Learn about the methodology used to develop the system of energy intensity indicators, as well as caveats and cautions related to interpretation and data issues.
Energy Intensity Indicators Methodology
The system of energy intensity indicators developed by DOE is based upon a hierarchical framework that begins with detailed indexes of energy intensity for various sectors of the economy, which are ultimately aggregated to an overall energy intensity index for the economy as a whole.
The definitions of energy used in the system depend upon how the transmission and generation losses associated with the use of electricity by various sectors are treated. Finally, the system recognizes that at an aggregate level the conventional measures of energy intensity are not all traceable to improvements in energy efficiency.
Many of these other "structural" factors influencing overall energy consumption are explicitly captured in the system developed here. These elements are discussed in summary fashion here.
The files listed below contain methodology documentation and related studies that support the information presented on this website. The files are available to view and/or download as Adobe Acrobat PDF files.
- 2003. Energy Indicators System: Index Construction Methodology
- 2004. Changing the Base Year for the Index
- Boyd GA, and JM Roop. 2004. "A Note on the Fisher Ideal Index Decomposition for Structural Change in Energy Intensity." The Energy Journal 25(1):87-101.
- Belzer, David B. and Katherine A. Cort. Statistical Analysis of Historical State-Level Residential Energy Consumption Trends
- Belzer, David B. 2014. A Comprehensive System of Energy Intensity Indicators for the U.S.: Methods, Data, and Key Trends. PNNL-22267.
The construction of indexes that show the performance of energy intensity over time and account for structural changes is an exercise in decomposition of effects. Advances have been made recently that allow for this decomposition so that many of the attributes of an "ideal index" are captured in the decomposition used here.
One of these attributes is "perfect aggregation," which allows all higher-level indexes to be constructed so that they include all the information available at the lower levels, and allows this information to be integrated at the higher level. Unfortunately, that attribute slightly distorts the year-to-year changes in the activity measure in a way that would make these measures deviate from published figures, so the approach used to construct intensity indices for this website was modified to allow almost perfect aggregation and yet have the activity measures conform to the published numbers.
A thorough discussion of the methods used for the construction of the indexes can be found in the methodology summary paper (2003) and in Section 2 of the 2014 comprehensive report above.
Caveats and Cautions When Interpreting Energy Intensity Data
This website contains a diverse collection of indicators that track changes in energy intensity at the national and end-use sector levels (after taking into account other explanatory factors). Indicators are based on readily available and publicly accessible data, although some of this data has been interpolated between published years, or extrapolated beyond the last published year.
To facilitate appropriate interpretation, below is discussion of data issues and of issues related to the treatment of other explanatory factors.
Interpretation of Energy Intensity Indicators
The purpose of this national system of indicators of energy intensity is to provide an understandable, readily available, and transparent measure of the performance of the economy with regard to its use of energy.
The intent is to periodically update this system of indicators and make the data available to the public and to analysts. The purpose of developing a new index of energy intensity is to have it reflect, as closely as possible, improvements in energy efficiency.
This can only be accomplished if other explanatory factors—things that would normally affect the energy-intensity index, but do not reflect improvements in energy efficiency—are taken into account. These other explanatory factors are often referred to as structural and behavioral components or effects, to differentiate them from underlying improvements in the use of energy. So in addition to looking at energy intensity, this system of indicators also tracks the important structural changes that affect the use of energy.
In the construction of this system of indicators, a number of issues have been addressed, but these issues need to be highlighted to the user, because they may affect any conclusions that are drawn from the indicators. More detailed discussion of these issues can be found in the 2014 comprehensive report that describes key trends, methodologies, and data sources.
Data Issues in Energy Intensity Indicators
Because all data are not collected every year, annual updating of the system of indicators requires that data that are not inherently annual be extrapolated. The documentation of data sources provides a description of how these interpolations were done, which are subject to analytical judgment, rather than scientific certainty.
The intent is to revise the system of indicators as missing pieces of data are published, so that extrapolated data are more closely matched to published data over time. The following items are of special note:
Data for the transportation sector are generally available from private and governmental source. However, some very small (in terms of energy use) transportation sectors do not have explicit energy consumption and activity measures. At present, these sectors include intercity and school buses. For the recent (2014) updates to the indicators, extrapolations are made from prior estimates.
Electricity consumption data for the manufacturing sector is taken from the Annual Survey of Manufactures (ASM) through 2011 and the fuel data is benchmarked to the various Manufacturing Energy Consumption Surveys (MECS), with latest data for 2010. (The MECS has been conducted in 1985, 1988, 1991, 1994, 1998, 2002, 2006, and 2010.)
The interpolations between MECS years and the extrapolation for 2011 are based upon developing aggregate fuel prices per million Btu for each 3-digit NAICS (North American Industrial Classification System) industry and using these fuel prices with purchased fuel information from the ASM to estimate fuel quantities in million Btu.
As there are no specific industry fuel prices or quantity information for the non-MECS years, these imputations necessarily contain more error than the data for the MECS years.
The estimated energy consumption estimates, for both electricity and fuels, for non-manufacturing portion of the industrial sector (agriculture, mining, and construction) is derived from various sources and reports. Expenditures for fuel for agriculture comes a special survey conducted by the U.S. Department of Agriculture.
The primary data for mining and construction are derived from the periodic Economic Census (in years ending "2" or "7"). The information for fuels is now collected only in terms of expenditures, and so an unknown element of error is introduced in assigning an appropriate fuel price to convert to a Btu measure.
Residential Building Energy
A methodological decision that has a major impact on the behavior of the energy intensity in the residential sector is the choice of energy per square foot as the primary intensity measure. This decision was prompted by the fact that space conditioning remains the largest single end use in the residential sector, making up more than 40% of total energy consumption.
As discussed in the 2014 comprehensive report, data from the biennial American Housing Survey (and earlier Annual Housing Survey) have been employed to estimate the average square footage for single-family and multi-family housing units. Average square footage for manufactured homes was derived from various editions of the EIA's Residential Energy Consumption Survey (RECS).
Given sampling variation that accompanies the values derived from any survey, and the need to interpolate or extrapolate estimates for non-survey years, one should recognize that the estimate of total residential floor area in any given year should be viewed as approximate.
Commercial Building Energy
There is a significant divergence between the aggregate commercial buildings sector energy consumption data, derived from utility-reported sales as published in the Annual Energy Review and the Monthly Energy Review, and the energy consumption, as collected in the end-use surveys, Commercial Building Energy Consumption Survey (CBECS). These differences, as well as the occurrence of some implausible changes in energy intensity in adjacent CBECS, inhibit the development of consistent energy intensity time series for subsectors of the commercial sector.
The user should recognize that the energy consumption aggregates in the Annual Energy Review and the Monthly Energy Review pertain to all commercial customers, some of which relate to non-building uses of energy (e.g., water and sanitary service, street lighting, and communications equipment). Another issue relates to the reclassification of electricity consumption by utilities between the industrial sector and the commercial sector, as exhibited in their reporting of sales by customer class to EIA.
Based upon a state-by-state analysis of reclassification of sales, commercial electricity consumption in recent years was reduced by about 3% and added to industrial consumption.
The methodology underlying this adjustment was described in a report published in 2007 by D. Belzer of Pacific Northwest National Laboratory. This report discusses non-building use of electricity as well as these reclassifications and is included on the indicators website (PNNL-16820, Estimates of U.S. Commercial Building Electricity Trends: Issues Related to End-Use and Supply Surveys).
Issues Related to Other Explanatory Factors
Although the system of indicators tracks energy intensity over time, it also identifies and isolates other changes that are going on in the economy that have no bearing on the efficiency with which energy is used. These changes are referred to as structural changes. These structural elements give rise to a change in energy use per unit measure of output, but do not reflect improvements in the underlying efficiency of energy use.
For example, the electronics industry may grow much faster than the steel industry over time, and since electronics are substantially less energy intensive than steel, this growth creates the impression that the manufacturing sector is less energy intensive than before.
For the intensity index to better reflect the efficiency of manufacturing energy use, there is a need to normalize the intensity measure so that these differential growth rates are taken into account.
Structural changes occur in each of the energy-using sectors of the economy, but they are quite different from one sector to another. For example, a transportation indicator constructed for that end-use sector will have a different interpretation than one constructed at the subsector level, or one constructed for another end-use sector.
A method of index construction has been chosen so that these structural changes can be aggregated from the bottom up, and integrated into the overall index, whether the index is at the end-use sector level (i.e., freight and passenger transport in the transportation sector) or at the aggregate, economywide level.