Grants from public agencies often support collaborative environmental management groups.  There is little evidence, however, about the extent to which these grants improve environmental outcomes. The extant literature hypothesizes that grants for ‘capacity building’ and other institutional purposes have indirect, long-term effects that are not immediately observable. For instance, the Puget Sound Partnership in Washington state and the Oregon Watershed Enhancement Board (OWEB) both provide grants to hire local watershed council coordinators, assuming that institutional changes generated by a full-time coordinator ultimately manifest in better environmental conditions. In this study I test this assumption using time-series Oregon watershed quality and grant data to explore the following question: To what extent do specific grant types -and features- affect long-term environmental conditions? 

Research Design

I use a Bayesian repeated-measurement multilevel regression model, with various watershed quality metrics as the dependent variable (e.g., chemical content), to model the temporal relationship between grant receipt and watershed condition. The three-level approach models time-specific observations as nested within a watershed, and watersheds as nested within basins. The advantage of a multilevel model relative to a fixed-effects method is that variation can be modeled as a function of covariates at each level of the model. Further, using a Bayesian formulation accounts for potential measurement errors and sparseness associated with empirical environmental modeling data by examining the probability of parameter estimates given observed data. 

The data I use for this study begin with 20 years of biennial Clean Water Act (CWA) Section 303(d) and 305(b) water quality data (1993 to 2013) obtained from the Oregon Department of Environmental Quality (ODEQ). I then collect the grants awarded to 87 local watershed councils from OWEB starting in 1995 and code for grant characteristics such as type, purpose, performance metrics, and deliverables. With these data, along with additional covariates such as land use, climate, and spatial relationships, I fit a nested three-level model in which: (1) the first level fits watershed condition for a specific biennium as a function of past condition, time-variant watershed characteristics (e.g., yearly precipitation, population), grant receipt and characteristics (both in current and preceding years) plus a random effect for each watershed; (2) the second level models this random watershed-specific effect as a function of time-invariant watershed characteristics (e.g., watershed size, council characteristics) and a large basin random effect (15 large basins); and (3) the third level models the large basin random effect as a function of basin features and basin-wide grant receipt and characteristics (OWEB awards grants collectively to whole basins as well). 

Implications for Research and Policy

The extant literature lacks evidence regarding the outcomes associated with collaborative management, in large part because of the indirect relationship between administrative and institutional changes and environmental outcomes. This analysis provides empirical evidence of the environmental outcomes resultant from public investment in collaborative approaches. Further, by comparing and contrasting the short and long term impacts of different grant types, this analysis directly informs policy makers about how to best allocate limited resources.