Since the adoption rate is unlikely to be an exoge-
nously determined variable, a second equation
(2) ADOPT = f(PMDP, PSUB, TREND)
is needed to complete the supply-response model. The
adoption rate is hypothesized to be directly related to
PMDP, inversely related to PSUB, and directly related
to a TREND variable. The trend variable is included
to capture the demonstration effect that occurs over time
as awareness of the technology spreads, as its effi-
ciency is improved, as regional, national, and foreign
markets develop, and as firms become convinced of the
stability of relevant regulations and Ieam to minimize
the attendant compliance costs.
Data, Time Frame, and Structure of
the Empirical Model
Data on the production of MDP are not collected
regularly by the USDA and are not readily available
from industry sources. Monthly data on the aggregate
U.S. production of MDP in 1978 and 1979 were ob-
tained by a special USDA survey of the 1,065 feder-
ally inspected poultry further-processing establishments
operating in 1980. Due to the time limitations of this
data series, the analysis must be constrained to the
short-mn time frame.
Monthly price data for MDP was constructed from
weekly quotes in the Urner Barry’s Price-Current
reports. A price index for the major substitutes in pro-
duction was Constrocted from the monthly prices for
chicken backs and necks and turkey necks {Poultry
Market Statistics). The quantity of ready-to-cook
poultry produced in federally inspected plants was used
as a measure of the availability of potential poultry raw
materials for producing MDP (Poultry and Egg Out-
look and Situation). Data on the adoption rate of the
technology, measured by the percent of establish-
ments operating deboning machines in each period, was
also obtained from the special USDA survey. Varia-
tion in the absolute number of firms operating debon-
ing machines each period ranged in a nonmonotonic,
though generally increasing, pattern from a low of 65
firms to a high of 95 firms as some of those initially
adopting the technology either abandoned it or went out
of business while other firms subsequently made the
adoption.
The TREND variable was assumed to be linear due
to the relatively short-run time frame of the study.
While a sigmoid relationship between the percent of
firms adopting a technology and time is often found in
technology-assessment studies, these studies gener-
ally span a much longer time frame. For a relatively
short time frame, a linear TREND variable should pro-
vide a reasonable approximation, even if the long-run
trend is nonlinear. There is also reason to believe that
during the period analyzed, the demonstration effect
referred to above (tending to accelerate the rate of
adoption by poultry processors) may have been mod-
erated by the regulatory uncertainty generated by the
controversy, negative publicity, and court battles over
mechanically deboned red meat regulations and the
subsequent near abandonment of the technology by red
meat processors.
The appropriate prices for supply analysis are the
prices expected by producers at the time production
decisions are being made (Gardner, p. 81). Futures
prices or lagged prices are often used as proxies for ex-
pected prices in supply-response analysis (Gardner;
Nerlove). Since MDP futures prices are not available,
one-period-lagged prices were used. This procedure
also avoids the problems created by the simultaneous
determination of supply and demand (Gardner).
Empirical Results
As emphasized by Morzuch et al. (p. 30), proces-
sors are concerned with relative prices in allocating
production capacity among alternative uses. Conse-
quently, the price variable was entered as the lagged
ratio of PMDP to PSUB to estimate the parameters of
equations (1) and (2). This formulation was also cho-
sen for its usefulness in modeling the impact of changes
in the regulations affecting MDP.
Since the two-equation system is recursive, the pa-
rameters were estimated using both ordinary least
squares (OLS) and two-stage least squares (2SLS)
regression. All coefficients are of the theoretically ex-
pected sign (Table 2). The quantity of MDP supplied
is directly related to its relative price, the quantity of
poultry inputs potentially available for producing MDP,
and the rate of adoption of the mechanical deboning
technology. The rate of adoption is directly related to
both the relative-price and the time-trend variable. All
coefficients are significant at the 0.10 level or lower.
The Durbin-Watson (DW) statistics indicate no prob-
lems with serial correlation, and the condition values
Table 2. Estimated MDP Supply Response Model
OLS and 2SLS Regression Coefficients3
(with standard errors)'3
Equation (PMDP/PSUB)^ QINPUTt ADOPTt TREND Rz DW COND(X)
(I)QMDPt
OLS 26.90 5.45 46.71 .87 2.26 39.59
(10.43) (2.49) (6.77)
** ** ***
2SLS 26.69 5.20 47.99 .87 2.30 41.91
(10.44) (2.54) (7.19)
** * ***
(2)AD0PT .36
t (.ID
***
.71 .92 1.80 14.44
(.05)
***
a) Intercept terms: equation (1) OLS—18,705.60 2SLS - 19,236.60
(3,761.16) (3,894.87)
*** ***
equation (2) 504.83
(19.54)
***
b) *** - significant at .01 level.
** - significant at .05 level.
* - significant at. 10 level.
Statements of significance are only approximate for 2SLS.
c) Units of measurement: QMDP - pounds per month, PMDM∕PSUB - ratio X 100, QIN-
PUT - pounds per month ÷ 10 ; ADOPT - percent firms X 100, TREND - months X
10.
135