1 See S.H. Wade, “Price Responsiveness in the NEMS Building Sector Models,” in Energy Information Administration, Issues in Midterm Analysis and Forecasting 1999, DOE/EIA-0607(99) (Washington, DC, August 1999).

2 For reference case projections see Energy Information Administration (EIA), Annual Energy Outlook 2003, DOE/EIA-0383(2003) (Wash-ington, DC, January 2003). For modeling assumptions and general techniques see EIA, Assumptions to the Annual Energy Outlook 2003, DOE/EIA-0554(2003) (Washington, DC, January 2003). For greater modeling detail see EIA, Model Documentation Report: Residential Sector Demand Module of the National Energy Modeling System, DOE/EIA-M067(03) (Washington, DC, January 2003); and Model Documentation Report: Commercial Sector Demand Module of the National Energy Modeling System, DOE/EIA-M066(03) (Washington, DC, March 2003).

3 When prices are referenced in this paper, the reference is always to “real” energy prices, adjusted to remove any effects from general inflation in the economy. The internal calculations of the residential and commercial models operate on real prices.

4 Equipment that is still capable of providing energy services but has operating costs (fuel and maintenance) that exceed the annualized capital and operating costs of newer equipment is “economically obsolete.” The retirement and retrofitting of economically-obsolete equipment is simulated in the commercial model, adding another dimension to its potential price responsiveness.

5 Both the residential and commercial models employ “myopic” expectations of future energy prices—that is, the current energy price is used in formulating equipment purchase decisions.

6 The residential model insulation upgrades are an example of an effect that does not reverse in response to lower energy prices. Once installed, the insulation is assumed to last for the life of the structure.

7 An exception is the unusual case of what are referred to as “Giffin goods.” By definition, for these goods, price reductions lead to reductions in demand. Real-world examples are hard to come by, but a good purchased primarily for its “conspicuous consumption” attributes might exhibit this type of price response.

8 C. Dahl, A Survey of Energy Demand Elasticities in Support of the Development of the NEMS, Contract No. DE-AP01-93EI23499 (Washington, DC, October 1993). The Dahl survey incorporated results from other survey articles and from newer studies, not reviewed previously. From prior surveys, the residential/commercial own-price elasticities for total energy ranged from -0.012 in the short run to -0.44 in the long run. Focusing on studies of aggregate time series data, demand elasticities for electricity from more recent studies averaged from -0.22 in the short run to -0.91 in the long run for residential buildings and from -0.22 in the short run to -0.82 in the long run for commercial buildings. For natural gas the averages from more recent studies were -0.13 (short run) to -0.68 (long run) for residential buildings and -0.26 (short run) to -0.99 (long run) for commercial buildings.

9 Equipment that is used only for short periods during the year (e.g., air conditioning in northern climates) will have relatively low energy consumption and thus low energy costs. In such cases, equipment choice will be less influenced by energy prices than in areas where equipment is more heavily used.

10 The residential model projects equipment choices using a “continuous function” approach to model the tradeoff between equipment cost and equipment efficiency, whereas the commercial model employs a “discrete algorithmic” approach. As will be seen from the simula-tion results, the overall behaviors of the two models are similar. For further details on equipment choice formulations, see the model docu-mentation reports and the AEO2003 key assumptions (cited above).

11 A 5-percent increase in energy prices is assumed to result in a 1-percent increase in the shell efficiency index for residential buildings. No adjustment to shell efficiency is made for price declines.

12 For the commercial model, the same end uses subject to the long-run price elasticity response are also covered by the efficiency rebound effect. For the residential model, space conditioning is covered by the rebound effect. For a discussion of the rebound effect, see J.D. Khazzoom, “Economic Implication of Mandated Efficiency Standards for Household Appliances,” Energy Journal, Vol. 1, No. 4 (1980), pp. 21-40.

13 Efficiency rebound effects for both the residential and commercial models are based on a parameter that results in a 0.15-percent increase in consumption for a 1-percent increase in efficiency.

14 See Energy Information Administration, A Look at Residential Energy Consumption in 1997, DOE/EIA-0632(97) (Washington, DC, November 1999); and A Look at Commercial Buildings in 1995, DOE/EIA-0625(95) (Washington, DC, October 1998), for more information on these surveys and results.

15 RECS 2001 and CBECS 1999 are not yet available in printed reports. Links to the currently available information are as follows: for RECS, see http://www.eia.doe.gov/emeu/recs/recs2001/detail_tables.html; for CBECS, see http://www.eia.doe.gov/emeu/cbecs/contents.html.

16 Arthur D. Little, Inc., “EIA Technology Forecast Updates: Residential and Commercial Building Technologies—Reference Case,” Ref-erence No. 8675309 (October 2001).

17 See the model documentation reports (cited above) for a description of the distributed generation modules. AEO99 modeled commer-cial cogeneration, but with a relatively simple single-equation representation that did not include explicit technologies.

18 Distributed generation natural gas-based technology characterizations are from ONSITE SYCOM Energy Corporation, The Market and Technical Potential for Combined Heat and Power in the Commercial/Institutional Sector (Washington, DC, January 2000). Photovoltaic technology characterizations are from U.S. Department of Energy, Office of Energy Efficiency and Renewable Energy, and Electric Power Research Institute, Renewable Energy Technology Characterizations, EPRI-TR-109496 (Washington, DC, December 1997).

19 For information on Energy Star homes, see web site www.energystar.gov. For information on PATH homes, see Partnership for Advancing Technology in Housing, web site www.pathnet.org.

20 The simulations are based on “stand-alone” runs of the commercial and residential models. This is appropriate, since the purpose of this paper is to describe the responses of these models. In an “integrated” NEMS model run, macroeconomic effects due to large swings in energy prices could affect the calculated elasticities, possibly increasing the own-price sensitivity of the integrated model results (i.e., higher energy prices reduce economic activity, leading to further consumption decreases). Elasticities are measured using the logarithmic percent-age change formula given by: elasticity = ln(q1/q0)/ln(p1/p0), where p0 and q0 are base prices and quantities, and p1 and q1 represent an alter-nate price-quantity combination.

21 The earlier paper, reporting AEO99 results, was based on simulations using a 10-percent price increase instead of a price doubling. A price doubling was chosen for this report on elasticities, because price paths with such large changes are relevant to current policy analysis. Higher energy prices increase the monetary value of energy savings that accrue to higher efficiency purchases and can thus lead to greater long-run consumption responses.

22 As mentioned above, the short-run behavioral adaptations are spread over a 3-year interval in AEO2003, whereas the entire effects were assumed to occur in the first year in AEO99. Fuel price changes also affect capital purchases for retiring equipment in the first 3 years of a simulated price change; however, no attempt has been made to isolate the capital purchase-induced component during the initial phase-in period. Capital purchases will build gradually over the forecast horizon as more equipment becomes available for replacement.

23 The 20-year horizon was chosen because NEMS currently runs through 2025, and the initial price increase is imposed in 2005. For equipment such as commercial boilers and residential furnaces, additional long-run effects could occur beyond 2025.

24 Examples of other miscellaneous uses include service station equipment, automated teller machines, telecommunications equipment, medical equipment, and elevators and escalators.

25 Some negligible negative cross-price elasticities were found for AEO99, but in all cases they rounded to 0.00.

26 There are also some effects of altered equipment purchases during the first 3 years beyond what would have occurred in the first-year AEO99 short-run results. The equipment-related components during this period are not separately identified.

27 Under the more recent technology characterizations used for AEO2003, distillate equipment is generally more costly relative to natural gas-based equipment than was the case for AEO99. This change probably is responsible for most of the reduction in long-run sensitivity.

28 Based on results prepared for the earlier paper reporting AEO99 results, it is estimated that roughly one-half of the reported differences in the long-run own-price elasticities of electricity for both sectors are due to the use of price doublings for analyzing the AEO2003 models.

29 C. Dahl, A Survey of Energy Demand Elasticities in Support of the Development of the NEMS, Contract No. DE-AP01-93EI23499 (Washington, DC, October 1993).