Machine Learning from Schools about Energy Efficiency

In the United States, consumers invest billions of dollars annually in energy efficiency, often on the assumption that these investments will pay for themselves via future energy cost reductions. We study energy efficiency upgrades in K-12 schools in California. We develop and implement a novel machine learning approach for estimating treatment effects using high-frequency panel data, and demonstrate that this method outperforms standard panel fixed effects approaches. We find that energy efficiency upgrades reduce electricity consumption by 3 percent, but that these reductions total only 24 percent of ex ante expected savings. HVAC and lighting upgrades perform better, but still deliver less than half of what was expected. Finally, beyond location, school characteristics that are readily available to policymakers do not appear to predict realization rates across schools, suggesting that improving realization rates via targeting may prove challenging.

[1]  Sébastien Houde,et al.  NBER WORKING PAPER SERIES HETEROGENEOUS (MIS-) PERCEPTIONS OF ENERGY COSTS: IMPLICATIONS FOR MEASUREMENT AND POLICY DESIGN , 2019 .

[2]  Jessica Granderson,et al.  Application of automated measurement and verification to utility energy efficiency program data , 2017 .

[3]  Lucas W. Davis,et al.  Do Energy Efficiency Investments Deliver at the Right Time? , 2017, American Economic Journal: Applied Economics.

[4]  W. Michael Hanemann,et al.  The economic impacts of climate change: Evidence from agricultural output and random fluctuations in weather: Comment , 2014 .

[5]  Pierre Desprairies,et al.  World Energy Outlook , 1977 .

[6]  Mateus Souza Predictive Counterfactuals for Event Studies with Staggered Adoption: Recovering Heterogeneous Effects from a Residential Energy Efficiency Program , 2019, SSRN Electronic Journal.

[7]  Maarten Jennen,et al.  The impact of energy labels and accessibility on office rents , 2012 .

[8]  R. J. Park Hot Temperature and High Stakes Cognitive Assessments∗ , 2018 .

[9]  Alan R. Ellis,et al.  The role of prediction modeling in propensity score estimation: an evaluation of logistic regression, bCART, and the covariate-balancing propensity score. , 2014, American journal of epidemiology.

[10]  K. Gillingham,et al.  Bridging the Energy Efficiency Gap: Policy Insights from Economic Theory and Empirical Evidence , 2013, Review of Environmental Economics and Policy.

[11]  Jonathan Hersh,et al.  Poverty in HD : What Does High Resolution Satellite Imagery Reveal about Economic Welfare ? , 2016 .

[12]  Kevin Novan,et al.  The Incentive to Overinvest in Energy Efficiency: Evidence from Hourly Smart-Meter Data , 2018, Journal of the Association of Environmental and Resource Economists.

[13]  Sang Michael Xie,et al.  Combining satellite imagery and machine learning to predict poverty , 2016, Science.

[14]  Matthew Neidell,et al.  Temperature and Human Capital in the Short and Long Run , 2015, Journal of the Association of Environmental and Resource Economists.

[15]  Xavier Jaravel,et al.  Revisiting Event Study Designs , 2017 .

[16]  Amitabh Chandra,et al.  Health Care Exceptionalism? Performance and Allocation in the US Health Care Sector. , 2016, The American economic review.

[17]  Michael Greenstone,et al.  Do Energy Efficiency Investments Deliver? Evidence from the Weatherization Assistance Program , 2015 .

[18]  Victor Chernozhukov,et al.  Inference on Treatment Effects after Selection Amongst High-Dimensional Controls , 2011 .

[19]  Susan Athey,et al.  Beyond prediction: Using big data for policy problems , 2017, Science.

[20]  Paul L. Joskow,et al.  What Does a Negawatt Really Cost? Evidence from Utility Conservation Programs , 1992 .

[21]  J. Reif,et al.  The Mortality and Medical Costs of Air Pollution: Evidence from Changes in Wind Direction , 2016, The American economic review.

[22]  Matthew J. Kotchen,et al.  Longer-Run Evidence on Whether Building Energy Codes Reduce Residential Energy Consumption , 2017, Journal of the Association of Environmental and Resource Economists.

[23]  John M. Quigley,et al.  The Economics of Green Building , 2010, Review of Economics and Statistics.

[24]  Nicholas Ryan,et al.  Energy Productivity and Energy Demand: Experimental Evidence from Indian Manufacturing Plants , 2018 .

[25]  Ryan Engstrom,et al.  Poverty from Space: Using High-Resolution Satellite Imagery for Estimating Economic Well-Being , 2017 .

[26]  Alberto Abadie,et al.  Choosing among Regularized Estimators in Empirical Economics: The Risk of Machine Learning , 2019, Review of Economics and Statistics.

[27]  Matthew E. Kahn,et al.  Carbon emissions from the commercial building sector: The role of climate, quality, and incentives , 2014 .

[28]  Michael Luca,et al.  Crowdsourcing City Government: Using Tournaments to Improve Inspection Accuracy , 2016 .

[29]  Paul J. Gertler,et al.  Cash for Coolers : Evaluating a Large-Scale Appliance Replacement Program in Mexico , 2014 .

[30]  Steve Cicala When Does Regulation Distort Costs? Lessons from Fuel Procurement in U.S. Electricity Generation , 2013 .

[31]  Michael Greenstone,et al.  Measuring the Welfare Effects of Residential Energy Efficiency Programs , 2017 .

[32]  Gabriel Cadamuro,et al.  Predicting poverty and wealth from mobile phone metadata , 2015, Science.

[33]  H. Allcott,et al.  Is There an Energy Efficiency Gap? , 2012 .

[34]  G. Imbens,et al.  Matrix Completion Methods for Causal Panel Data Models , 2017, Journal of the American Statistical Association.

[35]  R. Newell,et al.  Assessing the Energy-Efficiency Gap , 2015 .

[36]  S. Borenstein The Trouble With Electricity Markets: Understanding California's Restructuring Disaster , 2002 .

[37]  Hal R Varian,et al.  Causal inference in economics and marketing , 2016, Proceedings of the National Academy of Sciences.

[38]  Arik Levinson,et al.  How Much Energy Do Building Energy Codes Save? Evidence from California Houses , 2016 .

[39]  Steve Cicala Imperfect Markets Versus Imperfect Regulation in U.S. Electricity Generation , 2017 .

[40]  Andrew Goodman-Bacon Difference-in-Differences with Variation in Treatment Timing , 2018, Journal of Econometrics.

[41]  J. Robins,et al.  Double/Debiased Machine Learning for Treatment and Structural Parameters , 2017 .

[42]  Susan Athey,et al.  Recursive partitioning for heterogeneous causal effects , 2015, Proceedings of the National Academy of Sciences.

[43]  Jure Leskovec,et al.  Human Decisions and Machine Predictions , 2017, The quarterly journal of economics.

[44]  Sendhil Mullainathan,et al.  Machine Learning: An Applied Econometric Approach , 2017, Journal of Economic Perspectives.

[45]  Merrian Borgeson Financing Energy Upgrades for K-12 School Districts: A Guide to Tapping into Funding for Energy Efficiency and Renewable Energy Improvements , 2014 .

[46]  César A. Hidalgo,et al.  Cities Are Physical Too: Using Computer Vision to Measure the Quality and Impact of Urban Appearance , 2016 .

[47]  D. McCaffrey,et al.  Propensity score estimation with boosted regression for evaluating causal effects in observational studies. , 2004, Psychological methods.

[48]  Joshua A. Blonz,et al.  The Welfare Costs of Misaligned Incentives: Energy Inefficiency and the Principal-Agent Problem , 2019, Finance and Economics Discussion Series.

[49]  S. Borenstein,et al.  Do Two Electricity Pricing Wrongs Make a Right? Cost Recovery, Externalities, and Efficiency , 2018, American Economic Journal: Economic Policy.

[50]  Esther Duflo,et al.  Generic Machine Learning Inference on Heterogenous Treatment Effects in Randomized Experiments , 2017 .

[51]  A. Fuels,et al.  Electric power monthly , 1992 .

[52]  Paul J. Ferraro,et al.  Panel Data Designs and Estimators as Substitutes for Randomized Controlled Trials in the Evaluation of Public Programs , 2017, Journal of the Association of Environmental and Resource Economists.

[53]  Alberto Abadie,et al.  The risk of machine learning , 2017, 1703.10935.