A Spatially Integrated Economic-Ecological Model of Farmers’ Land Management Decisions and Water Quality Outcomes in Lake Erie

Agricultural runoff is a major source of non-point source pollution to the Great Lakes with harmful impact on the water quality. Nutrient runoff, especially phosphorus, from Maumee River watershed entering the western Lake Erie basin, has led to frequent and severe water quality crises, including harmful algal blooms (HABs), hypoxia in Lake Erie, and the 2014 Toledo water crisis (Lake Erie LaMP 2011; Scavia et al. 2014; Stumpf et al. 2012). To address these growing concerns, the Great Lakes Water Quality Agreement (GLWQA), committed by both United States and Canada, adopted targets to reduce total phosphorus (TP) and dissolved reactive phosphorus (DRP) entering affected areas of Lake Erie by 40 percent based on 2008 loading levels. In this paper, we calculate the tradeoff frontier between cost-share program costs and phosphorus reduction in Lake Erie by quantifying the impact of uniform and targeted cost-share policies on best management practices (BMP) adoption and the resulting water quality change. Previous analysis using multiple variations of a Soil and Water Assessment Tool (SWAT) hydrological model found that subsurface placement of fertilizer is one of the most successful pathways to reduce P runoff (Scavia et al. 2017). Therefore, we specifically focus on the cost-effectiveness of payment policies for subsurface placement adoption and their impact on P reduction.

A full understand of the coupled human-natural system requires not only a hydrological model of the watershed to capture nutrient loadings and impacts on water quality of the lake, but also a model of farmer decision making. While farmers are motivated by profit maximization, some evidence suggests that farmers are heterogeneous in their land management decisions and motivations for engaging in more environmentally friendly practices (Zhang et al. 2016). To account for the influence of farmer land management decisions on water quality in Lake Erie, we develop an integrated economic and ecological model that explicitly incorporates a discrete choice model of heterogeneous farmer decision making at the field level. We use survey data on farmer cropping and land management decisions taken from a random sample of over 1,800 corn and soybean farmers within the Maumee River watershed in the western Lake Erie basin (Zhang et al. 2016; Martin et al. 2011). We first quantify heterogeneous adoption costs of subsurface placement based on farmer and field level characteristics. We then use this adoption cost variable along with other socio-economic and spatial characteristics to estimate a discrete choice model of subsurface placement adoption at the field level and predict the likelihood of adoption of subsurface placement under different incentive policy scenarios, assuming a uniform payment that ranges from $1 to $80 per acre. We use county-level data on farmer characteristics to spatially allocate the field-level predictions and integrate the resulting predictions of land area management practices into a SWAT watershed model using Hydrological Response Units (HRU) as the spatial unit of observation. Holding other land use, management, and climate condition at baseline conditions, we use the SWAT model to simulate water quality outcomes in Lake Erie (Gildow et al., Aloysius et al. (in prep).

Results from these policy scenarios show that the subsidy payment program can reduce lake phosphorus by up to 13%. Among them, the most cost-effective policy is the $35 per acre payment program, which costs 120 million dollars in total and reduces TP and DRP by 5% and 10% respectively. Thus, while payment for a single BMP can reduce loadings substantially, a uniform cost-share program that subsidizes subsurface placement alone cannot reach the reduction goal set by GLWQA. In subsequent work we explore the cost-effectiveness of a cost-share program that incentivizes a combination of BMPs, including cover crops, and compare this with a policy scenario in which a fertilizer tax is also imposed. Finally, we compare the results of these uniform cost-share policies with targeted policies based on farmer, field characteristics, and nutrient runoff hotspots, and examine whether spatial targeting can achieve the desired policy goal and compare the cost-effectiveness of doing so.