Files in this item

FilesDescriptionFormat

application/pdf

application/pdfNOGUEIRAMEIRELLESDESOUZA-DISSERTATION-2020.pdf (25MB)
(no description provided)PDF

Description

Title:Three essays on the economics of energy efficiency and conservation
Author(s):Nogueira Meirelles De Souza, Mateus
Director of Research:Myers, Erica
Doctoral Committee Chair(s):Myers, Erica
Doctoral Committee Member(s):Christensen, Peter; Fullerton, Don; Khanna, Madhu
Department / Program:Agr & Consumer Economics
Discipline:Agricultural & Applied Econ
Degree Granting Institution:University of Illinois at Urbana-Champaign
Degree:Ph.D.
Genre:Dissertation
Subject(s):energy economics
energy efficiency
causal inference
environmental economics
impact evaluation
machine learning
randomized control trial
observational study
applied econometrics
behavioral economics
Abstract:In this dissertation, I analyze investment decisions and consumer behavior related to energy efficiency and conservation. This research is motivated by evidence that the benefits from energy savings can extend well beyond private monetary gains to consumers. By providing insight on these topics, I contribute to a broad literature on environmental economics. This dissertation is constituted of three chapters as follows. For the first chapter, I introduce a methodological contribution for statistical evaluation of the impact of policy changes, interventions, or general "treatments." Specifically, I focus on estimating treatment effect heterogeneity in event studies with staggered adoption: panel data settings where observational units experience treatment at different points in time. I propose using highly flexible machine learning algorithms to predict counterfactuals (unobserved outcomes in absence of treatment) in those settings. With simulations, I show that my proposed method can recover nuanced effects with more accuracy and with better statistical efficiency than traditional econometric models (such as fixed effects regressions). I conclude that chapter with an application of the ML approach to real data from the Weatherization Assistance Program (WAP), which is one of the largest residential energy efficiency programs in the US. I identify how energy savings differ substantially depending on housing structure and on which types of upgrades were performed. For example, I document how complete furnace replacements are associated with significantly higher savings than furnace repairs. Also, I show that measures related to wall, attic, and foundation insulation are among the strongest contributors to energy savings. Finally, I assess measure-specific cost-effectiveness. Those results are informative for efforts to improve the allocation of program funds. The second chapter of this dissertation consists of a randomized control trial (RCT) to assess if behavioral nudges can promote energy conservation in absence of direct monetary incentives. The RCT was conducted in a campus residence hall with students that payed a fixed fee for energy at the beginning of each term. In that setting, behavioral factors such as social norms and moral suasion are the main incentives for conservation, since the marginal monetary costs/savings for energy are zero. The RCT consisted of sending "home energy reports" to randomly selected students, revealing their heating/cooling energy consumption. The energy usage of their neighbors was also displayed. Results from analyzing high-frequency thermostat data suggest that those reports were not effective for changing students’ consumption patterns during the regular semester. On the other hand, nudges sent prior to school breaks resulted in significantly lower thermostats (thus lower energy consumption). That second finding suggests that the null effects during the regular semester are unlikely to be driven by students’ inattention or by lack of understanding on how to operate thermostats. Rather, students were more willing to lower setpoints before leaving for breaks, given that they would not be facing associated thermal discomfort in that case. Collectively, those results suggest that behavioral nudges alone may not be sufficient to promote energy conservation in settings were monetary incentives are lacking. For the last chapter of this dissertation, I investigate how and why rented dwellings are less likely to have energy efficient appliances. I use data from a representative sample of residences across the continental US. I document that, on average, dwellings are more efficient when landlords are responsible for paying energy bills, and in states with high energy prices. Those results are consistent with a well-documented problem of split incentives in residential markets: the "landlord-tenant problem." When tenants pay utility bills, landlords may have little incentive to invest in efficiency, especially if those investments do not translate into higher rents. That could happen due to information asymmetries that limit prospective tenants' abilities to fully compare rental units across the market. By analyzing the effects of tenancy duration on the adoption of efficient technologies, I find evidence that investments in owner-occupied homes are more likely to occur closer to move-in dates. On the other hand, investments in rented homes are more likely to occur at later periods of tenancy, when relations between landlords and tenants might be better established. Those results are suggestive of a sorting process in residential markets, where homeowners have a stronger preference for energy-efficient dwellings and may have smaller discount rates. Findings from this chapter reinforce that energy efficiency policies in rental markets should differ from those targeted at homeowners.
Issue Date:2020-07-10
Type:Thesis
URI:http://hdl.handle.net/2142/108459
Rights Information:Copyright 2020 Mateus Nogueira Meirelles De Souza
Date Available in IDEALS:2020-10-07
Date Deposited:2020-08


This item appears in the following Collection(s)

Item Statistics