MODELLING ENERGY SAVING TECHNOLOGY CHOICES IN DUTCH GLASSHOUSE HORTICULTURE - Economics of Sustainable Energy in Agriculture - page 50

the management level and personal preferences. Firm specific effects are
related to the location of the firm (affecting climate and access and
availability of hired labour) and the firm financial structure. Serial
correlation accounts for adjustment costs that cause a persistence of
technology choices over time. A problem with the Simulated Maximum
Likelihood method in this study was that not all parameters reflecting the
proposed error structure could be identified.
Future research should extend the scope of the research by
modelling the choice of the technology and the size of the required
investment simultaneously. Different results may be expected, if required
investments substantially differ between technologies. Furthermore, future
research should pay attention to the role of other factors in analysing
technology choice decisions and in particular to the role of information. In
a situation where investments are (partly) irreversible and where the firm
operator has the possibility to postpone the investment, information that
reduces uncertainty may give rise to an option value (Dixit and Pindyck,
1994). Such an option value implicitly increases the costs of investments
and may hamper the adoption of new (energy saving) technologies.

NOTES

1 Co-generators produce electricity and heat simultaneously.
2 In the literature, computational problems of solving the structural model are reduced by
using convenient functional forms for the reward functions and error distributions ( e.g. Rust
1987), or by using simulation and interpolation techniques (Keane and Wolpin 1994) and
methods that do not require a full solution of the dynamic programming model (Hotz and
Miller 1993).
3 The accuracy of the reduced form approximation is a standard functional specification
problem. Approximation error can be decreased by using flexible functional forms.
However, in the empirical application of this chapter, augmenting the model by quadratic
terms resulted in numerical identification problems.
4 The specification in (4.5) is consistent with irreversible and reversible investments, since
the estimation method maximizes the likelihood of the observed technology choices.
Therefore, the reversibility or irreversibility of investments will show up in the parameter
estimates.
5 Other energy saving options available in horticulture at present are heat pump and heat
deliveries by electricity plants. However, these options were used by none and only few
firms in the sample period, respectively.
6 McFadden's

is calculated as: where log is the value of log-

likelihood function subject to the constraint that all regression coefficients except the
constant term are zero, and log is the maximum value of the log-likelihood function
without constraints (Veall and Zimmerman 1996). McFadden

in the range of 0.2 to 0.4

are typical for logit models (Sonka et al. 1989).
7 An ad valorem tax on energy increases the price that energy users pay by a given
percentage.