower choice (Greene). The model was estimated
level of control in the grocery store setting using 245 observations—less than the total
compared to the lab. In particular, laboratory number of participants (313) because some
valuation studies typically require partici- respondents failed to answer one or more
pants to be the consumers of the good. In this survey questions. Marginal effects evaluated
experiment the intended consumer may have at the mean values are shown in table 3.
been someone other than the participant— Age and education have a statistically sig-
perhaps a spouse or other family member— nificant positive influence on the probabil-
in which case the preference revealed by the ity that a consumer prefers tender steak.
bid might not necessarily correspond to the The estimates indicate that for every one-
stated preference based on the taste test. This year increase in age, a participant is 0.4%
will be an issue for future attempts to move more likely to prefer the tender steak. Thus,
experimental valuation to settings that are a 45 year old would be expected to be 8%
more familiar for the participant.10 more likely to prefer tender steak than would
Average willingness-to-pay (of those con- a 25 year old. For a one “unit” increase
sumers who were willing to pay) was $1.23 in the respondent’s level of education (e.g.,
per pound in treatment 1 and $1.84 per from “some college” to “B.S., B.A. com-
pound in treatment 2. The distribution of plete”), a respondent is 4% more likely to
willingness-to-pay values for all consumers prefer the tender steak. Perhaps the most
that preferred tender steak is presented in notable result is the relatively large, posi-
figure 2. tive, and statistically significant estimate for
the “Treatment 2” dummy variable. Partici-
pants in the second treatment, where the ten-
10
Zero bids might also be due to cash constraints (i.e., the der and tough steaks were identified prior
respondent may have only had a credit card or check and not
cash), or time constraints (some participants probably did not to tasting, were 18% more likely to pre-
want to take the time to complete the bidding portion). fer tender steak than participants in treat-
546 August 2001 Amer. J. Agr. Econ.

Figure 2. Distribution of willingness-to-pay premiums for tender steak

Table 3. Multinomial Logit–Marginal Probabilities of Factors Affecting

Consumers Preference for Guaranteed Tender Steaks
Steak preference category
Variable Tender Tough Neither
Age 0
004∗∗ −0
003∗ −0
001
(0
002) (0
002) (0
001)
Education 0
040∗∗ −0
027∗∗ −0
013
(0
014) (0
012) (0
008)
Gender 0
071 −0
027 −0
044
(0
060) (0
053) (0
033)
Income −0
005 0
007 −0
002
(0
011) (0
010) (0
007)
Treatment 2 0
181∗∗ −0
142∗∗ −0
039
(0
069) (0
061) (0
041)
Doneness −0
044∗ 0
035 0
009
(0
025) (0
022) (0
014)
Steak home 0
027 −0
047 0
020
(0
034) (0
032) (0
016)
Steak away −0
003 0
006 −0
003
(0
034) (0
029) (0
019)
Knowledge of grade −0
004 0
032 −0
028
(0
066) (0
059) (0
035)
Note: Marginal effects evaluated at the means of the independent variables.
Number of observations = 245; log likelihood = −172
55; percentage of correct predictions = 74%.
∗ , ∗∗ Denote significance at the 10% and 5% levels, respectively.
a Values in parentheses are standard errors.
b Knowledge of grade = 1 if respondent knew the quality grade of beef typically purchased, 0 otherwise.
Lusk et al. In-Store Valuation of Steak Tenderness 547

ment one where tenderness levels were not where φ is the standard normal density func-
identified. The only steak preference vari- tion and
2 is a vector of coefficients. Given
able that influenced steak choice was cooking equations (1) and (2), the likelihood for the
doneness—consumers who preferred their random sample as shown by Haines, Guilkey,
steaks cooked to a higher degree of doneness and Popkin is
were less likely to prefer the tender steak.

(3) L = (−
1 Xi )(1−ti ) (
1 Xi )[(1/σ)
This is logical in that beef cooked to a higher t
degree of doneness is inherently tougher than ×φ([Pi −
2 Xi ]/σ)]/(
2 Xi /σ) i
beef cooked to a lower internal temperature
(Milligan). where ti takes the value of 1 if Pi > 0 and
When respondents indicated a preference 0 otherwise. This version of the double hur-
for the Red (guaranteed tender) steak, they dle model assumes that the error terms in the
were given the opportunity to bid to upgrade two steps (the participation and consumption
from their Blue (probably tough) steak to decisions) are independent and normally dis-
the Red (guaranteed tender) steak. With tributed. Results of the double-hurdle model
a significant proportion of the willingness- are presented in table 4.11
to-pay observations at the zero limit, an The first column of results in table 4 shows
appropriate model for estimating willingness- the effects of consumer demographics on the
to-pay is the tobit model. However, tobit probability that the consumer will pay for
estimation is restrictive in that it assumes the Red (tender) steak. All variables are sta-
that the probability of a limit outcome (a tistically insignificant at standard significance
zero willingness-to-pay bid) and positive out- levels. The positive coefficient on “Knowl-
edge of Grade” suggests that consumers who
comes are affected identically by the same
knew which quality grade of beef they typi-
determinants. Haines, Guilkey, and Popkin
cally bought may be more likely to pay for
showed that consumption of many food items
tenderness—but the effect is only marginally
often involves a two-step process. A common
significant (p = 0
21). Although both income
model to use in this instance is the double
and treatment were expected to have posi-
hurdle model developed by Cragg. Cragg’s
tive effects, neither coefficient is statistically
model is more general than the tobit in that
significant.
it allows for different determinants of limit
Estimates for the second hurdle indicate
outcomes and positive outcomes. that females and younger consumers were
Following Cragg, let Pi be the ith con- willing to pay more for the upgrade to the
sumer’s bid to upgrade to the Red (tender) tender steak. Surprisingly, income level did
steak. The first hurdle is the consumer’s deci- not significantly affect the amount respon-
sion of whether to pay for the Red (tender) dents were willing to pay. The most impor-
steak. If the consumer chooses not to bid, tant determinant of willingness-to-pay was
then Pi = 0. The probability of this outcome the information treatment—consumers in the
is second treatment, where the tenderness lev-
(1) Prob(Pi = 0) = (−
1 Xi ) els were revealed, were willing to pay about
$0.82 per pound more for tender steak and
where  is the standard normal distribu- this effect was statistically significant (p =
tion function, Xi is a vector of consumer 0
001). The significance and magnitude of
i’s economic and demographic characteristics this estimate suggests that information and
as previously indicated, and
1 is a vector labelling have important economic impacts
of coefficients. The second hurdle determines on consumer willingness-to-pay for a quality
the effect of the independent variables on Pi characteristic.
given that Pi > 0. Thus, the second hurdle
involves the decision of how much to pay for
the Red (tender) steak given that the con- Implications and Conclusions
sumer has decided to pay. The distribution of
Pi conditional on being positive is truncated Current USDA quality grading standards do
at zero and assumed normal with mean
2 Xi not provide consumers with a direct mea-
and variance σ2 . The second hurdle is formu- sure of tenderness. As a result, consumers
lated as

  
(2) f(Pi |Pi > 0) = (1/σ)φ Pi
2 Xi /σ 11
A Lagrange Multiplier test, proposed by Lin and Schmidt,
  indicated that the double hurdle model was preferred to the
/ 
2 Xi /σ tobit.
548 August 2001 Amer. J. Agr. Econ.

Table 4. Double Hurdle Model—Determinants of Consumer

Willingness-to-Pay for Guaranteed Tender Steak
First Hurdle
Probability Second Hurdle
Variable of Paying Amount Paid
Constant 0
275 1
478∗∗
(0
575)a (0
613)
Age −0
004 −0
016∗∗
(0
007) (0
008)
Education 0
044 0
011
(0
056) (0
062)
Gender −0
150 0
409∗
(0
210) (0
237)
Income −0
044 0
045
(0
039) (0
047)
Treatment 2 0
019 0
816∗∗
(0
212) (0
258)
Household Size −0
060 −0
083
(0
089) (0
091)
Steak home 0
044 0
028
(0
121) (0
135)
Steak away 0
101 −0
115
(0
132) (0
143)
Knowledge of grade 0
286 0
009
(0
230) (0
292)
Sigmab 0
911∗∗
(0
094)
Number of Observations 182 100
a Values in parentheses are standard errors.
b Sigma is the error variance; log likelihood of first hurdle = −121
8; log likelihood of second hurdle = −110
2.
∗ , ∗∗ Denote significance at the 10% and 5% levels, respectively.

are uncertain when they purchase a steak to 69%) and willingness-to-pay for the “guar-
(or other beef cut) whether it will be tender. anteed tender” steak increased by $0.82 per
In this experiment, participants tasted both pound. Third, younger consumers and women
tough and tender steaks and, if they preferred were willing to pay significantly more to
the tender steak, bid to upgrade from a tough upgrade to a tender steak than were other
to a tender steak. The willingness-to-pay consumers. Surprisingly, consumer income
values demonstrate some important points did not affect either the probability or the
about consumer preferences for tenderness. amount that participants would pay.
First, some consumers are willing to pay Overall, our results indicate that a quality-
large premiums to obtain a “guaranteed ten- grading standard based upon (or supple-
der” steak instead of a “probably tough” mented with) tenderness measurements may
steak. Of the participants who were informed be a viable alternative to the current system.
that the steak they were bidding for was Having established that the upgrade from a
“guaranteed tender,” 20% were willing-to- tough to a tender steak has value for con-
pay a premium of $2.67 per pound or more. sumers, future research is needed to explicitly
Second, providing consumers information via compare the proposed tenderness based sys-
labelling has value. Participants in treatment tem with the current system—i.e., values for
1 relied on a taste test alone to differenti- the upgrade from Select or Choice steaks to
ate between the two steak samples. Partici- steaks of known tenderness—in order to bet-
pants in treatment 2 were told which steak ter estimate the overall benefits.
was tender and which was tough. Given that Historically, the vast majority of beef has
information, more participants preferred the been marketed as a generic commodity with
“guaranteed tender” steak (84% compared few branded products. Beef processors, who
Lusk et al. In-Store Valuation of Steak Tenderness 549

for many years primarily focused on being Acceptance of Meat Products.” Proc. Recip.
low cost providers, are beginning to consider Meat Conf. 40(1987):93.
product differentiation and branding. Differ- Eales, J.S., and L.J. Unnevehr. “Demand for Beef
entiation of beef via tenderness is beginning and Chicken Products: Separability and Struc-
to occur among some industry participants, tural Change.” Amer. J. Agr. Econ. 70(August
albeit at a slow rate. Information regarding 1988): 521–32.
the value of steak tenderness in addition to . “Simultaneity and Structural Change in U.S.
improvements in technology, that lower costs Meat Demand.” Amer. J. Agr. Econ. 75(May
of tenderness identification, segregation, and 1993): 259–68.
alteration are the catalysts in this evolving Flake, O.L., and P.M. Patterson. “Health, Food
market. Safety and Meat Demand.” Paper presented
at the AAEA annual meeting, Nashville,
[Received October 1999; TN, August 1999.
accepted August 2000.] Fox, J.A., J.F. Shogren, D.J. Hayes, and
J.B. Kliebenstein. “CVM-X: Calibrating Con-
tingent Values with Experimental Auction
Markets.” Amer. J. Agr. Econ. 80(August
References 1998):455–65.
Greene, W.H. Econometric Analysis. Upper Saddle
Becker, G.M., M.H. DeGroot, and J. Marschak. River, NJ: Prentice-Hall, 1997.
“Measuring Utility by a Single-Response Haines, P.S., D.K. Guilkey, and B.M. Popkin. “Mod-
Sequential Method.” Behav. Sci. 9(July eling Food Consumption Decisions as a Two-
1964):226–32. step Process.” Amer. J. Agr. Econ. 70(August
Boleman, S.J. et al. “Consumer Evaluation of Beef 1988):543–52.
of Known Categories of Tenderness.” J. Anim. Hayes, D.J., J.F. Shogren, S.U. Shin, and J.B.
Sci. 75(June 1997):1521–4. Kliebenstein. “Valuing Food Safety in Experi-
Boyce, R.R., T.C. Brown, G.H. McClelland, G.L. mental Auction Markets.” Amer. J. Agr. Econ.
Peterson, and W.D. Schulze. “An Experimen- 77(February 1995):40–53.
tal Examination of Intrinsic Values as a Huffman, K.L., M.F. Miller, L.C. Hoover, C.K. Wu,
Source of the WTA-WTP Disparity.” Amer. H.C. Brittin, and C.B. Ramsey. “Effect of Beef
Econ. Rev. 82(December 1992):1366–73. Tenderness on Consumer Satisfaction with
Brester, G.W., and T.C. Schroeder. “The Impacts Steaks Consumed in the Home and Restau-
of Brand and Generic Advertising on Meat rant.” J. Anim. Sci. 74(January 1996):91–7.
Demand.” Amer. J. Agr. Econ. 77(November Kachelmeier, S.J., and M. Shehata. “Examin-
1995):969–79. ing Risk Preferences Under High Monetary
Brester, G.W., and M.K. Wohlgenant. “Estimating Incentives: Experimental Evidence for the
Interrelated Demands for Meats Using New People’s Republic of China.” Amer. Econ.
Measures for Ground and Table Cut Beef.” Rev. 82(December 1992):1120–40.
Amer. J. Agr. Econ. 73(November 1991): Kinnucan, H.W., H. Xiao, C.-J. Hsia, and J.D.
1182–94. Jackson. “Effects of Health Information and
Buzby, J.C., J. A. Fox, R.C. Ready, and S.R. Generic Advertising on U.S. Meat Demand.”
Crutchfield. “Measuring Consumer Benefits Amer. J. Agr. Econ. 79(February 1997):13–23.
of Food Safety Risk Reductions.” J. Agr. Appl. Koohmaraie, M., C.N. Cutter, W.J. Dorsta, S.D.
Econ. 30(July 1998):69–82. Shackelford, G.R. Siragusa, and T.L. Wheeler.
Capps, O., D.S. Moen, and R.E. Branson. “Con- “A Summary of Beef Quality and Meat Safety
sumer Characteristics Associated with the Research.” Unpublished Report, USDA-ARS,
Selection of Lean Meat Products.” Agribusi- U.S. Meat Animal Research Center, Clay Cen-
ness 4(November 1988): 549–57. ter, NE. 1996.
Chavas, J. “Structural Change in the Demand for Lamb, R.L., and M. Beshear. “From the Plains
Meat.” Amer. J. Agr. Econ. 65(February 1983): to the Plate: Can the Beef Industry Regain
148–53. Market Share?” Federal Reserve Bank of
Cragg, J.G. “Some Statistical Models for Limited Kansas City, Econ. Rev., Third Quarter
Dependent Variables with Application to the 1998, Internet address: http://www.kc.frb.org/
Demand for Durable Goods.” Econometrica publicat/econrev/ermain.htm.
39(September 1971):829–44. LMIC. Livestock Marketing Information Cen-
Dikeman, M.E. “Fat Reduction in Animals and ter, Lakewood, Colorado. Internet address:
the Effects on Palatability and Consumer http://lmic1.co.nrcs.usda.gov.
550 August 2001 Amer. J. Agr. Econ.

Lin, T.F., and P. Schmidt. “A Test of the Tobit Spec- Savell, J.W., R.E. Branson, H.R. Cross, D.M.
ification Against an Alternative Suggested by Stiffler, J.W. Wise, D.B. Griffin, and G.C.
Cragg.” Rev. of Econ. and Statist. 66(February Smith. “National Consumer Retail Beef
1984):174–77. Study: Palatability Evaluations of Beef Loin
Melton, B.E., W.E. Huffman, J.F. Shogren, and J.A. Steaks that Differed in Marbling.” J. Food Sci.
Fox. “Consumer Preferences for Fresh Food 52(May/June 1987):512–17.
Items with Multiple Quality Attributes: Evi- Shackelford, S.D., T.L. Wheeler, and M.
dence from an Experimental Auction of Pork Koohmaraie. “Tenderness-Based Classifica-
Chops.” Amer. J. Agr. Econ. 78(November tion of Beef.” Unpublished manuscript, U.S.
1996):916–23. Meat Animal Research Center, USDA, Clay
Miller, M.F., K.L. Huffman, S.Y. Gilbert, L.L. Center, NE. 1996.
Hammon, and C.B. Ramsey. “Retail Con- Shogren, J.F., S. Cho, C. Koo, J. List, C. Park, P. Polo,
sumer Acceptance of Beef Tenderized with and R. Wilhelmi. 2000. “Auction mechanisms
Calcium Chloride.” J. Anim. Sci. 73(August and the measurement of WTP and WTA.”
1995):2308–14. Resour. and Energy Econ., in press.
Milligan, S.D., M.F. Miller, C.N. Oats, and C.B. Shogren, J.F., S.Y. Shin, D.J. Hayes,
Ramsey. “Calcium Chloride Injections and J.B Kliebenstein. “Resolving Differences in
Degree of Doneness, Effects of Sensory Char- Willingness to Pay and Willingness to Accept.”
acteristics of Beef Inside Round Roasts.” J. Amer. Econ. Rev. 84(March 1994):255–70.
Anim. Sci. 75(March 1997):665–72. Smith, G.C. et al. The National Quality Beef Audit.
Moon, W., and R.W. Ward. “Effects of Health Colorado State University, Texas A&M Uni-
Concerns and Consumer Characteristics on versity, College Station, and Oklahoma State
U.S. Meat Consumption.” Paper presented at University for the National Cattleman’s Beef
the AAEA annual meetings, Nashville, TN, Association. 1995.
August 1999. Unnevehr, L.J., and S. Bard. “Beef Quality: Will
Purcell, W.D. “Measures of Changes in Demand Consumers Pay for Less Fat?” J. Agr. Resour.
for Beef, Pork, and Chicken, 1975-1998.” Econ. 18(December 1993): 288–95.
Research Institute on Livestock Pricing, U.S. Census Bureau. http://www.census.gov.
Research Bulletin 3-98, Blacksburg, VA, Vickrey, W. “Counterspeculation, Auctions, and
October 1998. Competitive Sealed Tenders.” J. Finance
Roosen, J., D.A. Hennessy, J.A. Fox, and A. 16(1961):8–37.
Schreiber. “Consumers’ Valuation of Insec- Wheeler, T.L., L.V. Cundiff, and R.M. Koch.
ticide Use Restrictions: An Application to “Effect of Marbling Degree on Beef Palata-
Apples.” J. Agr. Resour. Econ. 23(December bility in Bos taurus and Bos indicus Cattle.”
1998):367–84. J. Anim. Sci. 72(December 1994):3145–51.