Those who collect taxes use transfers better: Evidence of decentralization design and service outcomes in Bolivia

Abstract

This paper examines whether the design elements of decentralization—own-revenue effort, operations and maintenance (opex), capital expenditure (capex), and total execution—help explain municipal differences in poverty-relevant service conditions and their downstream influence on human capital and the local economy. A recursive SEM is estimated within departments, with services defined as an SDG1-based composite. Robustness replaces the mediator with a basic-infrastructure services composite (biservices1) and reparameterizes execution as total executed expenditure per capita. Four results stand out: (i) own-revenue effort is the strongest predictor of services, while execution scale is positive but smaller; in composition, opex—not capex—supports services; (ii) capex influences the economy directly, consistent with an investment pass-through; (iii) higher services raise the predicted level of human-capital and economic outcomes in the model, with the former path larger; and (iv) population scale and density matter. The pattern is consistent with a flypaper-with-effort interpretation: where fiscal effort and operations and maintenance (O&M) discipline are present, available resources—including transfers—translate more effectively into poverty-relevant service conditions. At the same time, investment directly influences the economy. Estimates are directed influences within the maintained model, not counterfactual causal effects. Policy implications: (i) embed effort-compatible transfers so own-source revenue unlocks additional grant resources; (ii) protect O&M floors to keep assets working; (iii) pair new capex with credible O&M plans; and (iv) keep services abreast of agglomeration.