Machine learning from schools about energy efficiency

Fiona Burlig; Christopher Knittel; David Rapson; Mar Reguant; Catherine Wolfram

doi:10.1086/710606

Machine learning from schools about energy efficiency

Fiona Burlig, Christopher Knittel, David Rapson, Mar Reguant, Catherine Wolfram

Economics

Research output: Contribution to journal › Article › peer-review

40 Scopus citations

Abstract

We use high-frequency panel data on electricity consumption to study the effectiveness of energy efficiency upgrades in K–12 schools in California. Using a panel fixed effects approach, we find that these upgrades deliver between 12% and 86% of expected savings, depending on specification and treatment of outliers. Using machine learning to inform our specification choice, we estimate a narrower range: 52%–98%, with a central estimate of 60%. These results imply that upgrades are performing less well than ex ante predictions on average, although we can reject some of the very low realization rates found in prior work.

Original language	English (US)
Pages (from-to)	1181-1217
Number of pages	37
Journal	Journal of the Association of Environmental and Resource Economists
Volume	7
Issue number	6
DOIs	https://doi.org/10.1086/710606
State	Published - Nov 2020

Keywords

Energy efficiency
Machine learning
Schools

ASJC Scopus subject areas

Economics and Econometrics
Nature and Landscape Conservation
Management, Monitoring, Policy and Law

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Access to Document

10.1086/710606

Cite this

@article{44dc540a510a480db72294bab09a1b39,

title = "Machine learning from schools about energy efficiency",

abstract = "We use high-frequency panel data on electricity consumption to study the effectiveness of energy efficiency upgrades in K–12 schools in California. Using a panel fixed effects approach, we find that these upgrades deliver between 12% and 86% of expected savings, depending on specification and treatment of outliers. Using machine learning to inform our specification choice, we estimate a narrower range: 52%–98%, with a central estimate of 60%. These results imply that upgrades are performing less well than ex ante predictions on average, although we can reject some of the very low realization rates found in prior work.",

keywords = "Energy efficiency, Machine learning, Schools",

author = "Fiona Burlig and Christopher Knittel and David Rapson and Mar Reguant and Catherine Wolfram",

note = "Publisher Copyright: {\textcopyright} 2020 by The Association of Environmental and Resource Economists.",

year = "2020",

month = nov,

doi = "10.1086/710606",

language = "English (US)",

volume = "7",

pages = "1181--1217",

journal = "Journal of the Association of Environmental and Resource Economists",

issn = "2333-5955",

publisher = "University of Chicago Press",

number = "6",

}

TY - JOUR

T1 - Machine learning from schools about energy efficiency

AU - Burlig, Fiona

AU - Knittel, Christopher

AU - Rapson, David

AU - Reguant, Mar

AU - Wolfram, Catherine

PY - 2020/11

Y1 - 2020/11

N2 - We use high-frequency panel data on electricity consumption to study the effectiveness of energy efficiency upgrades in K–12 schools in California. Using a panel fixed effects approach, we find that these upgrades deliver between 12% and 86% of expected savings, depending on specification and treatment of outliers. Using machine learning to inform our specification choice, we estimate a narrower range: 52%–98%, with a central estimate of 60%. These results imply that upgrades are performing less well than ex ante predictions on average, although we can reject some of the very low realization rates found in prior work.

AB - We use high-frequency panel data on electricity consumption to study the effectiveness of energy efficiency upgrades in K–12 schools in California. Using a panel fixed effects approach, we find that these upgrades deliver between 12% and 86% of expected savings, depending on specification and treatment of outliers. Using machine learning to inform our specification choice, we estimate a narrower range: 52%–98%, with a central estimate of 60%. These results imply that upgrades are performing less well than ex ante predictions on average, although we can reject some of the very low realization rates found in prior work.

KW - Energy efficiency

KW - Machine learning

KW - Schools

UR - http://www.scopus.com/inward/record.url?scp=85092790121&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85092790121&partnerID=8YFLogxK

U2 - 10.1086/710606

DO - 10.1086/710606

M3 - Article

AN - SCOPUS:85092790121

SN - 2333-5955

VL - 7

SP - 1181

EP - 1217

JO - Journal of the Association of Environmental and Resource Economists

JF - Journal of the Association of Environmental and Resource Economists

IS - 6

ER -

Machine learning from schools about energy efficiency

Abstract

Keywords

ASJC Scopus subject areas

UN SDGs

Access to Document

Other files and links

Fingerprint

Cite this