Date of Award
2025
Degree Type
Dissertation
Degree Name
Doctor of Philosophy in Environmental and Natural Resource Economics
Department
Environmental & Natural Resource Economics
First Advisor
Pengfei Liu
Abstract
As climate change and resource scarcity drive a global shift toward cleaner energy systems and environmental conservation, understanding how individuals value sustainability-related interventions has become increasingly important. This dissertation contributes to this growing body of knowledge by applying and advancing discrete choice experiment (DCE) methodology to assess consumer and public preferences for two key environmental policy domains: (1) repurposed electric vehicle (EV) batteries for residential energy storage and (2) aquatic invasive species (AIS) management in New York State. Across all three chapters, the study examines how traditional econometric models and emerging machine learning (ML) methods can complement each other in modeling choice behavior, enhancing prediction accuracy, and improving the transferability of results.
Chapter 1 investigates consumer preferences for second-life EV batteries repurposed for residential photovoltaic (PV) systems. As EV adoption rises globally, so too does the need to manage the retirement of lithium-ion batteries in a manner that is both environmentally sustainable and economically viable. Using a DCE, respondents were asked to choose between battery options that varied in cost, lifespan, safety, storage capacity, and whether the battery was new or repurposed. The analysis, based on a conditional logit model, finds that while consumers generally prefer longer-lasting and lower-cost batteries, many are open to using repurposed batteries, particularly when they understand the tradeoffs involved. Demographic factors such as income, education, and homeownership influence willingness to adopt second-life batteries. These insights can guide both public policy, through incentives for battery reuse, and private sector strategies aimed at expanding access to affordable and sustainable energy storage solutions.
Chapter 2 shifts focus to environmental conservation and the management of aquatic invasive species. Employing a separate DCE targeting recreational users in New York’s Hudson and Mohawk Valleys, this chapter quantifies preferences and willingness to pay for various AIS management strategies, with a particular focus on controlling European water chestnuts. Key attributes include water clarity, fishing catch rate, ease of boat launch, and the control method used (mechanical, manual, or chemical). Findings indicate that environmental quality attributes, especially water clarity and fish catch rate, strongly influence preferences. Demographic factors, along with the framing of information (information treatment), also shape support for different control strategies. This chapter provides insights for environmental managers and policymakers, suggesting that emphasizing visible ecological benefits may improve public buy-in for AIS control programs.
Lastly, Chapter 3 introduces machine learning (ML) models as methodological complements to traditional discrete choice models in analyzing DCE data. Using the same AIS dataset from Chapter 2, the study compares the predictive performance of a conditional logit model with four ML approaches: Lasso regression, Ridge regression, Elastic Net, and Random Forest. Results show that while the conditional logit model provides a solid baseline with interpretable coefficients, ML models demonstrate improved predictive accuracy, better handling of attribute collinearity, and enhanced variable selection. This chapter also highlights how ML techniques can support benefit transfer, a practice commonly used in environmental economics to apply estimated preferences from one study context to another. ML models, by capturing complex patterns in high-dimensional data, may increase the reliability of value estimates in new settings where conducting a full DCE is not feasible.
In summary, this dissertation makes both substantive and methodological contributions to the field of environmental and natural resource economics. Substantively, it deepens understanding of public preferences for two timely environmental challenges: battery repurposing and invasive species management. Methodologically, it demonstrates the value of integrating machine learning with traditional econometric models to improve the robustness and transferability of stated preference results. By combining rigorous quantitative methods with real-world environmental applications, this work supports more informed, data-driven decision-making in the pursuit of sustainability.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Liu, Lufan Hellen, "CONSUMER AND ENVIRONMENTAL PREFERENCES: DISCRETE CHOICE EXPERIMENTS AND MACHINE LEARNING APPLICATIONS IN EV BATTERY ADOPTION AND AQUATIC INVASIVE SPECIES MANAGEMENT" (2025). Open Access Dissertations. Paper 4510.
https://digitalcommons.uri.edu/oa_diss/4510