±∆Qi= α↑ + У.,⅛∆Pi,+ У ηj∆Z∣ + ∑tΦik∆Wk i =1, . . . ., m3, (7)
i i ii i ij j i
A constant was added to account for a technology-shifter (replacing the linear trend variable in
the levels-model). (Estimation results are available upon request from the authors). As compared to the
levels version of the model, goodness of fit has been substantially lowered (in particular for the arable
supply function). The goodness of fit for the arable, meat, milk, hired labour, and fertilizer equations
are respectively 0.11, 0.64, 0.41, 0.32 and 0.45, which are still satisfactory given that the model is now
in first differences4. All parameters associated with own price responses have the appropriate sign, and
4 out of 5 were significant. Evaluating the t-values, 22 out of 42 two parameters are significant. The
significance levels of the price (and other) parameters have declined in comparison with estimating the
same model in level-terms. This confirms the problems with non-stationarity (but no cointegration) in
the levels model, which is known to lead to overstated R-square and t-values.
In order to determine the long-run elasticities, the remaining part of the profit function was
estimated. Firstly, however, (see equation 6) the equation had to be respecified in terms of first
differences as
∆∏ = yα∖P. + yιμ,∆Zj + У δl∆W
+ 22 У,УГ Y,,∆(. ZlZf) + 22 У к У χ-∆WW-) + У ∣ У k θ,t ∆(ZW ) (8)
and where Π differs from ∏ because it includes as an additional term ∑i αiPi. The estimation
yields a high goodness of fit, which is exceptional for a model estimated in first differences. Also the
Durbin Watson statistic is reasonable, indicating that autocorrelation is not a serious problem. Crucial
parameters with respect to generating the long-run elasticities are γKK , γLL and γKL . Although all
three parameters have a plausible sign (γKK and γLL should be non-positive), only γLL is
significantly different from zero.
Discussion
The associated short-run price and fixed factor elasticities are given in Table 3. In comparison
with the levels version of the model, as well as in comparison with the elasticities found for France in
the Bouchet et al study, the own-price responses are very low. With exception of the own price
responses for meat and fodder, all absolute own price elasticities are smaller than 0.1 and in 3 out of 6
cases even below 0.05. As such our results confirm the results found for the period 1949-1983 by
Oskam (1991, 68). With respect to arable and dairy farming the connectedness to land (ARABL and
GRASSL) is clear. This is not only reflected in the significant (positive) production elasticities, but
might also explain the low own price responses and non-significant cross price elasticities between
both sectors. Arable land and pasture land have both a rather permanent character in the Netherlands
and operate as stable quasi-fixed factors. Meat production, which is less connected to land, has the
highest relative own price response (0.133), although it is still highly inelastic. Non-surprisingly, meat
output shows the most pronounced sensitivity with respect to feed price. Depending on the type of
meat production, feed costs can have a share up till 70% of total production costs.