Association for Public Policy Analysis & Management

Management of common pool resources generally requires a method for providing consistent feedback on resource conditions and adherence to rules-in-use (Ostrom 1990). Monitoring strategies are often designed to provide empirical estimates of resource quality and maintenance; however, citizens can also co-produce monitoring with government (Bovaird 2007). Does citizen co-production of species monitoring enhance species protections in oil and gas drilling plans? We provide a rationale for how co-production could promote different oil and gas drilling plans. First, monitoring (even without co-production) can provide new information to policy discussions, or trigger implementation of additional regulatory requirements (Scott 2019). However, citizen monitoring can also provide an avenue to citizen participation in governance, improve citizen knowledge of impacts (Cornwell and Campbell 2012; Scott 2019), and facilitate self-organized communities of citizens with common interests (Carton and Ache 2017). Thus, we hypothesize citizen participation in monitoring will correspond to outputs that reflect management practices that are more likely to protect community assets.

We test this theory in a specific case: co-production of wildlife monitoring and site-specific planning for energy development in the intermountain west. Focusing on Colorado, New Mexico, Utah, Montana, and Wyoming, we evaluate how co-production of wildlife occurrence observations affects state and federal policy adoption of oil and gas drilling controls. We utilize year-specific spatial grid cells as the unit of analysis. For an individual grid of land in a year, we model the probability of state or federal policy protecting species as a function of monitoring program characteristics. For the dependent variable, we use text from land use plans, drilling permits, and activity permits, with the proportion of text referring to wildlife management protocols and externality mitigations as the measure of a policy output. This probability is generated via supervised machine learning via Keras, with word embeddings and the natural language structure of text as the key textual inputs. Finally, we utilize a multilevel Bayesian model to evaluate how use of co-production, private contractors, or public monitors, the diversity and density of participants in monitoring (count of unique organizations, utilization of citizens v. professionals), and identification of species (occurrence) relates to changes in surface development mitigation plans at individual oil and gas leases. The results of this study are critical for understanding how co-production as a form of participation can inform downstream public policy outputs.