Program 1

Tradeoffs in Agriculture, the Environment, and Farmer Health
C.C. Crissman¹, J.M. Antle², and S.M. Capalbo²

An Integrated Agriculture/Environment Model
The Pesticides Effect Study
Presentation of Results
Implications for Policy/Technology Analysis
Conclusions
Selected Reading

Over the past two decades, sustainability has come to mean for many the maintenance and improvement in human living standards, standards that are affected by changes in the environment and in human health and economic status. Incorporating sustainability criteria into the mandate of CIP implies that the Center should critically examine not only the productivity but the potential environmental and health effects of its research. In the heterogeneous and frequently fragile environments that are among the priority regions for CIP, there are considerable tradeoffs between productivity gains and their potential environmental and health effects. This paper reports on a general approach to measure the economic, environmental, and health tradeoffs associated with agricultural technologies and how those tradeoffs may be altered through technology or policy changes. The research reported here demonstrates this approach in a study of the environmental and health consequences of pesticide use in potato production in Ecuador.

An Integrated Agriculture/Environment Model

Policy analysis is typically undertaken nationally or regionally. The analysis frequently uses secondary data that reflect aggregates. However, policy analysis with sustainable agricultural development objectives must include environmental effects that are location-specific. Policy analysis in the heterogeneous biophysical setting of the tropical mountains of Ecuador suffers serious deficiencies from the averaging effects of using data that are not location-linked. One objective of the research on pesticide effects is to improve the potential for policy analysis with sustainable agriculture criteria by developing and implementing a framework that links macrotype policy to microtype effects.

The method used here is to define a common unit of measurement valid to the different disciplines and predict the effects of technological or policy changes on those units. In this case, the unit is a farm field. By describing the population of these units statistically and estimating effects on each unit, it is possible to aggregate those effects to a level useful for policy analysis. Using this information, one can define aggregate tradeoffs between economic and environmental or health outcomes in the form of a tradeoff curve.

Figure 1 depicts a model of land use and crop management decision-making. The upper part of the figure pertains to the analysis of a unit of land at the farm level. Prevailing prices, policies, technologies, and the physical attributes of the unit of land affect the farmers’ management decisions on land use and input use. These decisions affect agricultural production, but also may affect the environment and human health through two distinct but interrelated mechanisms. Farmers first determine which units of agricultural land are put into production-the land use decision. Then, on the land in production, farmers make management decisions that determine the application rates of chemicals, water use, and land management practices-the input use decision. Physical relationships between the environmental attributes of the land in production and the management practices then jointly determine the agricultural output, and human health and environmental effects associated with the particular unit of land in production. Thus, the land use and input use decisions of farmers form the linkage between policy and technology and the environmental and health consequences.

 

The Pesticides Effect Study

   The case study
The case study was conducted in a 70-km² watershed in the principal potato-producing zone of Ecuador. Production data were collected in a farm-level survey on 40 farms. Because crops are planted and harvested continuously throughout the calendar year, data were collected for parcels, where a parcel is defined as a single crop cycle on a farmer’s field. Detailed parcel-level data were collected monthly. Potato production in Ecuador is management-intensive, and there are as many as 20 distinct operations during the 6-month crop cycle.

Late blight caused by Phytophthora infestans is the principal disease, and the tuber-boring Andean weevil (Premnotrypes vorax) and several foliage-damaging insects are the principal pests affecting production. They are controlled with dithiocarbamate fungicides and neurotoxin insecticides from the organophosphate and carbamate families. The case study focused on groundwater contamination through leaching and the occupational health issue of the pesticide applier’s exposure as the expressions of environmental and health consequences of pesticide use.

   The simulation model
The decision-making model of Figure 1 is embedded into a simulation model (Figure 2) for empirical application. The simulation model is based on econometric revenue and production models, a soil pesticide leaching model, and a health effects model. Figure 2 illustrates the flow of the economics, pesticide leaching, and health portions of the simulation model. A policy or technology scenario is imposed on the economic model where the unit of analysis is a parcel of land. The economics portion consists of four components. First, the model is initiated by sampling the economic and physical characteristics of the fields in the study watershed. Second, we obtain net returns distributions of the principal crops in the rotation (potatoes and pasture for milk production), which determine the land use decision. In the third component, a potato production model produces estimates of the quantity and frequency of pesticides used. The fourth component is a restricted revenue function used to predict the value of production. As shown in Figure 2, the economic model generates three types of outputs that are used in the subsequent portions of the simulation model.

 

The leaching model is a detailed process model that uses soils and other physical data from the watershed, the chemical characteristics of the pesticides, the pesticide applications from the economics portion of the model, and more than two decades of daily rainfall, temperature, and evapotranspiration data from weather stations in the area. The leaching model predicts the downward movement of pesticide active ingredients through the different soil horizons.

The health component of the simulation model consists of an estimated health production function that specifies health as a function of the total number of pesticide applications, total quantity of applied neurotoxic substances to which an individual was exposed, and other factors. We measure health as an individual’s mean neurobehavioral score, an index constructed from a series of neurobehavioral tests. Neurobehavioral tests measure specific aspects of cognitive function such as attention span, visuospatial memory, and reaction time, which are important for decision-making and daily performance of farm work. We also measured poisoning rates to establish the size of the problem.

Because we use mutually compatible data sets, the three separate disciplinary models produce results that are linked. As the last box in the figure shows, we use the linked results to compare the tradeoffs in gains and losses among farm revenues, water table contamination, and applicator health.

Note that the model is stochastic. At several points, the model samples distributions that are derived from physical and economic data. Thus, any two runs of the model are not expected to produce precisely the same results. This is advantageous in two ways. First, since economic and physical processes are stochastic, the model can produce all outcomes in addition to the expected average outcomes. Second, by using distributions constructed from the sample data, you can appeal to rules of aggregation to obtain summary statistics for the group. Reasonable assumptions about the structure of production permit statistically valid extrapolation beyond the data collection site, an important implication for policy analysis.

Presentation of Results

Use of tradeoff curves
To facilitate interpretation of the model results, the tradeoffs between agriculture, health, and the environment are presented in a series of pairwise comparisons. The hypothetical tradeoff curves in Figure 3 illustrate the tradeoffs between agricultural output and environmental effect. The tradeoff curve represents all the possible pairs of outcomes for a given technology. Thus, different curves are available for different technologies. The move from T1 to T2 shows a change in technology that everywhere maintains output while reducing environmental effect. The slope of the tradeoff curve provides information about the opportunity cost of environmental quality in terms of lost output. As curve T3 shows, different technologies may be more or less damaging at different levels of output.

 

Construction of an empirical tradeoff curve
An empirical curve is constructed by imposing different policy scenarios on the model. Figure 4 is a scatter diagram of the comparison that relates the level of fungicide leaching with the value of agricultural production. The squares are the base case produced with simulations using the actual data set. Each square represents the average outcome of 30 fields each with 5 production cycles. This is repeated 30 times. Changing relative prices causes movement along a curve. For example, when the levels of fungicide tax are imposed on the model, there is a reduction in fungicide use with a consequent reduction in fungicide leaching and the value of agricultural production.

 

The reduced value of potato production comes from two sources. First, farmers leave more area in pasture, which produces less revenue than potatoes; second, they use less fungicide on potatoes, which reduces yields. The figure also depicts increases in potato prices. Higher prices increase profitability, thus causing farmers to plant more area to potato and to use more fungicides. That in turn increases fungicide leaching and the value of production.

Implications for Policy/Technology Analysis

Several issues relevant to policy analysis can be addressed by tradeoff curves. Politicians implicitly use tradeoff curves every day. By the nature of their jobs, they are concerned with winners and losers resulting from policy decisions. The tradeoff curve is simply a concrete expression of what is usually a mental calculation. In the example, the politician or analyst can readily see what the sacrifice of a single unit of environmental quality will gain in units of agricultural production. Whether the size of the sacrifice is acceptable becomes a political decision.

Case study results
First, because of the short half life of the pesticides in use, the fixation of these pesticides to the organic matter in the soil, and the frequent but light rainfall that moves the pesticides slowly downward, water contamination from pesticide leaching is not significant. At four parts per billion, carbofuran contamination of the water table is a full magnitude below the USEPA tolerance limit of 40 ppb. Second, at 171 per hundred thousand inhabitants, the rate of work-related pesticide poisonings ranks among the highest recorded in the world. Third, the health effects of chronic exposure severely depress the neurobehavioral performance of a large segment of the applicators and farm families. Fourth, our economic efficiency analysis shows that farmers are not overspending and applying pesticides irrationally.

These results run counter to public perceptions in Ecuador. Farmers are thought to make irrational, excessive use of pesticides with resulting widespread environmental contamination. Environmental lobbyists put forward proposals to ban certain classes of pesticides, including carbofuran, based on the presumption of environmental contamination. However, the research shows that a ban on carbofuran would reduce potato production, but would only minimally improve the environment.

The worker-safety lobby in Ecuador is not nearly as well organized as the environmental lobby, so there is little public discussion about occupational health issues. With pesticide use an essential part of efficient potato production, the policy debate should be centered primarily on the safety of the pesticide applicator and farm families and integrated pest management (IPM) techniques.

Heterogeneity of environment
Policy analysis that incorporates environmental effects must also consider the variability of the impact of the policy itself. The research watershed was classified into four microregions (Figure 5) where we plot lines through the scatter plots to simplify their interpretation. Three of the microregions have similar low-leaching characteristics; the fourth zone is much more susceptible to leaching. There is considerable heterogeneity present even in small areas. We can differentiate such phenomena by linking environmental data to economic data. With such differentiation, regulations to improve the environment could be based on agricultural zoning rather than on taxes or broad-based prohibitions on pesticides that uniformly affect all production zones.

  

Technology effect analysis
The current technology to control the Andean weevil relies on applications of carbofuran. CIP entomologists have developed a set of IPM technologies that reduce the reliance on carbofuran. Here we simulate the effect on health risk of an 80% adoption of IPM practices that reduce carbofuran use by 40%, combined with a worker education program for safe handling practices (Figure 6). Health risk is the percent chance of a one standard deviation decrease in neurobehavioral score below the mean score of the nonfarming urban control population. The analysis suggests that the combination of IPM and improved safety practices could reduce health risk by 50% or more.

 

Conclusions

Including sustainability criteria into policy decisions in both developed and developing countries has spurred substantial changes in the mandates of agricultural research institutions and the manner in which research is conducted. One outcome of these changes is an increased emphasis on systems modeling. A weak link in most systems modeling work is the lack of economic criteria in the models.

The research reported in this paper developed a conceptual model that provides linkages between macropolicy and microtype effects. The linkage is provided by including a farmer decision-making model. That model shows how farmers react to policy or technology changes through adjustments to land use and input use decisions. The simulation model produces results that can be aggregated for extrapolation beyond the actual case study area, thus making the model useful for policy analysis. Tradeoff curves are introduced as an analytical tool for summarizing large amounts of data and for illustrating the multiple effects of a given technology.

Although the conceptual model is applied in an economics, pesticide leaching, and health study of pesticide effects, the model is flexible and can be applied to other research questions such as fertilizer use, erosion, or output price adjustments. Since the model is statistically based, it can use data sets generated from various sources.

Selected Reading

Antle, J.M., S.M. Capalbo, and C.C. Crissman. 1994. Econometric production models with endogenous input timing: An application to Ecuadorian potato production. J. Agric. Res. Econ. 19(1):1-18.

Antle, J.M., C.C. Crissman, R.J. Wagenet, and J.L. Hutson. 1996. Empirical foundations for environment-trade linkages: Implications of an Andean study. In: M.E. Bredahl, N. Ballenger, J. Dunmore, and T.L. Roe (eds.). Agriculture, trade, and the environment: Discovering and measuring the critical linkages. Boulder, Col.: Westview Press. p. 173-197.

Crissman, C.C., D.C. Cole, and F. Carpio. 1994. Pesticide use and farm worker health in Ecuadorian potato production. Am. J. Agric. Econ. 76(3):593-597.

---

1 CIP, Ecuador Liaison Office.
2 Montana State University, Bozeman, Montana, USA.