Price elasticities of retail energy demands in the United States: New evidence from a panel of monthly data for 2001–2016
Introduction
Energy demand growth projections are necessary for energy policy modeling [1] and resource planning [2], [3]. Following [1,p.5], an intuitive representation of a growth projection is the sum of: (a) the demand change due to price changes; (b) the demand change due to changes in such non-price factors as income and appliance stock; and (c) the demand change attributable to demand-side management (DSM) initiatives (e.g., a government’s energy efficiency (EE) standards and an electric utility’s DSM programs [4]).
This paper focuses on price-induced changes in demand (a). Eq. (1) below shows that a price-induced demand change can be calculated as:where = demand change over one period; = demand in period ; = demand price elasticity with respect to the price ; and = percent change in over one period. For clarity, we use the first price to denote the own-price and to denote the own-price elasticity. If is the price of a substitute (complement), the cross-price elasticity is positive (negative).
To appreciate the real-world significance of price elasticities, consider an electric utility that is planning to meet its customer demand over the next 10 years and forecasts sustained increases in electricity prices due to the cost of complying with a renewable portfolio standard (RPS). If customers are price inelastic, the utility will need to procure more renewable generation to comply with the RPS. If customers are more price elastic, the utility can procure less renewable generation and still comply. If the utility uses an inaccurate price elasticity estimate, it will over- or under-procure renewable energy, leading to excessive costs or compliance penalties. This example illustrates how price elasticity estimates used in investment planning translate into investment decisions, with implications for costs and risk allocation.
Despite the importance of price elasticities, elasticity estimates vary widely. For instance, to inform national energy policies, the U.S. Energy Information Administration (EIA) publishes an Annual Energy Outlook (AEO) pursuant to the Department of Energy Organization Act of 1977 [5]. Its AEO is based on the national energy modeling system (NEMS) [6], which uses price elasticity assumptions to develop energy demand growth projections [7]. For retail electricity demands, the 2017 AEO assumes a price elasticity of −0.15 for the residential class and −0.25 for the commercial class [7, Chapters 4 and 5]. Fifteen years ago, the AEO assumed a price elasticity of −0.20 for the residential class and −0.10 for the commercial class [8, Table 4].
While the 2017 AEO’s elasticity assumptions resemble the −0.16 to −0.25 estimates used by the U.S. Environmental Protection Agency (EPA) [9], they sharply contrast with those adopted by some North American electric utilities. For instance, BC Hydro, which serves the Canadian province of British Columbia, uses an own-price elasticity estimate of −0.10. This estimate is comparable to the estimates used by two Pacific Northwest utilities: −0.05 to −0.15 for the residential class and −0.10 for the non-residential class [10, Table 4]. An electric cooperative in the state of Kentucky uses −0.25 for the residential class, −0.10 for the commercial class, and −0.05 for the industrial class [11]. In general, utility estimates appear to be lower than those used by federal government agencies and are consistent with a conventional assumption of inelastic demand.
Beyond regional differences in price elasticities [12], utilities’ lower price elasticities may be explained by their aversion to the risk of insufficient supply. Large electricity price elasticity estimates can cause them to underestimate electricity demand growth, which may in turn lead to capacity shortages. Hence, an electric utility’s elasticity assumptions reflect a tradeoff between supply capacity costs and the political and economic costs of customer outages, as explained by [13] in the context of efficient resource planning under demand and supply uncertainties.
Are higher or lower energy price elasticity estimates more consistent with recent empirical evidence? Has energy demand become more price elastic over time as a result of, for instance, more advanced metering and more consumer education around energy efficiency opportunities? To answer these questions, this paper uses a Generalized Leontief (GL) demand system [14] and a newly constructed panel of monthly data for the lower 48 United States over the 16-year period 2001–2016 to estimate price elasticities of retail energy demand for electricity, natural gas and fuel oil for three major customer classes: residential, commercial and industrial.
Our key findings are as follows. First, our electricity price elasticity estimates are closer to those used by some North American utilities, implying that electricity demand is largely unresponsive to changes in price. Second, retail electricity demand is generally less price-responsive than retail natural gas and fuel oil demand. Third, electricity and natural gas are substitutes, with small and positive cross-price elasticity estimates. Finally, retail fuel oil demands are unresponsive to changes in electricity prices though not to changes in natural gas prices.
Our empirical analysis makes three contributions to the literature on energy demand estimation. First, we use a GL demand system and a newly constructed panel of monthly data to estimate price elasticities of retail energy demand for the 48 lower United States. Our choice of the GL specification is motivated by its formal test of the hypothesis of zero substitution among the three principal energy types: electricity, natural gas and fuel oil. To the best of our knowledge, no other demand study has done a joint estimation like ours for the three major customer classes.1
Second, we show that retail energy demands in the United States are highly price-inelastic, more so for electricity than for natural gas and fuel oil. This finding suggests that higher energy prices—for instance, resulting from environmental regulation or higher fuel prices—will have a limited impact on energy demands. While the retail price increases will only minimally curtail energy demands, there are well-known and time-proven measures for achieving discernable demand reductions. A good case in point is that mandatory EE standards can increase the market penetration of highly energy-efficient appliances (e.g., air conditioners and water heaters with high EE ratings) that may not have been bought by some households despite escalating energy prices. Similarly, an electric utility may implement such DSM programs as energy audits, public education and financial rebates to promote energy-efficiency investments by its customers. Thus, cost-effective EE standards and DSM programs still have an important role to play in managing energy demands, mitigating energy price risk, and reducing emissions and other environmental impacts from energy use.
Finally, our method is general. It can be readily implemented in other countries with similar data availability. One case that has significant academic and policy interest is the world’s largest energy consuming country, China, whose annual energy consumption data by sector and province are also publicly available.2 To be sure, constructing a Chinese data file is by no means easy, especially in assembling suitable price data series necessary for a system estimation of energy demand. With improved data availability, however, we posit that our approach can be used for estimating the price elasticities of retail energy demand in China.
This paper proceeds as follows. Section 2 contextually links our paper to the energy demand literature, presents our GL demand system, and describes our panel data. Section 3 reports our empirical results. Section 4 concludes.
Section snippets
Contextual linkage
To contextually link our paper to the vast literature on energy demand estimation, this section reviews retail demand studies of electricity, natural gas and fuel oil for three customer classes: residential, commercial and industrial. As such, it is not a review of time series modeling of electricity loads [15], [16] or the estimation of capital-labor-energy-material models [17], [18], [19], [20], [21], [22]. It also excludes demand models for transportation fuels (e.g., gasoline, diesel and
Regression results
Obtained under the non-negativity restrictions of , Table 5 reports the ITSUR regression results for the static GL system of energy intensities. All seven regressions have adjusted values above 0.80, indicating their reasonable fit with the monthly data sample. In other words, the regressions’ fit is unlikely to materially improve by including additional regressors that may be formed by second-order terms of the RHS variables to account for possible nonlinearities in the energy
Conclusion
This paper uses the GL demand system and a panel of monthly data covering 2001–2016 for the lower 48 United States to estimate own-price and cross-price elasticities of retail demand for electricity, natural gas and fuel oil for residential, commercial and industrial customers. Our results indicate that retail energy demand in the lower 48 states was highly price inelastic during this period.
Our electricity own-price elasticity estimates are consistent with those used by some U.S. electric
Acknowledgements
We thank the reviewers and editors for their very helpful comments, which have greatly improved our analysis and exposition. This paper is funded by the first author’s research grant (R3698) from Education University of Hong Kong. Without implication, all errors are ours.
References (76)
Optimization methods for electric utility resource planning
Eur J Oper Res
(1995)- et al.
Survey of Western U.S. electric utility resource plans
Energy Policy
(2014) Electricity price forecasting: a review of the state-of-the-art with a look into the future
Int J Forecast
(2014)- et al.
The translog function and the substitution of equipment, structures, and labor in US manufacturing 1929–68
J Econom
(1973) - et al.
A meta-analysis on the price elasticity of energy demand
Energy Policy
(2017) - et al.
A review of electricity product differentiation
Appl Energy
(2014) The welfare econometrics of peak load pricing for electricity
J Econom
(1984)- et al.
The effect of utility time-varying pricing and load control strategies on residential summer peak electricity use: a review
Energy Policy
(2010) - et al.
Consistency of residential customer response in time-of-use electricity pricing experiments
J Econom
(1984) - et al.
Residential winter kWh responsiveness under optional time-varying pricing in British Columbia
Appl Energy
(2013)