The effect of U.S. electricity prices on the purchase of energy- efficient appliances and implications for the effects of carbon pricing Peter Schwarz, University of North Carolina Charlotte, 704.687.7614, pschwarz@uncc.edu Craig Depken, University of North Carolina Charlotte, cdepken@uncc.edu Michael Herron, Premier Healthcare Alliance, mherron@uncc.edu Benjamin Correll, Pricewaterhouse Coopers, benjcorrell@gmail.com Overview There is considerable disagreement, both theoretical and empirical, and consumer willingness to pay for energy efficiency. We are among the few studies to use data, rather than surveys, to estimate the effect of electricity prices on energy-efficient purchases. Most analyses rely upon survey data, which has wellknown biases of which the most fundamental is that consumer response is based upon stated rather than revealed preferences. Consumer responses to surveys need not coincide with actual purchasing decisions. We use U.S. Energy Information Administration Data on purchases of energy-efficient appliances – appliances that qualify for the U.S. Environmental Protection Agency designation as ENERGY STAR appliances—that save 15-20% of the electricity used by standard appliances. Our key independent variable is average electricity price for each of the fifty U.S. states for the years 2000-2009, which is provided by the U.S. Department of Energy. In addition to contributing to the research on the effect of electricity prices on the purchase of energy-efficient appliances, our second focus is on the consequent effect on carbon emissions with and without a price on carbon. Many studies estimate the potential savings from avoiding costs associated with greenhouse gas emissions. But relatively few look at the reduction in emissions from using energy-efficient appliances, with the majority of studies based on building efficiency and automobile efficiency. Methods The first step is to estimate the effect of variation in state electricity prices on variation in state market share of energy efficient appliances. Then we predict the change in the market share of energy-efficient appliances after incorporating a carbon price into the price of electricity, In order to estimate the first step: Percent ENERGY STARi = β0 + β1 Electricity Pricei + β2 Per Capita Incomei + β3 Percent Owner Occupiedi + β4 Percent Bachelorsi + εi where Percent ENERGY STARi is the mean of the percent of all refrigerators sold that were ENERGY STAR certified in state i, Electricity Pricei is the mean residential price of electricity (in cents) in state i, Per Capita Incomei is the mean personal income in thousands of dollars per capita in state i, Percent Owner Occupiedi is the mean of the percent of households that are owner occupied in state i, and Percent Bachelorsi is the mean of the percent of the population over age 25 with a bachelor’s degree or higher in state i. We also consider variables that reflect a state’s attitude towards energy efficiency, alternatives and renewables. States that have a more favorable view towards these attributes may have constituents who are more likely to purchase energyefficient appliances. Two approaches are used to account for this variation between states: in one variation of the model, regional dummy variables are added; and in another variation, the American Council for an Energy-Efficient Economy (ACEEE) score representing a state government’s commitment to policies that support energy conservation is added. Regional dummy variables identify states by region as defined by the U.S. Census Bureau. The U.S. EIA cautions against pooling the data for 2004-2009, as the sample of retailers reporting the sale of energy-efficient appliances changes from year to year. Our initial results are based upon averages over the five years. Ordinary least squares appears to provide a better fit of the data than log-linear estimation. Also, the error terms appear to fit classical OLS assumptions, as we did not see evidence of heteroskedasticity or spatial relationships in the residuals. We also use the data in panel form, which gives give 500 rather than 50 observations, but question panel data results where there intrastate variability within the ten years is much smaller than interstate variability as in using state averages for the OLS estimates. We estimate consequent reductions in carbon by relating energy use to carbon emissions, and the added effect of a carbon price on the electricity sector. Results We find average price elasticities of demand for room air conditioners, refrigerators, clothes washers, and dishwashers of 0.29, 0.22, 0.04, and 0.01 respectively. Resources for the Future (2010) estimates that a carbon price would increase the average electricity price by at most $0.04/kWh, which would increase the market share of ENERGY STAR room air conditioners by 5%, 3% for refrigerators, and less than 1% for clothes washers and dishwashers. In turn, a carbon price on the order of $0.04 per kilowatt hour would only increase market share by at most 5% for room air conditioners and 3% for refrigerators and in turn, reduce carbon emissions by 0.2%. Conclusions The market share of ENERGY STAR appliances show an inelastic response to electricity prices, while also responding to the share of adults with a bachelor’s degree, the percent of homes that are owner occupied, and cultural attitudes. One factor that may dilute the effect of electricity prices is the availability of rebates on ENERGY STAR appliances. Consumers showed a significant response to rebates, and an upfront incentive may be a more powerful stimulus than electricity price, which factors in over the lifetime of the appliance. To the extent that these results can be extrapolated to other types of ENERGY STAR appliances, or to other investments in energy efficiency in general, reliance on a Pigovian tax or other mechanism to internalize the social costs of pollution from using energy may not have as much impact as some might hope. If other market failures exist that prevent the efficient level of investment in energy efficiency, such as asymmetric information, credit market inefficiencies, bounded rationality, or the landlord-tenant problem, then several targeted policy approaches may be needed.