Evidence of a Homeowner-Renter Gap for Electric Appliances.

• Author: Davis, Lucas W.

1. INTRODUCTION

   There is a growing consensus that building electrification will play a crucial role in reducing carbon dioxide emissions. Several recent large-scale studies of the United States, for example, conclude that virtually all feasible pathways to decarbonization include widespread residential electrification (Larson et al., 2020; National Academies of Sciences, Engineering, and Medicine, 2021; Williams et al., 2021). Already many U.S. cities have banned natural gas for new homes in what the Wall Street Journal aptly describes as "a growing fight unfolding across America." (1)

   While most of the academic literature and policy discussion surrounding building electrification emphasizes owner-occupied housing, this paper focuses on rental housing. In the United States, over one-third of homes are rented and these homes consume almost one trillion cubic feet of natural gas annually. (2) Moreover, the incentive issues which arise between landlords and tenants mean that the economics of rental housing is different from owner-occupied, and of considerable independent interest.

   This paper provides the first empirical analysis of the homeowner-renter gap for electric appliances. Using U.S. nationally representative data, the analysis shows that renters are significantly more likely than homeowners to have electric heat, electric hot water heating, an electric stove, and an electric dryer. The gap is statistically significant at the 1% level for all four appliance categories, prevalent across regions, and persists after controlling for the type, size, and age of the home, as well as for climate and household characteristics.

   This gap likely arises from the same split incentives that lead to underinvestment in energy-efficiency. Researchers have long bemoaned the "landlord-tenant problem," pointing out that landlords have too little incentive to invest in energy-efficiency when their tenants pay the energy bills (Blumstein et al., 1980; Jaffe and Stavins, 1994; Gillingham et al., 2009; Allcott and Greenstone, 2012; Gillingham and Palmer, 2014; Gerarden et al., 2017). (3)

   By the same argument, landlords tend to prefer electric appliances because they are less capital-intensive. Electric resistance heating is cheaper to install than a natural gas furnace, and electric dryers and electric hot water heaters are cheaper to install than natural gas. (4) Although in theory, the higher capital cost of natural gas appliances could be passed on in the form of higher rents, it can be difficult for landlords to effectively convey this type of information (Myers, 2020).

   These findings are relevant for an emerging set of policies aimed at reducing carbon dioxide through building electrification. In California, more than 40 cities have passed measures prohibiting or restricting natural gas in new homes, and policymakers are retooling state building codes to favor all-electric homes. (5) In addition, the Biden administration announced in May 2021 its support for building performance standards and other initiatives aimed at building electrification. (6)

   This paper is related to an existing literature on split incentives and energy-efficiency (Levinson and Niemann, 2004; Maruejols and Young, 2011; Davis, 2012; Gillingham et al., 2012; Krishnamurthy and Kristrom, 2015; Aydin et al., 2019). These studies have tended to find that homeowners are more likely than renters to have energy-efficient technologies. For example, Gillingham et al. (2012) shows that California homeowners are 20 percent more likely than renters to have attic and ceiling insulation and Krishnamurthy and Kristrom (2015) show that homeowners are more likely than renters to have energy-efficient appliances and other energy-efficient technologies using data from 11 OECD countries. (7)

   This paper also contributes to a broader literature on energy consumption in rental housing. Best et al. (2021) shows that U.S. rental homes use 35% less electricity than owner-occupied homes, but that this negative unconditional effect turns into a positive 9% conditional effect after controlling for location, household, and appliance quantity characteristics. They attribute this positive conditional effect to lower energy-efficiency, behavioral factors like more television watching, differences in bill payment responsibilities, and increased reliance on electric space and water heating.

   The paper proceeds as follows. Section 2 describes the data and presents baseline evidence on electric appliances for homeowners and renters. Section 3 discusses alternative explanations and performs regression analyses with a large number of controls. Section 4 performs additional analyses distinguishing between tenant-pay and landlord-pay housing units, estimating models separately by housing type, and corroborating the main results with evidence from an alternative dataset. Section 5 concludes with a discussion of economic and policy implications.

2. BASELINE EVIDENCE

   2.1 Data

   The primary dataset for these analyses is household-level microdata from the 2015 Residential Energy Consumption Survey (RECS) from the U.S. Department of Energy. (8) These data are nationally representative of the United States' 83 million owner-occupied housing units and 43 million renter-occupied housing units. RECS provides rich household-level information about household appliances, as well as about household income and other characteristics. The RECS sample is selected using stratified sampling, so RECS sampling weights are used in all results.

   The 2015 RECS has a total sample size of 5,686 households, with 3,928 homeowners and 1,758 renters. The analyses throughout exclude households if they do not have a particular category of appliance. Among homeowners, 97%, 100%, 99%, and 95% have heat, hot water, stove, and dryer, respectively. Among renters, the saturation rates are 93%, 100%, 98%, and 56%. Thus, in the regressions which follow the sample sizes for the four appliance categories are 5,428, 5,686, 5,622, and 4,750. In practice, excluding households without a particular category of appliance only substantively impacts the results for dryers.

   2.2 Comparing Means

   Figure 1 plots the percentage of U.S. homeowners and renters with four different categories of electric appliances. Across categories, renters are significantly more likely than homeowners to have electric appliances. The biggest gap is for electric heating. Whereas 49% of U.S. renters heat their homes primarily with electricity, only 29% of U.S. homeowners do the same. There is a considerable homeowner-renter gap for all four categories, with renters between 9 and 20 percentage points more likely to have electric appliances.

   2.3 Results by U.S. Region

   Table 1 presents estimates of the homeowner-renter gap for the entire U.S. and for the four Census regions. Estimates and standard errors are reported from twenty separate least squares regressions of the following form,

   1 [(Electric Appliance).sub.i] = [[alpha].sub.0] + [[alpha].sub.1] 1[(Renter).sub.i] + [[epsilon].sub.i]. (1)

   In all regressions, the dependent variable 1 (Electric Appliance) is an indicator variable equal to one if the household has an electric appliance of the category indicated in the panel heading. The table reports the coefficient [[alpha].sub.1] corresponding to 1 (Renter), an indicator variable for renters. This coefficient is the difference in electric appliance saturation between renters and homeowners, with a positive coefficient indicating that renters are more likely to have an electric appliance. For these results no additional control variables are included, so this is equivalent to a two-sample t-test.

   The estimates reveal a pronounced homeowner-renter gap across appliances and regions. The national estimates are equivalent to the gaps presented in Figure 1. Point estimates range from 9 percentage points for electric hot water heaters and electric stoves, to 20 percentage points for electric space heating. Electric heating has the largest point estimate across all regions, ranging from 12 percentage points in the Northeast to 23 percentage points in the South. Across appliance categories point estimates tend to be smaller in the Northeast, and larger in the Midwest. Of the 20 estimates, 16 are positive and statistically significant at the 1% level.

3. ALTERNATIVE EXPLANATIONS

   ...