The Household Appliance Stock, Income, and Electricity Demand Elasticity.

AuthorOhler, Adrienne

    Many energy policies specifically target low-income households for assistance. The most well-known of these policies, Low Income Home Energy Assistance Program (LIHEAP), is a federally funded program that is administered through the states to help low-income households with their home energy bills through a one-time payment. In addition, many state programs administered through utilities have a Percentage of Income Payment Plan (PIPP) program that provides monthly assistance for utility bills to the extent that a household's total bill exceeds a cap on their bill as a percentage of their income. Accordingly, PIPP programs provide the greatest assistance to lower income households. Federal weatherization and other energy efficiency programs also provide free or sliding scale benefits such as energy audits and insulation to lower income households.

    Even though policies such as the ones above are implemented with concern for low-income households, often such policies fail to consider the impact that income changes have on a household's electricity consumption as well as a household's appliance stock. Given that the residential sector accounts for over one-fifth of the total primary energy consumption in the U.S., understanding the factors that impact electricity consumption can help inform energy policy. This paper examines household electricity demand using data from the Energy Information Administration's (EIA's) 2015 Residential Energy Consumption Survey (RECS). The analysis examines the interactions between income, price, and the appliance stock to illustrate the importance of the appliance stock on household responses to price and income changes.

    Several studies have explored the relationship between prices, income, and household energy consumption, with a large amount of literature focusing on demand elasticities (Alberini et al., 2011; Bohi, 1981; Brounen et al., 2012; Carter et al., 2012; Cuddington and Dagher, 2015; Herriges and King, 1994; Labanderia et al., 2017; Matsukawa, 2001, 2004; Munley et al., 1990; Ohler and Billger, 2014; Reiss and White, 2005; Schulte and Heindl, 2017). The literature provides a variety of income and price elasticity estimates at the household level; however, several studies find that appliance ownership and usage explain more variation in electricity consumption than socio-demographic variables such as price and income (Bedir et al., 2013; Kavousian et al., 2013; Huebner et al., 2016; Reiss and White, 2005; Sanquist et al., 2012; Wiesmann et al., 2011). This is due to the fact that most households experience very little price variation, along with electricity consumption being an intermediate good. Cooking, running appliances, space heating and cooling, and operating electronics for entertainment are all final consumption goods in a household production function.

    We contribute to the literature on electricity demand in three important ways. First, we demonstrate that low-income households are more price responsive than high-income households, consistent with the work performed by Reiss and White (2005). We contribute to this research by using EIA's comprehensive survey data. The magnitude and robustness of the survey allows us to examine detailed household appliance characteristics not considered by previous research with smaller samples. This result suggests that low- and high-income households respond differently to price changes in electricity.

    Second, basic summary statistics show that low-income households use less electricity than high-income households, illustrating that the differences in price elasticity estimates are likely due to differences in overall electricity use. Thus, a 1 % price increase is likely to result in similar reduction levels of approximately 69-79 kWh for high- and low-income households. Thus, energy efficiency policy and conservation efforts implemented through price changes are likely to lead to similar absolute reductions in usage across income levels. However, if policymakers wish to have a response across income levels that is proportional to their usage, an across-the-board price increase will not yield the desired effect.

    Finally, we show that appliance ownership is a useful metric to predict differences in overall kWh electricity reduction through price changes. Based on various appliance characteristics, we estimate the electricity reduction from a 1% price increase. As price increases, households relying on electricity for cooking, heating, and/or water heating are much more responsive than their non-electric counterparts. A non-electric household responds to a 1% price increase by reducing electricity between 31-37 kWh, on average. In comparison, an all-electric household responds to a 1% price increase by reducing electricity between 142-164 kWh, on average. These all-electric households tend to be lower income households. Using the same analysis, we consider the appliance stock through televisions, AC ownership, and fridges, and observe a greater inelasticity among 'high stock' households, which tend to be high-income households. Our results show that the mechanism through which low-income households adapt to price changes differs from high-income households because of heterogeneity in their appliance stock.

    The results can help inform policymakers on the impact that energy policy can have on electricity demand. Current policies aimed at reducing electricity consumption, such as information-based marketing, home energy audits, energy efficiency rebates, dynamic pricing, and technical consumption feedback are more likely to be utilized by high-income households because of the greater perceived benefits. These programs will cause greater disparities between low- and high-income households and their behavior in the marketplace unless designed specifically to address these divergences.

    The paper proceeds as follows. Section 2 reviews the demand estimation typical in the literature, the data used in the analysis, and methods for approximating income when income is not directly observable. In section 3, we present the results from a simple electricity demand model to compare price elasticities and calculate average kWh reductions across income levels and appliance ownership. In section 4, we examine the differences in appliance ownership between high- and low-income households. Section 5 concludes with policy recommendations.


    This study analyzes household level data from the EIA's 2015 RECS, which is a national survey that collects energy-related information from U.S. households on dwelling characteristics, energy related expenditures, and household consumption for over 700 variables. Participants are selected at random using a multistage sample design with more than 5,600 eligible households completing the survey. Ten Census regions were considered in the survey design, and the survey was designed to ensure that the selected sample statistically represents the entire population of the U.S. (EIA, 2017). Additionally, the 2015 RECS offers more accuracy and coverage for understanding energy consumption in all regions and divisions because of its multistage area probability design. The survey excludes information on secondary homes, vacant units, military barracks, and common areas in apartment buildings. The American Community Survey (ACS) was also considered for the analysis but given the research focus on appliance ownership, RECS provides a more thorough examination of the appliance stock of households as well as energy use behaviors.

    The REC survey asks respondents about their dwelling attributes (e.g. size of house, number of bedrooms, level of insulation, age of house), their energy consuming appliances and equipment (e.g. number of refrigerators, TVs, computers, and use of electricity for cooking, heating, and water heating), and socio-economic characteristics (e.g. income, age, education, and number of individuals in household). In the 2015 RECS Energy Supplier Survey, electricity and natural gas suppliers were asked to provide monthly billing data between August 2016 and February 2017. This billing data was matched with households to capture energy consumption and expenditure.

    We focus on the relationship between income and a household's appliance stock and consequently aggregate annual electricity consumption. The final sample includes households that live in single-family homes with more than 330 reported days of energy supplier billing data. Respondents with partial or incomplete billing data are dropped from the analysis. Additionally, we exclude renters from the analysis due to the split incentive problem between renters and landlords because much of the discussion centers around appliance choices over which renters often have little or no control. Finally, to simplify the analysis, we exclude households in multi-family units to avoid households where electricity is combined with other housing units. Respondents with missing data are also excluded.

    The final sample includes 3,080 households. Figure 1 shows the income distribution of our sample relative to the U.S. population in 2015. U.S. estimates are collected from the Census Bureau's Current Population Survey and Annual Social and Economic Supplement. When compared with the U.S. population, the income distribution of our sample is a reasonable representation. However, we note that our sample appears to have a lower representation of individuals with very low incomes when compared to the U.S. population. This disparity is likely driven by our focus on homeowners and single-family residences. Additionally, the demand model is estimated such that the sample cases are weighted to represent the U.S. population of single-family households, including the residences not in the sample (EIA 2017).

    Table 1 provides the average annual electricity usage by income level. Again, because the survey is designed as 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