Intergovernmental Collaboration and Policy Outcomes: A Causal Network Approach to Drinking Water Governance

In the environmental domain, intergovernmental collaboration—formal partnerships between municipalities, regional authorities, and national actors—has become a central mechanism for governing complex socio-ecological problems such as water governance. Despite its widespread adoption, however, empirical evidence on the effectiveness of these collaborative ties remains limited. The key identification challenge is that collaboration networks are rarely formed at random: governments self-select into partnerships based on shared problems, political connections, and institutional constraints, making it difficult to distinguish the causal effects of network participation from selection processes. This paper advances political network research by developing a causal strategy to estimate the effects of intergovernmental collaboration networks on policy outcomes while explicitly addressing endogenous tie formation and multiplex network dynamics. Substantively, the paper examines intergovernmental collaboration in municipal drinking water governance in Colombia, a decentralised system characterised by overlapping jurisdictions and strong interdependence across tiers of government. Drinking water provision requires municipalities to coordinate with regional environmental authorities and national agencies through formal interadministrative agreements, producing a complex, multilevel network of collaborative ties. Drawing on the universe of public contracts recorded in Colombia’s national procurement system, I identify and classify more than 4,700 intergovernmental collaboration agreements using an LLM-based text classification procedure. These data are used to construct municipality-level measures of network participation along both the extensive margin (entry into collaboration) and the intensive margin (number of collaborations and total collaborative spending). To analyse the causal effects of this network, I develop a novel instrumental-variables strategy for causal network analysis. To address the endogeneity of collaboration ties, I exploit the fact that municipalities participate simultaneously in multiple policy networks. Specifically, I construct instruments from an exogenous network of non-water collaborations (e.g. land use, waste management, forestry), which shape structural opportunities to collaborate but do not directly affect drinking water quality. Using a friction-only gravity model based on geographic and administrative distance, I estimate each municipality’s exogenous propensity to form collaborative ties. I then combine these propensities with partner-level spending shocks in a shift-share design to instrument collaboration intensity. This approach isolates variation in network exposure driven by exogenous network frictions and external shocks, rather than by political selection or policy need. The results show that entering collaboration entails significant short-term costs, reflected in initial declines in water quality, consistent with high coordination and setup costs. However, deeper and more sustained collaboration—measured by additional partnerships and greater collaborative investment—yields substantial improvements in service outcomes. Methodologically, the paper contributes a new approach to estimating network effects using multiplex administrative networks and exogenous exposure designs. Substantively, it demonstrates that collaboration networks can generate both negative and positive policy effects over time: while entry into collaboration may initially worsen outcomes, sustained and intensive collaborative engagement can produce meaningful performance gains. These findings highlight the importance of distinguishing network formation from network effects when evaluating the political consequences of collaborative governance.