A Theoretical and Empirical Investigation into Payments for Watershed Ecosystem Services

Abstract

This thesis contains three research chapters on the economics of Payments for Watershed Services (PWS). While each chapter covers a different aspect of PWS schemes, all three provide insights into the management of uncertainty in ecosystem services. The first chapter serves as an introduction to the problem and the main research question addressed by each paper.

Forested watersheds provide a variety of ecosystem services. Their economic valuation has increased significantly over the past decades, but the literature is fragmented and heterogenous and little has been done to systematically analyse estimated values. This study presents a global analysis of the economic valuation of forested watershed services. We address two methodological issues in the literature: the impact of ecosystem service classification on value estimates and sensitivity to scale. The latter is measured as the forested watershed area size compared to common practices to measure overall area size including other land cover and use. In the former case, we compare the detailed Common International Classification of Ecosystem Services with more simple and informal classifications in the literature based for example on the Millenium Ecosystem Assessment. We show that both the explanatory and predictive power of the estimated meta-regression models increase as we include more details about the valued ecosystem services and use more accurate estimations of the forested area size. Findings are, where possible, cross-validated with the existing forest hydrology literature. The study highlights the economic significance of maintaining forest cover in watershed areas and emphasises the role economics can play in determining high-value uses for ecosystem services.

The second chapter utilises the 2015 Survey of Drinking Water Plants, employing spatial regression models and mediation analysis to examine the relationships between land use, raw water quality, and treatment costs. The study reveals a significant and economically substantial impact of forest cover on reducing treatment costs, primarily through its influence on turbidity levels. Forest cover significantly reduces turbidity levels, thereby decreasing treatment costs. The results indicate that converting agricultural land to forest within a 5km buffer zone around a treatment facility can generate savings of $18.92 CAD per hectare per year, while the savings relative to urban cover can reach up to $21.37 CAD per hectare per year. These savings diminish as the distance from the facility increases, with lower per hectare savings observed at the 10km buffer and sub-sub drainage basin scales. Further, the study accounts for spatial auto-correlation and the effects of wildfires on treatment costs. Wildfires are shown to lead to substantial increases in turbidity, significantly impacting treatment costs. The spatial error and lag models used highlight the importance of considering spatial dependencies in ecosystem service valuations. This research underscores the economic value of forest conservation and management in supporting water treatment processes and provides valuable insights for policymakers and stakeholders in water resources management. The study represents a significant step forward in understanding the interplay between land use, water quality, and treatment costs in Canada.

The main objective of the final study presented here is to develop a novel modelling framework in the context of Payments for Ecosystem Services (PES) to mimic and simulate behaviour of agents (e.g., landowners) providing ecosystem services and a principal (e.g., government, municipality) buying them under uncertainty. Uncertainty is defined as the case where the principal and agents lack precise knowledge on the parameters that govern ES output, but knows the range these parameters belong to. We compare contracts under two different decision-making paradigms, namely standard and robust optimisation. With robust optimisation, an uncertainty-averse principal designs a contract to maximise their worst-case outcome and obtain a performance guarantee, i.e., a minimum acceptable performance. The results show that with standard, input-based contracts, the only way for a principal to achieve this guarantee is to invest conservatively in ES. In this setting, the result holds with adverse selection and moral hazard. However, when input is unobservable, using output-based payments, the principal can achieve this performance guarantee only by sharing some of the value of ES output with the landowner.