Predicting the supply of ecosystem services from agriculture

A socially efficient agricultural policy wouldprovide incentives for farmers to supply theappropriate combination of market and non-market goods demanded by society (Antle andCapalbo 2002). Ecosystem services (ES) arethe most important type of public goods pro-vided by agriculture. To support the transi-tion from commodity-based subsidy policiesto policies based on the efficient provisionof ES, policy decision makers will need toknow how to design efficient mechanisms forthe provision of ES from agriculture, and willneed estimates of their environmental and eco-nomic effects. For example, the 2002 Farm Billcreated the Conservation Security Program,which pays farmers who adopt environmen-tally beneficial practices. The ten-year costs ofimplementing this program were estimated tobe $2.1 billion in 2002, but rose to $8.9 billionin 2004. This cost underestimation was due inpart to inaccurate estimates of farmer partici-pation (General Accounting Office 2006). Thisexample suggests that estimates of benefits andcosts of these policies with a reasonable degreeof accuracy will be needed to facilitate the tran-sition to policies based on provision of ES. Atthe same time, to be useful for policy decisionmaking, information needs to be provided ina timely manner. Researchers will inevitablyneed to trade-off cost, timeliness, and accu-racy in making decisions about the appropriatemodeling approach.The purpose of this paper is to present aconceptual framework for the analysis of ESsupply, and to discuss some of the data andmodeling issues that arise in predicting farmers’ participation in ecosystem service con-tracts and the supply of ES resulting fromthem. Two key features of agro-ecosystemsthat have been identified in the scientific lit-eratures are the complexity of the biophys-ical and human environments in which theyoperate, and the complexity of the ecologi-cal and economic processes governing the sys-tems. To frame the discussion, we outline amodel of the supply of ES that illustrates theways that spatial and system complexity canaffect it. We use this model to discuss thedata and modeling issues that arise in empiri-cal implementation. We conclude with recom-mendations regarding modeling strategies thatcan provide sufficiently accurate and timely in-formation needed to support informed policydecision making.