Wilson
Hedonic Prices
PROit is the percent protein in sam-
ple i in t;
PLUit is the percent plumpness in
sample i in t;
FDBARt is the price of feed barley
in t;
eit is the error term.
Coefficients of particular importance are
β1 and β2 which can be used to derive val-
ues of marginal implicit prices for plump-
ness and protein. Alternative versions of
equation (5) were used to test for 1) the
appropriateness of pooling, 2) homoge-
neity of the marginal implicit prices across
the classification variables, and 3) constan-
cy of the marginal implicit prices.
Data Sources and Estimation
The Minneapolis Grain Exchange is the
only organized public market for malting
barley; consequently, price discovery at
this market plays an important role in es-
tablishing prices and price relationships
throughout the United States and other
countries. The data used in this study were
for spot transactions of malting barley at
this market. Malting barley is sold on the
basis of samples displayed by commission
firms on the floor of the Exchange. Most
samples represent a railroad car located at
country elevators in North Dakota, South
Dakota, and Minnesota. Accompanying
each sample is a “pan ticket” on which
results of the official inspection and other
information important to the sale are re-
corded. The inspection includes data on
both grade (i.e., the assigned grade) and
nongrade quality factors (e.g., variety,
plumpness, and protein). The Daily Mar-
ket Record quotes variety, numerical
grade, percent plumpness, protein con-
tent, -and price for each carlot sold on the
Exchange floor. This information was col-
lected for every Wednesday over the pe-
riod 1978/79 to 1981/82 crop years. The
last crop year contains only the first six
months when this analysis was under-
taken. There were 1,101, 1,218, 1,032, and
699 carlots used in the analysis in crop
years 1978/79, 1979/80, 1980/81, and
1981/82, respectively.
Separate hedonic price functions were
estimated for each of the four crop years
in order to reduce the potential problems
of inter-crop year variability in the mar-
ginal implicit prices. These could be at-
tributed to changes in the supply and/or
demand for the characteristics which
would largely stem from the varieties pro-
duced, weather, and agronomic practices.
Varieties produced and marketed were not
the same throughout the time periods cov-
ered in this analysis, and each has poten-
tially different yields of quality character-
istics, as well as inherent varietal attributes.
Agronomic practices (e.g., fertilization
which is positively related to protein) and
weather during the growing and harvest
seasons affect the yield of quality char-
acteristics and therefore supply. Conse-
quently, estimation of one equation for all
four years of data would potentially suffer
from aggregation bias because of the in-
ability to account for these unmeasurable
supply side influences.7 Estimation of sep-
arate hedonic price functions for each year
allows for a different equilibrium value
for the marginal implicit prices, rather
than constraining them to be equal across
years.
The data consisted of a cross section of
observations for Wednesday of each week.
However, the number of cross-sectional
observations was not equal across each
time period. Separate regression coeffi-
cients could have been estimated for each
week, but the large number of parameter
7 Pyler does indicate qualitatively that in the first
three years of the study, the supply of the charac-
teristics changed. However, this information was
very aggregated because it was an average across
samples collected from North Dakota, South Da-
kota, and Minnesota and was only available for crop
years 1977/78 to 1980/81. Consequently, it was not
possible to include these effects in the model.
33