between innovations. The effect of shelf-life extension could not be tested as it
was unclear if this innovation would be used by farmers or middlemen; therefore, we assessed what the price premium would have to be in order to maintain
the existing litchi area.
Outcome indicators used to assess the simulation output included the area under
litchi trees, net household incomes, the potential risk of pesticide use as calculated by
the Environmental Impact Quotient (EIQ) method (Kovach et al. 1992) and erosion
soil loss quantified using the Revised Universal Soil Loss Equation (RUSLE).
10.7.3.2 Data
All economic information for designing and parameterizing the agent decision
model came from farm level data based on a random sample of 303 farm
households interviewed in October/November 2006. Different from previous MPMAS applications (Berger and Schreinemachers 2006), we used sample weights to
duplicate farm households, creating a total population of 1,309 agents. We furthermore used secondary data to parameterize crop water requirements, precipitation
and the potential environmental impact of pesticides.
10.7.4 Main Findings
We validated the model by comparing observed and simulated values for land use,
and found a close fit (R
2 above 90 %) between these values at various levels of
aggregation, but the fit at the agent/household level was not as good
(Schreinemachers et al. 2010). This shows that each agent in the model did not
exactly represent a real-world farm household, but that at the watershed, village or
group level, agents on average were a good representation of reality.
The impact of the innovations was assessed by comparing a baseline scenario –
representing current conditions in the absence of innovations, with alternative
scenarios which included the innovations. The results of the baseline scenario
showed that the litchi area would decline annually by 2.3 %, while the general
intensification of land use would increase incomes by 3.4 %, pesticide loads by
about 3.5 % and soil erosion by 1.3 %.
As can be seen from Fig. 10.19, the introduction of each innovation reduced the
decline in litchi area, but the effect was weakest for improved irrigation and
strongest for cooperative fruit drying. If simultaneously introducing all three
innovations, then the average area under litchi was found to be 6 % greater than
in the baseline, but still declined over time in spite of a relatively successful
adoption of the three innovations, especially AFI. The three innovations were;
therefore, able to limit the decline in litchi orchards, but not reverse the trend,
416
C. Marohn et al.
was unclear if this innovation would be used by farmers or middlemen; therefore, we assessed what the price premium would have to be in order to maintain
the existing litchi area.
Outcome indicators used to assess the simulation output included the area under
litchi trees, net household incomes, the potential risk of pesticide use as calculated by
the Environmental Impact Quotient (EIQ) method (Kovach et al. 1992) and erosion
soil loss quantified using the Revised Universal Soil Loss Equation (RUSLE).
10.7.3.2 Data
All economic information for designing and parameterizing the agent decision
model came from farm level data based on a random sample of 303 farm
households interviewed in October/November 2006. Different from previous MPMAS applications (Berger and Schreinemachers 2006), we used sample weights to
duplicate farm households, creating a total population of 1,309 agents. We furthermore used secondary data to parameterize crop water requirements, precipitation
and the potential environmental impact of pesticides.
10.7.4 Main Findings
We validated the model by comparing observed and simulated values for land use,
and found a close fit (R
2 above 90 %) between these values at various levels of
aggregation, but the fit at the agent/household level was not as good
(Schreinemachers et al. 2010). This shows that each agent in the model did not
exactly represent a real-world farm household, but that at the watershed, village or
group level, agents on average were a good representation of reality.
The impact of the innovations was assessed by comparing a baseline scenario –
representing current conditions in the absence of innovations, with alternative
scenarios which included the innovations. The results of the baseline scenario
showed that the litchi area would decline annually by 2.3 %, while the general
intensification of land use would increase incomes by 3.4 %, pesticide loads by
about 3.5 % and soil erosion by 1.3 %.
As can be seen from Fig. 10.19, the introduction of each innovation reduced the
decline in litchi area, but the effect was weakest for improved irrigation and
strongest for cooperative fruit drying. If simultaneously introducing all three
innovations, then the average area under litchi was found to be 6 % greater than
in the baseline, but still declined over time in spite of a relatively successful
adoption of the three innovations, especially AFI. The three innovations were;
therefore, able to limit the decline in litchi orchards, but not reverse the trend,
416
C. Marohn et al.