UC Davis

Energy markets are at the forefront of the conversation as policymakers worldwide experience a paradigm shift. Technological changes like the shale revolution and the efficiency gains in electricity generation through renewable resources have created abundant energy. In addition, the 2015 Paris Agreement has catalyzed a future where energy generation is less carbon intensive. Through these tectonic changes, policymakers slowly have abandoned the notion that market conditions will be dominated by a binding energy supply, commonly known as “peak oil” (Hubbert 1956). Instead, they are bracing for a world where climate policies and technology are pivotal in shaping demand. Yet, a past full of well-intentioned but perfectible policies is evidence that we are far from a good understanding of what drives consumer demand for energy. My dissertation expands our understanding of consumer demand for gasoline, the primary fuel source for transportation, and shows how public policies affect consumer welfare.

My research studies two distinct settings, but in both cases, I exploit regulatory constraints to causally identify fundamental parameters of consumer preferences. The first setting is the newly liberalized gasoline market in Mexico, where gas station operators can choose their pricing and branding strategies for the first time in more than 80 years. Mexico’s regulatory environment before the liberalization allowed me to identify key preference parameters such as consumers’ price sensitivity, willingness to drive, and the value for the convenience of product availability.

The market-level price elasticity of demand for Mexican consumers is between -0.42 and -0.64. Similarly, I estimate that, on average, consumers would need to save 13.8 MXNc/L (3.3 USDc/gal) at the pump to be indifferent to driving an extra kilometer to visit a gas station. However, considerable heterogeneity exists across households, driven chiefly by their income level. For example, low-income households are 55% more price sensitive than high-income households.Given my parameter estimates for heterogenous households, I estimate the welfare impact that consumers experienced between 2015 and 2019. During this period, retail prices were liberalized, and initially, there were substantial gas station openings. However, from late 2016 onwards, entry was constrained through regulatory backlogs. I find that for every peso gained in welfare from increased product availability, roughly two pesos are lost in welfare from increased prices. I estimate an annual net loss of 7% of the markets’ annual revenue or 1.43 billion MXN/year. However, this loss is not distributed evenly. Households in the top three income deciles consume 60% of the gasoline, while households in the bottom three deciles only consume 13% of the annual volume. Therefore, high-income households are affected the most by price increases, while low-income households benefit the most from the increased taxation by gas station openings. These policies have been regressive on a net basis, with low-income households being worse off by roughly 2.5% of their disposable income while high-income households are worse off by 1.5%.

In the second setting, I study California’s gasoline market and consumers’ price sensitivity. In this mature market, environmental regulations segment California from the rest of the contiguous U.S. I use this feature along with the rich data available for the state to control for current and persistent demand shifts. However, I purposely do not control for simultaneity bias which results in estimating a lower bound for the price elasticity of demand. This lower bound coincides with previous studies that have tried to estimate the price elasticity of demand, suggesting that their instruments may be weak or endogenous.

I further propose refinery outages as a new set of instruments to estimate the price elasticity of demand. These instruments yield estimates that are statistically larger in magnitude than previous state-of-the-art instrumental variable estimates. The estimate of my preferred specification, -0.52, is 40% larger than what was reported in earlier research. Due to the segmented nature of California’s gasoline market, refinery outages have a strong explanatory power. Due to their unexpected nature, these instruments are conditionally mean-independent of unobserved demand shocks. They satisfy the four main documented shortcomings of previously used instruments: a weak first stage, endogeneity with economic activity, endogeneity with consumers’ anticipatory behavior, and they do not elicit long-term adaptative behavior.

The methods I use in this dissertation are varied, yet they are all grounded in causal inference. In the first essay, I use structural consumer behavior modeling through a random coefficients model to estimate fundamental preferences using cross-sectional market data. In the second essay, I use Monte Carlo simulations to build plausible counterfactual pricing scenarios had the markets not been liberalized. In the third essay, I use time series and local projection techniques to causally estimate consumers’ price sensitivity.

The common focus of my research is to expand our understanding of demand by estimating fundamental consumer preferences. In the process, I find actionable insights that can be used by policymakers, businesspeople, and academics alike. For example, having estimated consumer preferences in the Mexico setting, I compute the welfare effects of the policies that followed the price liberalization. In the California setting, I propose a new set of instruments that can be used to inform policy design better and answer questions beyond fuel markets, including agricultural markets, cost passthrough and monetary policy, and even environmental policy.