HEDONIC PRICES IN THE MALTING BARLEY MARKET

Western Journal of Agricultural Economics

TABLE 1. Tests of Hypotheses of Constancy in Implicit Prices across Varieties and Grades
(F Ratio).

* Indicates rejection of the null hypothesis at the 5 percent level of significance.

estimates would have made presentation
and interpretation of the results unneces-
sarily voluminous. As an alternative, the
cross-section data for each Wednesday
were pooled throughout the crop year, re-
sulting in one estimated hedonic price
function for each year. Analysis of co-
variance was used to test the appropriate-
ness of pooling following Maddala (pp.
322-325). Hypotheses were posed that the
intercept terms and slope coefficients were
equal across months.⁸ The hypotheses of
equal intercepts across months was reject-
ed at the 5 percent level, but that of equal
slopes could not be rejected. To account
for the heterogeneity in the intercept term
across months, 11 monthly dummy vari-
ables were added to the empirical hedonic
price function that was then estimated us-
ing the pooled data.

Standard regression procedures were
used to estimate the regression coeffi-
cients. Problems associated with using
pooled data are the potential for serial
correlation and the heteroscedasticity in
the error terms throughout the ranges of
protein and plumpness. The estimated
models were tested for constancy of the
error terms using the Goldfeld-Quandt test
which is applicable to large samples. In all
cases the assumption of homoscedasticity
could not be rejected. It was not possible
to test for the existence of serial correla-
tion or to use recently developed proce-

⁸ The alternative would have been to test for homo-
geneity in coefficients across weeks, but because
there were unequal numbers of Wednesdays in each
year, month was used as the classification variable.
dures for estimation with pooled data
because of the unequal number of obser-
vations in each cross section.

Empirical Results and Hypothesis
Testing

The empirical model is unrestricted
across several parameters and provides a
framework for testing hypotheses about
the equality of some of the regression coef-
ficients. In particular, the empirical equa-
tion represents a three-way fixed effects
analysis of covariance (ANOVA) model
with two covariates. Variety, grade, and
months are the three class factors with
four, three, and twelve levels, respective-
ly. Analysis of covariance was used to test
hypotheses about the equality of the slope
coefficients and equality of the intercept
shifters. In the first case it was hypothe-
sized that the slope coefficients were ho-
mogeneous across varieties and grades.⁹
The homogeneity test determines whether
the implicit prices estimated from the
regression model are statistically different
across these classification variables. Hy-
potheses of homogeneity in implicit prices
for both plumpness and protein across va-
rieties and grades were tested for each
year, and the results are presented in Ta-
ble 1. These results indicate that 1) im-

⁹ Hypotheses about other interactions were also posed
and tested. In all cases these were insignificant and
are not reported here (see Crabtree) because they
were neutral with respect to the marginal implicit
prices and did not affect specification of the empir-
ical model.

<<   <   3   4   5   6   7   8   9   >   >>