The Future of Energy: Three Essays on the Renewable Transition, Energy Conservation, and Electricity Planning

Abstract

This dissertation offers insights into the sustainable energy transition across diverse contexts, each presenting unique challenges that other regions and cities are likely to encounter as countries work to reduce global greenhouse gas emissions. To mitigate these emissions, many regions have chosen to reduce their reliance on fossil-fuel powered electricity generation in exchange for clean, renewable sources. However, because renewable energy power plants have a lower energy density than fossil fuel plants, the amount of land required to sustain this transition is significant. Rather than increasing renewable generation assets, other countries have sought to instead focus on electricity conservation and efficiency measures to limit carbon emissions by restraining load growth. This is particularly relevant for countries that rely mainly on carbon-intensive fossil-fuel powered generation. A third sustainable energy transition strategy involves the adoption of decentralized electricity generation assets, such as solar home systems (SHS). In addition to providing a clean source of electricity, these systems can also improve electricity connection reliability, particularly essential for countries with unreliable grid infrastructure. The challenge of land intensity for renewable power plants is particularly evident in California, where state legislation targets 2045 for 100% clean generation. These ambitious renewable energy targets will require that considerable land be allocated to solar-powered electricity generation. Due to its high-quality solar resource, the Central Valley (CV) is likely to play a key role in this transition. However, land conversion in the region could substantially reduce crop yields and revenue in one of the most agriculturally productive areas in the country. The decision of whether and how to protect farmland will thus be crucial to the state's agricultural output and environmental sustainability. My first chapter examines how land allocation to solar development and cropland differs under two decision-making perspectives, one that optimizes private benefits (the Land Manager) and another that considers social welfare (the Central Planner). We use a two-stage, spatially-explicit optimization model to simulate solar power capacity targets ranging from 5 to 50 GW, considering agricultural profits, solar revenues, and environmental benefits. This allows us to quantify economic tradeoffs across these various benefits and the gap between the social optimum and the arrangement preferred by private landholders. Our findings indicate that economic incentives will be needed to reduce the misalignment between privately and socially optimal solar siting, especially for capacity targets over 25 GW. In addition, socially optimal PV siting has the potential to reduce the overpumping of groundwater in critically overdrafted basins by up to 19% as the PV target increases.My second chapter explores heterogeneity in the impact of electricity conservation contests in Hanoi, Vietnam, using a combination of econometric and machine learning methods. Leveraging high-frequency electricity consumption data from a month-long conservation contest, in which participants were randomly assigned to one of three treatments or the control group, I estimate average treatment effects, their temporal dynamics, and heterogeneity. Fixed effects modeling presents parametrically constrained treatment effects, while my machine learning model uncovers more nuanced variation over time and across individuals. Our results show significant but modest effects on electricity conservation across all treatments, and a prominent deadline effect near the conclusion of the contest for the highest-intensity treatment. These findings provide insight into energy conservation in lower-income countries and suggest there needs to be considerable rethinking of similar interventions in this setting.To evaluate the impact of the third sustainable energy transition strategy of increasing the adoption of clean, decentralized energy systems, I focus on Cape Town, a city characterized by deep and persistent inequality. Households with sufficient financial means are increasingly adopting SHS, partly to mitigate experienced and anticipated future load shedding. These capital-intensive solutions are out of reach for many lower-income households. In this chapter, I investigate whether SHS adoption significantly affects household electricity consumption from the national grid, to examine the extent to which SHS enhances household resilience to unreliable power. I first built a deep learning model using the Mask2Former infrastructure to identify the locations and capacities of all SHS within the city. I then merge SHS data with detailed information on household electricity transactions and load shedding over time. Our fixed-effects econometric model results indicate that SHS reduce electricity consumption from the grid by approximately 14% on average. This effect dissipates over the winter, however, and is likely a lower bound due to uncertainty in our SHS data. Households with and without SHS mostly experience a modest rebound effect as outages increase, which suggests that consumption that would have occurred during outages is largely shifted in time.