Information Technology Reference
In-Depth Information
with similar characteristics: returning high profits, but requiring high labor loads and
having high water needs (resulting in illegal behavior). The data used for this model
slightly favor irrigated garlic especially for small farmers due to little higher profits
and less risks. This advantage by a narrow margin has strong effects in the utility-
based model: at the end of the simulation the area of irrigated garlic is almost twice as
large as that of irrigated melons, whereas the amount of irrigated garlic and irrigated
melons is almost equal throughout the simulation in the satisficing version. If farmers
don't optimize they are almost indifferent between options that contain almost equal
characteristics (such as melons and garlic). The cautious conclusion from the
comparison of model performance is that the most important factors influencing
farmers' decision-making in the satisficing model may be closer to reality than those
which determine farmer behavior in the utility-based model. The following analyses
the differences.
When comparing the models a striking difference can be observed in the model
dynamics. The use of satisficing with repeated decisions leads to less sensitive model
dynamics. If no constant increase of profit thresholds is assumed in the satisficing
model, a farmer has no incentive to change once he has found a satisfying pattern
(unless external influences worsen his situation). Even with the assumption of
increasing profit thresholds farmers rethink their decisions less often compared to the
utility-based model, where farmers optimize and rethink their decision every year.
This leads to a slow-down of the diffusion process of innovations in the satisficing
model. Thus, parameters that slowed down the diffusion process in the utility-based
model to an empirically observable speed
6
, namely the distance parameters α
s
and α
c
(cf. section 3.2), are estimated much lower in the satisficing model. The interpretation
in the utility-based model is that switching costs and learning efforts delay the
diffusion process, while the satisficing model suggests that farmers not actively
looking for improvements are the main reason for slow diffusion.
Another specific characteristic of the satisficing concept is that it is a non-
compensatory strategy, which means that a very high value of an option with regard
to one objective cannot compensate other values that are below the respective
threshold. For example, if a land-use pattern is evaluated as being very risky, other
attributes such as very high profit cannot compensate this, the option will not be
chosen. Thus satisficing favors “all-round options", which contain no negative outlier.
For example, the area of irrigated garlic is significantly higher in the utility-based
model since it returns higher profits and therefore compensates for high labor loads
and for an illegal amount of necessary groundwater extractions. The satisficing model
shows less irrigated garlic due to the fact that irrigated garlic includes high labor loads
and illegal behavior. As a consequence of the compensatory nature of the utility-based
model the maximum work-load κ (cf. section 3.3) played a significant role to fit the
model to comparably small areas of horticultural crops. These provide a very high
profit which overcompensates for illegal behavior, high risk and high labor loads in
the utility-based model. Setting κ to a value that does not allow for larger areas of
horticultural crops due to non-availability of labor was virtually the only possibility to
roughly approximate empirical data. The interpretation then is that the available labor
on (family and part-time) farms sets a limit to the area of irrigated horticultural crops.
6
Cf. the sensitivity analysis in [10].
