March 5, 1996
EIA Session Summary
Energy efficiency is an interesting and useful
concept for analyzing energy use. Energy efficiency is analyzed in relative terms; that
is, "is energy efficiency increasing or decreasing," rather than "what is
the energy efficiency of a particular energy-requiring product or service?"
Considered in relative terms, energy efficiency is a fairly simple concept to describe
theoretically--energy-efficiency improvement occurs when more or enhanced goods or
services are provided with level energy inputs. Energy-efficiency loss occurs when more
energy input are required to produce the same or reduced products or services.
In reality, however, efficiency is
difficult to measure. Because energy needs to be assessed relative to the amount of
product or service provided, energy-use rates, commonly called energy intensities, are the
measures ordinarily used to assess efficiency trends. Depending on the type of measure,
energy intensities may reflect not only energy efficiency but also changes in other
effects such as changes in industry mix and energy services used. However, the advantage
of considering individual energy intensities is that you may easily measure and track them
How Should Energy Efficiency be Defined in the
Individual workshop participants argued that:
- Energy Efficiency is the Reciprocal of Energy
Intensity. Energy efficiency may be merely the reciprocal of energy intensity, but
only if it is done correctly.
- Definition Depends on the Level of Analysis.
Sometimes, considering energy intensity is more rational at the process or step level
only, while at the industry level, energy intensity may mask the structural changes which
are not resulting from efficiency improvement. Across industries, energy-intensity
measures are preferable to energy-efficiency measures.
- Need to Consider the Total Production Equation.
The relevant concept of energy efficiency in the industrial sector may be energy
efficiency with respect to getting production out the door. This requires consideration of
both energy efficiency and the production equation.
- Need to Consider "Best Practice" Energy
Consumption. Energy efficiency is the difference between the current or average
practice energy consumption and the "best practice" energy consumption.
- "Best Practice" Does Not Work in Real
Life. The concept may work at the process level, but not at the industry level.
Furthermore, it cannot be used in an aggregate analysis. Exxon used to do this kind of
analysis in its refineries, but it found that the "Best Practice" was not a
meaningful concept; there may be no practical way to get to the "Best Practice."
Who is the Audience for Industrial Sector Energy-Efficiency Indicators?
The Energy Information Administration (EIA) has an
expanded sense of audience. EIA is currently involved with business process reengineering
and employing Total Quality Management with the goal of satisfying its customers.
Nevertheless, the ability to satisfy the industrial-sector customer poses quite a
challenge to EIA, especially in an environment of reduced resources. EIA's main source of
manufacturing data are collected via the Manufacturing Energy Consumption Survey. The
nonmanufacturing sector was not addressed due to a serious lack of data. The workshop
participants, although diverse in background, shared a common need--the need for
disaggregated data for the entire industrial sector.
The following are comments from workshop
participants representative of different EIA audiences:
- Academic and Research and Development (R&D)
Audience.Two major uses by the academic and R&D management communities is
appropriate: long-term industrial sector forecasting (for which it is always preferable to
have end-use data if you can afford it); and total market forecasting based on end-use
demand and market shares. Both uses require a disaggregated approach, with more detail
than 4-digit SIC. Ultimately, this audience needs to look at energy-efficiency indicators
by process. For example, separate energy-efficiency indicators for aluminum melters,
aluminum dryers, aluminum heat-treaters, etc.
- Policy or Program-Oriented Audience. A
policy-oriented audience needs to identify opportunities or target markets within the
industrial sector. Are we on a trend? Will opportunities present themselves or have we
already found them all? Only with disaggregated analysis, however, can policymakers learn
where to focus their efforts.
- Plant Manager Audiences. Plant managers
require far finer end-use detail than EIA can provide with existing data. Some plant
operators would scoff at indicator analysis, because the aggregate data makes it difficult
to use at the plant level.
- Utility Audience. Electric utilities looking
to offer conservation or other energy services may use energy consumption data in their
marketing efforts, but they don't necessarily use efficiency data. They look simply for
the biggest energy consumers, not for the efficiency opportunities.
What is the Appropriate Aggregation for Analysis of
Energy Efficiency in the Industrial Sector?
Several workshop participants differed in the level
and type of aggregation at which EIA should produce energy-efficiency indicators:
- Produce Input/Output Tables. Input/output
accounts distinguish between industry-basis and commodity-basis classification. These
might be useful distinctions for this kind of efficiency analysis. Input/output tables can
measure the gross output and final demand by over 500 categories of commodity. However,
these are not commodities in a physical sense; they are hodge-podge groupings.
- EIA Should Disaggregate by Process. EIA should
consider disaggregating industries by products or processes instead of SIC. A group of
experts could be called in to define the appropriate groupings. Currently, significant
changes to the SIC groupings in the 1997 Census of Manufactures are anticipated. For
example, there will be significant shifts among the manufacturing subsectors. At the
process level, energy intensity is revelant, but at the whole-industry level over time
intensity does not reveal efficiency improvement. Metal fabrication industries, as an
example, should be aggregated by process, not SIC.
- Report by 3-Digit SIC instead of 4-Digit.
Reporting Table 38 of MECS by 3-digit SIC (instead of 4-digit SIC with too many withheld
cells) would be a start. There were 22,000 observations for the 1994 MECS; certainly that
should be enough to provide that level of detail without compromising companies' privacy.
- EIA Should Provide Primary and Site (Delivered )
Energy Data. It would also be helpful if the MECS was to distinguish between
electricity and other fuels or between primary and site energy. (Note: Site (delivered)
energy is the amount of energy delivered to a household. Energy generation, transmission,
and distribution losses are not included. EIA used only estimates of site energy in the
report Measuring Energy Efficiency in the United States' Economy: A Beginning, used as the
"straw man" for the EIA workshops.)
- EIA Should Use Subsector Deflators. Deflators
should be designated at subsectoral levels. At least EIA should use 3-digit SIC-specific
deflators weighted up to the 2-digit level. (Note : The EIA uses 4-digit specific
deflators from the Bureau of Economic Analysis, U.S. Department of Commerce.)
What Would be the Most Appropriate Demand
for Industrial Sector Efficiency Analysis?
A demand indicator is the number of energy-consuming
units, or the amount of service or output, for which energy inputs are required--the
denominator in the energy-intensity ratio (energy requirements/demand indicator). The
demand indicator in energy-efficiency analysis is critical; the EIA report offers a
selection of economic demand indicators for the industrial sector.
Workshop participants were very vocal when
discussing the choice of the demand indicator:
- Value-added Demand Indicators. Value-added
demand indicators (including Gross Product Originating (GPO)) may be the wrong choice;
value-added calculated intensity and GPO-calculated intensity vary significantly (in the
report), when they should be measuring the same thing.
- Value of Shipments. It would be preferable to
use value of production or value of shipments as the demand indicator.; the value of
energy consumption may be subtracted from the value of production. The capacity
adjustments used in the report are geared to a collection of industries, but in reality
the industries within the EIA report's clusters may be affected very differently. The
clustering used in the EIA report is flawed; if adjustments are to be done in this matter,
it may be necessary to cluster the industries differently. The Marc Ross model, which
clusters industries by their energy intensities, may be more useful.
- Physical-Based Demand Indicators. The report
covers every demand indicator imaginable except those that would be most useful to
National Energy Modeling System (EIA forecasting model) analysts: something based on
physical flows instead of valuations. The model is calibrated to physical flows. From an
engineer's perspective, physical measures of output, not value should be used. These value
indicators are too subject to price volatility. With necessary adjustments, value of
shipments may be the next best if physical-based indicators are not available.
- Use Different Indicators at Different Levels.
At the disaggregated level, value of output or shipments masks when companies decide to
outsource certain processes. EIA should use value added at the aggregate level and
physical or volume measures for individual industries (4-digit level for SIC's 20-33)
where they are available. Process level indicators would be preferable for SIC's 34-39.
GPO is also an appropriate measurement of demand, although the difference between GPO,
based on return to factors, and value added, based on shipments data, is unclear.
What Are the Most Serious Data Deficiencies in the
Sector that Impede Energy-Efficiency Analysis?
Workshop participants overwhelmingly agreed that
there are very little nonmanufacturing data available within the industrial sector.
Nonmanufacturing includes agriculture, mining, and construction. Workshop participant
commented as follows:
- Need more Nonmanufacturing Data. EIA's
energy-efficiency studies should include a more comprehensive treatment of manufacturing
versus nonmanufacturing groups. Nonmanufacturing is truly a black hole. For modeling
purposes, EIA needs more satisfactory data.
- Agriculture, Construction, and Mining Data Are
Available. The Economic Research Service of the US Department of Agriculture has
end-use consumption and fuel use for agriculture. The dearth of data is not as bad as
people think. The Censuses of Mining and Construction are available for benchmarking.
- Could Use EIA Supply Data. EIA's petroleum
sales data may provide additional information about nonmanufacturing consumers.
Are the Tradeoffs between Frequency of Collection and
Detail of Industrial
(Manufacturing) Sector Data?
Although the workshop was energy-efficiency
focused, it was an excellent forum for EIA to obtain MECS customer input. In light of
budget reductions, EIA may have to decide which direction it has to take, decrease survey
frequency or reduce the level of detail. Overwhelmingly, the workshop participants
preferred detailed data even if the data were collected less frequent; preference was that
an expanded-sample MECS be administered at least during the years of the Census of
Manufactures administered by the U.S. Census Bureau.
Other workshop participants' comments are as
Some Data Can be Eliminated. EIA
could eliminate information on fuel switching, DSM participation, and square footage.
Removing these items from the survey may allow EIA to economize or add other data. EIA
should look at the questions asked in the survey. Which agenda is being measured? There is
more information collected in MECS than just the data used to construct energy-efficiency
EIA Needs to Keep the Methodology
Constant. EIA should keep a constant methodology or if there are changes,
go back and redo the data from previous surveys using the new methodology so that users
can use the data without too much trouble.
- EIA and the U.S. Census Bureau Should Collaborate.
EIA and the U.S. Census Bureau should harmonize the energy survey to reduce manufactures'
burden. Measuring and tracking selected respondents over time can be done annually, but
the big measures should be done once every 5 years.
What is the Role of EIA in Industrial Sector
As follows, the workshop participants displayed many
diverse opinions as to what EIA's role in the provision of energy-efficiency indicators
- Important EIA Role. There is an appropriate
and important role for EIA in this kind of analysis. EIA should interpret data, not just
- Steps EIA Should to Follow. Next steps for EIA
might be as follows: (1) look for better disaggregation, (2) develop and publish a few
indicators, with some description of what's appropriate for what, and (3) add of
physical/volumetric demand indicators.
- Need Multiple Indicators. Several indicators
may be needed as well as a fairly aggregate indicator for policy makers. EIA needs to
first identify clearly the most important questions it can and wants to ask and the
appropriate indicators to generate the respective answers. analysis. Keep in mind that the
needs of different parties will require different answers; for example, the steel industry
is not just interested in volume of steel. They also care about what kind of steel they
sell, as expressed by the going price for a quality/kind of steel.
- Improve the MECS and Leave Efficiency Analysis to
DOE's Policy Office. EIA could leave the efficiency indicator and analysis issues to
the Policy Office and devote its limited resources to improvement of MECS data quality.
(Note: A show of hands indicated that a minority of the participants supports the
proposition that EIA should devote itself to data improvement even if it means abandoning
efficiency indicators if budget limitations dictated a choice.)
- EIA Should Provide the Statistical Tools.
Another option is for EIA to focus completely on data quality but also provide statistical
tools (both formulas and instructions for application) to external analysts, perhaps on a
questions on this topic may be directed to:
Modified: October 17, 1999