An Examination of How Energy Efficiency Incentives Are Distributed Across Income Groups.

AuthorJacobsen, Grant D.

    Increasing energy efficiency has been a prominent public policy goal in recent years. A variety of policies that target energy efficiency have been enacted or strengthened to this end. Standards, such as building energy codes, have been used to set minimum allowable efficiency levels. Taxes, including energy taxes or carbon taxes, have been used to indirectly encourage investment in energy efficiency by raising energy prices. Energy efficiency incentives, which typically offer subsidies for high-efficiency goods through rebates or tax credits, have been used to subsidize the costs of energy efficiency investments. Labeling programs, such as Energy Star, have been used to help households identify high-efficiency products.

    Partly as a result of these policies, energy efficiency has been linked to major economic and environmental changes. Globally, about $250 billion were invested in energy efficiency in 2017 (IEA, 2018). This investment has spurred job growth in certain sectors. For example, in the United States alone, energy efficiency has been linked to the creation of 2.5 million jobs (NASEO, 2019). From an environmental perspective, some simulations indicate that policies that aggressively support investment in energy efficiency could lead to an 1,830 MMT carbon dioxide decrease in U.S. greenhouse gas emissions by 2050 (Gowrishankar and Levin, 2017).

    As energy efficiency policies have become more prominent, researchers have increasingly sought to carefully evaluate these policies. Many evaluations have focused on effectiveness and efficiency. (1) While effectiveness and efficiency are important factors, they may mask variation in the distributional effects of policies, which have been a key element in analyses of many environmental and energy policies (e.g., Grainger and Kolstad, 2010; Bento et al., 2006). Distributional effects are important because, as modeled in optimal tax theory, policies that lead to a more equal distribution of resources will enhance social welfare, holding all else equal (Diamond and Saez, 2011). Additionally, distributional effects are often an important factor in determining whether enacting or retaining policies is politically feasible. Perhaps because of political factors, the distributional effects of conservation programs are often of direct interest to utility managers and policymakers (Wichman et al., 2016).

    In this paper, I investigate an important component related to the distributional effects of energy efficiency policy: how energy efficiency incentives are distributed across income groups. I focus on energy efficiency incentives because, as I describe below, the distributional effects of energy efficiency incentives have been the subject of relatively few studies and because energy efficiency incentives are a large and growing component of energy policy. By 2025, spending on incentives for energy efficiency is expected to be about $10 billion annually in the United States, doubling relative to 2010 levels (Barbose et al., 2015). (2)

    The analysis focuses on a uniquely well-suited version of the Residential Energy Consumption Survey (RECS) from 2009 that includes a large set of detailed questions related to energy efficiency incentives. (3) Using a variety of empirical techniques, I evaluate how the probability of receiving an incentive relates to household income across three different types of subsidies (manufacturer or retailer rebates, utility rebates, and tax credits) and eight different types of equipment (refrigerators, dishwashers, clotheswashers, space heaters, central air-conditioners, light bulbs, windows, and insulation).

    The results indicate that almost all forms of incentives are concentrated in higher-income households, but there is substantial heterogeneity in the magnitude. Tax credits are the most concentrated type of subsidy and utility rebates are the least concentrated. Incentives for appliances that are not always present in residences, such as dishwashers and clotheswashers, are more concentrated than are incentives for equipment that tends to be universally-owned, such as refrigerators. The levels of concentration that are estimated are substantial. For example, regression models indicate that a household with an income of $80k is three times more likely than a household with an income of $20k to receive an incentive. The concentration of incentives in higher-income households is driven by differential rates across income groups in equipment presence and turnover, willingness to purchase Energy Star models, and homeownership. Utility rebates are no longer concentrated in higher-income households after controlling for these factors, but manufacturer / retailer rebates and tax credits remain concentrated.

    The results are helpful for informing how energy efficiency incentives should be structured. As I describe in Section 3, many policies lead to the provision of energy efficiency incentives. The main implication of the results for policymakers is that incentives are more likely to go to lower-income households if policies are structured such that the incentives are provided through utility rebates and such that the incentives avoid appliances that are more likely to be owned by higher-income households. Optimal policy design will require consideration of a broader set of factors; such as cost-effectiveness, free-ridership, producer price responses, and effects on innovation, but distributional differences in who receives incentives are an important factor as policymakers evaluate policy options and the associated trade-offs across multiple different criteria. (4) It should further be noted that the analysis is descriptive in nature and focuses on capturing tendencies with respect to the average distributional effects of different types of energy efficiency incentives. The analysis is not necessarily predictive of distributional effects of any individual program, which will depend on a large variety of factors which are not embedded in the present analysis.


    This paper contributes to the literature on the distributional effects of energy efficiency policies. In this section, I describe earlier studies in this area. I focus especially on four studies that have provided evaluations related to the distributional effects of energy efficiency incentives, although I also briefly describe work on the distributional effects of other policies related to energy efficiency.

    Borenstein and Davis (2016) use U.S. tax return data to examine the socioeconomic characteristics of individuals who recently received federal tax credits for a variety of "clean energy" investments, including residential energy investments for energy efficiency and renewable energy. They find that the bottom three income quintiles received only about 10 percent of all credits. They conclude that tax credits are less attractive on distributional grounds than other market-based policies that could reduce emissions.

    In another study using tax return data, Neveu and Sherlock (2016) find that federal tax credits for residential energy investments are distributed inequitably across groups. In addition to finding that tax credits are more likely to go to higher-income households, they also find that taxpayers in colder climates and in areas with higher electricity costs are more likely to take advantage of tax credits.

    Sutherland (1994) presents survey evidence from 1990 that higher-income households are more likely to participate in demand-side management (DSM) programs, including utility rebates. Households with newer homes and newer heating and cooling equipment are also more likely to participate in utility programs. Additionally, participants in utility DSM programs are more likely to undertake conservation measures other than those incentivized through rebates, indicating that rebate programs may serve as substitutes for other conservation investments.

    Using a discrete choice-model, Bruegge (2017) analyses a refrigerator and clotheswasher rebate program offered by a large utility. He focuses especially on the role that fundraising plays in utility-based programs. The results indicate that price changes induced by the rebate program enhance the energy savings attributable to the program yet reduced its welfare effects. Overall, the program created a loss in consumer surplus and the loss was greatest for low-income households.

    Other studies have evaluated the distributional effects of other types of policies related to energy efficiency, including carbon taxes, gasoline taxes, fuel economy standards, and building energy codes. Grainger and Kolstad (2010) use data from the consumer expenditure survey and an input-output model to present evidence that a carbon price is regressive. Bento et al. (2009) evaluate the gasoline tax and find that the distributional impacts differ substantially depending on how the revenue is recycled. Davis and Knittel (2016) evaluate fuel economy standards and present evidence that the implicit tax imposed by the policy, measured as a share of income, is greater for lower-income households. Levinson (2019) focuses on a comparison of fuel economy standards to gas taxes and presents theory and evidence that both are regressive, but that taxes are less so. Bruegge et al. (2019) evaluate building energy codes and find that they result in more undesirable distortions for lower-income households, partly because codes lead to the construction of smaller residences.

    The primary contribution of this paper relative to existing work on the distributional effects of energy efficiency incentives is that I examine incentives administered through several forms of subsidies and for multiple different types of equipment using the same sample and empirical framework. This feature allows me to describe how energy efficiency incentives are distributed across income groups in a...

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT