UC Berkeley

Most human decisions are made without consideration for the associated energy use. But our actions, from choices about where to live and how to commute, to the routine decision about when to sleep or whether to go to church, have energy implications. Collectively, these decisions form cities filled with gasoline-consuming cars and dictate when power plants turn on and off. Choices, even when made subconsciously, are made subject to constraints formed by past societies' decisions or attempts to coordinate with family, friends, and colleagues. This dissertation broadly investigates how technologies, norms, and incentives affect human behavior, energy use, and, ultimately, climate change. I develop novel and large datasets to investigate unanswered questions in energy and urban economics.

In Chapters 1 and 2, I ask how transportation technology affects urban growth. A broad set of interacting factors, such as physical features and land use policy, cause particular spatial organizations of households and firms in rapidly growing urban areas. Transportation costs and real estate prices drive individuals' decisions about where to work and live. These choices have tremendous welfare implications, costing time and energy, and resulting in externalities including air pollution and traffic congestion. Despite high social costs, there is little empirical evidence about the effect of changes in transportation costs on city structure. I estimate the effects of two transportation innovations - (1) a speed limit increase, and (2) ridesharing services - on residential real estate prices and development. I find that prices respond quickly and significantly to transportation cost changes. Consistent with my theoretical model, an increase in speed limits decreases housing prices by over 3% on average, with the largest effect in the city center. Subsequent housing development is farther from the central city. In contrast, the launch of Uber increases housing prices by over 2% after the first year, with a larger immediate effect in the central city. Housing development occurs closer to the central city after treatment. Both treatments change the ability of households to access surrounding markets. Applying the concept of "market access" from the trade literature, I show that the distribution of business establishments around a property dramatically changes the magnitude of each effect.

Chapter 3, co-authored with Solomon Hsiang and Terin Mayer, shows that electricity use can be predicted from human activities. System operators predict electricity loads in order to schedule power plants and allocate transmissions resources to ensure grid reliability. In the long term, forecasting dictates whether new power plants will be contracted or built. We combine hourly data on electricity load with the American Time Use Survey and show that, with just three time-use variables, we can predict over 90% of variation in electricity use. In an increasingly data-rich world, we know more about what individuals' locations and activities, making our finding a potentially valuable tool for improving prediction as well as offering ways to reduce electricity usage by shifting human activity.

Chapter 4, also co-authored with Solomon Hsiang, explores energy use on holidays and weekends. Government policies that coordinate labor and leisure have profound economic implications. However, manipulating the structure of our workweek, weekend, and holiday calendar in order to improve economic outcomes is a policy lever that has been largely unused. Setting optimal coordinating mechanisms and allocations may present a useful tool for reducing carbon emissions, particularly if individuals are constrained in the number of days they have available to take coordinated leisure. We provide evidence that there is an environmental externality associated with labor relative to leisure. We empirically estimate the effect of weekends and holidays on electricity loads, vehicle travel, and air travel in the U.S. We observe large reductions in electricity load, air travel, and vehicle travel on many holidays, as well as reductions on surrounding days. Using time use data and exogenous variation in when a holiday is observed, we provide evidence that, beyond labor being less carbon intensive than leisure, agents enjoy less carbon-intensive activities on holidays. Holidays during the summer all result in significant savings. New holidays may be more economically valuable if they are scheduled for hot days that are likely to require high marginal cost generators to meet electricity demand.