results. Alternative estimation methods, i.e., stepwise and adjusted R2, were used in selecting the
model to report.
Independent variables were characteristics of the feedlots, marketing and pricing practices, and
response to statements regarding concerns with grid pricing or factors that affected their
likelihood of using grid pricing. Responses to the thirteen statements were included in three
alternative ways, with similar results. Alternatives included: (1) including a continuous variable
(rating 1-7) for each statement just as respondents rated them; (2) creating a dummy variable for
those agreeing (ratings 1-3) and disagreeing (ratings 5-7) with each statement compared with
those neither agreeing or disagreeing (rating 4); and (3) creating a dummy variable for those
most agreeing (ratings 1-3) or most disagreeing (ratings 5-7) with each statement compared to
those neither agreeing nor disagreeing and responding in the opposite direction, thus comparing
the majority responses (either agreement or disagreement) with all other respondents.
Ordinary least squares (OLS) estimation results are shown in Table 4. Results for the ordered
logit (OL) model which included the same variables as the OLS model were essentially identical
in terms of signs and significance with two minor exceptions, both relating to the significance
level being just above or below 0.10. While ordered logit model results are not reported here,
selected results are mentioned.
The same two variables were significant in every version of both the OLS and OL models. As
the percent of fed cattle sold to the largest buyer increased, both the extent of grid pricing and
the probability of a higher usage of grid pricing increased. Similarly, as the percent of fed cattle
marketed with an agreement, contract, or part of an alliance or cooperative increased, both the
extent of grid pricing and the probability of higher usage of grid pricing increased. The latter
result was consistent with the 2002 survey of cattle feeders (Schroeder et al. 2002). The former
result was not surprising to the author but was initially thought to relate to size of feedlot.
However, feedlot size was rarely if ever significant in models estimated and feedlot size was not
highly correlated with extent of marketings to the largest buyer. Neither was the correlation
between extent of marketings to the largest buyer and use of a marketing agreement or related
commitment especially high (r = 0.504).
Recall from above that based on mean ratings, four of the five most important statements
determining whether or not to use grid pricing related to the Cattle Characteristics group. For
the regression results, just one statement was significant from this group. As agreement
increased that a determining factor was the expectation cattle would dress well (have a high
dressing percentage), the extent of grid pricing increased. While this variable was marginally
significant in the OL model, feeders agreeing with the statement were 3.4 times more likely to be
in the group using grid pricing more frequently than in the group using grid pricing for half or
less of their marketings.
Three significant factors determining the extent of grid pricing came from the Price and Market
Conditions group, suggesting that market conditions are indeed a driving force in determining
the extent of grid pricing. As agreement increased regarding the importance of a wide Choice-