6. The Industrial Sector
The industrial sector encompasses more than 3 million establishments engaged in manufacturing, agriculture, forestry, fishing, construction, and mining. These industries require energy to light, heat, cool, and ventilate facilities (end uses characterized as energy needed for comfort). They also use energy to harvest crops, process livestock, drill and extract minerals, power various manufacturing processes, move equipment and materials, raise steam, and generate electricity. Some industries require additional energy fuels for use as raw materials (feedstocks) in their production processes. Many industries use byproduct fuels(72) to satisfy part or most of their energy requirements. In the more energy-intensive manufacturing and nonmanufacturing industries, energy used by processes dwarfs the energy demand for comfort.
The value of industrial output has continued to increase, while total energy consumed by the industrial sector has fallen. This relationship is shown in Figure 6.1, where the consumption index for both primary and site energy is greater than the output index before 1980 and less afterward, with the gap consistently widening in the late 1980's. New energy-efficient technology and the changing production mix from the manufacture of energy-intensive products to less intensive products account for much of this difference.(73)
In this chapter, only the manufacturing sector is considered in the development of energy-intensity indicators. Data are insufficient to permit the development of robust energy-efficiency indicators for the nonmanufacturing sector.(74)
Historically, data on the manufacturing sector's energy use has also been scarce. This lack of data has made energy-use analysis more difficult here than in the other sectors. At the same time, an opportunity is available. A well-conceived efficiency analysis could be more important and more beneficial here than any other sector because of the amount of energy consumption represented in the sector and the perceived potential for additional efficiency improvements.
Industrial energy consumption and associated output data are classified by Standard Industrial Classification (SIC) in most surveys of establishments.(75) After the transportation sector, the manufacturing sector consumes the most energy in the United States. In 1991, 20.3 quadrillion Btu of energy for all purposes including use as feedstocks, or about one-third of the total end-use energy was consumed in the manufacturing sector (Figure 6.4).
Of the 20 major industry groups in the manufacturing sector, in 1991, 6 groups accounted for 88 percent of the consumption of energy for all purposes: Food and Kindred Products; Paper and Allied Products; Chemical and Allied Products; Petroleum and Coal Products; Stone, Clay, and Glass Products; and Primary Metals. These six account for only 40 percent of the output value for manufacturing, and as a result, are very energy intensive in their production, the exception being the Food and Kindred Products major group. This SIC industry is a high-energy consumer, but not very energy intensive as we will see later in this chapter.
For the purpose of this analysis, the manufacturing sector is divided into three major groups; high-energy consumers; high value-added consumers; and low-energy consumers. These are summarized in Table 6.1. The most important group, high-energy users, is presented in detail.
First, the major data sources are described in detail. As in the previous chapters, the trend in energy consumption in the manufacturing sector is shown, followed by a discussion of the demand indicators that influence the amount of energy consumed, namely gross output, value of shipments, industrial production, value added, and gross product originating. Changes in these indicators will be discussed as well as possible adjustments such as capacity and inventory adjustments. The most frequently used energy-intensity indicators will be compared, followed by a comparison of capacity and inventory-adjusted intensity indicators as a form of "closing in on energy-efficiency indicators." Last, the strengths and limitations of these energy-intensity indicators will be explored.
Table 6.1. Type of Manufacturing Industry Group
U.S. Department of Commerce: Bureau of Census
Census of Manufactures/Annual Survey of Manufactures: Historical and Current Series
The Census of Manufactures (CM) and the Annual Survey of Manufactures (ASM) conducted by the U.S. Department of Commerce, Bureau of the Census, provide economic data, such as sales, employment, and expenditures by SIC. The CM and ASM collect the same information, and together provide an annual series. The CM is conducted every 5 years, and collects the same information from essentially the entire population of manufacturing establishments. The CM does not collect data from very small establishments, which are represented instead by administrative records from other sources. In the years when the CM is not conducted, the ASM collects the same information from a sample of 45,000 to 55,000 establishments.
Between 1974 and 1981, the ASM collected data on the amounts of individual fuels and electric energy produced offsite, and expenditures for those fuels by SIC. Since 1981, the only energy data provided by the ASM are total expenditures for offsite-produced fuels, expenditures for electricity, and the amount of electricity produced offsite. The ASM continues to provide information on employment, value of shipments, and other important economic characteristics.
Manufacturing Energy Consumption Survey: Current Triennial Series
The Manufacturing Energy Consumption Survey (MECS), conducted by the Energy Information Administration (EIA), provides more detailed energy data than the ASM previously did or currently does. The MECS is the most comprehensive source of national-level data on energy-related information for the manufacturing sector. The MECS provided three different measures of manufacturing energy consumption. These measures differ in terms of how offsite-produced energy, feedstocks, and byproduct energy are accounted for. The MECS measure of offsite-produced energy corresponds to the ASM "purchased fuels" definition.
The MECS is a sample of approximately 12,000 (increased to 16,000 for 1991) establishments subsampled from the ASM sample. Thus the population represented by the MECS matches that covered by the ASM. However, because the MECS is only a sample from the ASM, the MECS estimates do not exactly coincide with the ASM for a given survey year, due to sampling variability.
MECS data are available for data years 1985, 1988, and 1991. The next MECS will provide data for 1994.
Annual Production Indices
The Federal Reserve Board (FRB) produces an annual series of production indices, by two-digit manufacturing SIC. The basis of this index is specific to each SIC. In many cases, it is tons of product, indexed to a base year. In other cases, a different production measure is used, indexed to the same base year. The FRB production indices are given for manufacturing as a whole, and for most two-digit SIC manufacturing groups. For SIC 23 (Apparel) and SIC 39 (Miscellaneous), however, no FRB production index is defined.(76)
U.S. Department of Labor, Bureau of Labor Statistics, provided the data on the costs of production in the manufacturing sector. The U.S. Department of Commerce, Bureau of Economic Analysis, provided the gross product originating data as well as the deflators used to obtain constant dollar estimates.(77)
Measures of Consumption
Manufacturers use energy sources in two major ways. The first use is to produce heat and power and to generate electricity. The second way in which manufacturers use energy is as a raw material input to the manufacturing process or for some other purpose usually referred to as nonfuel use. Box 6.2 describes the three general measures of energy consumption used by EIA. According to the 1991 MECS, the amount of Total Site Consumption of Energy for All Purposes was 20.3 quadrillion Btu. About two-thirds (13.9 quadrillion Btu) of this was used to produce heat and power and to generate electricity, with about one-third (6.4) quadrillion Btu being consumed as raw material. This does not include byproduct fuels produced from previously counted energy inputs. The Total Inputs measure, which does include byproduct fuels, is most useful in discussions of how energy use in the manufacturing sector compares with energy use in the residential and commercial sectors. It measures only the energy used for its energy content and not as an input into a manufacturing process. Therefore, the Total Inputs measure is the measure that will be used in the development of energy-intensity indicators in this chapter. As in the other end-use sectors, the energy consumption measures used in this analysis do not include the energy lost in the transmission and generation of electricity. These losses are dealt with in Chapter 7, "The U. S. Economy."
An ideal trend analysis would include lengthy historical consumption trends for all three measures for the manufacturing sector. This is not possible since the first comprehensive data series (MECS) was only first fielded in 1985. There is one measure, Offsite-Produced Energy, where trend data can be displayed. The MECS's Offsite-Produced Energy measure can be used to continue the data series, "Purchased Fuels and Electric Energy," which was previously collected for EIA by the Bureau of Census as a supplement to its Annual Survey of Manufactures.(78) Since at least 70 percent of Total Inputs of Energy is made of energy sources produced offsite, the Offsite-Produced Energy measure does provide some insight into changes in energy consumption in the manufacturing sector from 1977 to 1991.
As mentioned previously, the 20 SIC's have been grouped into three groups: High-Energy Consumers, High-Value Added Energy Consumers, and Low-Energy Consumers. Figure 6.5 shows the consumption trends for each of these groups from 1977 to 1991 for Offsite-Produced Energy. The High-Energy Consumers, as a group, reduced energy consumption between 1979 and 1982, but since then and until recently, consumption seemed to have been growing but only at a small rate. The High-Value Added and Low-Energy Consumers seem to have maintained a consistent level of consumption over time. In 1991, for all manufacturing groups, Offsite-Produced Energy represented an insignificant 2 percent of the total production costs faced by U.S. manufacturers.(79)
Although this percentage does increase to 3.5 percent for the High-Energy Consumers, it still is a very small portion of all of the costs of production that manufacturers face. However, in 1991, the costs of energy totaled $32.9 billion (constant 1987 dollars) for the High-Energy Consumers.
As in the past chapters, the most recent trends of energy consumption in manufacturing are placed into two intervals, 1985 to 1988 and 1988 to 1991, the intervals of growth/growth and growth/recession. Using these intervals in the analysis in the remaining portion of this chapter will allow comparisons of Total Inputs of Energy for heat and power and electricity generation (including Offsite-Produced Energy) between the intervals of growth/growth and growth/recession. This will allow the usage of the most comprehensive data available, the MECS, in the actual development and discussion of energy-intensity indicators in the manufacturing sector.
Total Inputs of Energy for heat, power and electricity generation is shown in Figure 6.6 for each of the three manufacturing groups. The High-Energy Consumers consume most of the energy used in the manufacturing sector. All three groups are consistent in that during the interval of growth/growth, their energy consumption increased but, during the growth/recession interval, it declined (Figure 6.6a). The largest percentage decline of Total Inputs of Energy was experienced by the High-Value Added consumers. These consumers' energy consumption declined by 10.4 percent (0.1 quadrillion Btu). During the growth/recession interval, the greatest decrease in energy demand was experienced by firms engaged in producing goods that fluctuate with changes in personal income, such as jewelry, bicycles, computers, apparel, and leather.
During this interval, High-Energy Consumers faced a 0.4 quadrillion Btu decline in consumption (2.4 percent decline). This group, however, did not experience declines in consumption for each member of the group. Paper and Allied Products and Chemical and Allied Products continued to grow after 1988, despite reductions by almost all other industry groups. These two industry groups actually experienced increases in energy consumption during the growth/recession interval.
Chemicals manufacturers exhibited the fastest growth in energy inputs throughout the entire 1985 to 1991. (24 percent) (Table 6.2) Petroleum and Coal Products industries were responsible for almost 30 percent of the 1.4 quadrillion Btu increase in Total Inputs of Energy despite the extent of the consolidation and plant closings that occurred. During the 1980's, mergers and acquisitions contributed to a 15-percent reduction in the total capacity of operable refineries in the United States.(80) Seventeen less sophisticated refiners (i.e., with older, less energy-intensive capital equipment in place) were put out of business in the aftermath of the 1986 oil price crash. Additionally, refiners have increased capacity utilization as plants were upgraded to meet additional environmental regulations governing gasoline and diesel fuel.
Table 6.2. Total Inputs of
Energy for High Consumers in the Manufacturing
In the industrial sector, the diversity of processes and ways in which energy is consumed makes it difficult to single out characteristics that drive energy consumption activities for all industries. At the two-digit SIC level, there are no consistent physical units that can be used to measure demand, e.g., tons could be used in the steel industry, but horsepower may be used in another industry. The demand indicators that are presented in detail in this chapter are considered surrogates for production. The five different dollar-denominated production surrogates considered as potential demand indicators are: gross output, value of shipments, industrial production, value added, and gross product originating (GPO).
All five are presented in constant 1987 dollars to compensate for inflation-induced price fluctuations. However, these measures can still fluctuate due to changes in energy prices, cost of capital, domestic and international taxes, consumer demand, and production cycles. Descriptions of the demand indicators are presented briefly in Box 6.3 and also described in the industrial sector section of the glossary in this report.
The number of establishments, number of workers, and weight of manufactured goods could be considered as potential demand "drivers" or indicators of energy consumption in the manufacturing sector. These three, though, have characteristics that would render them unsuitable for usage in the development by EIA of indicators of energy-intensity in the manufacturing sector. The following describes some of the reasons why each of the three potential demand indicators may not be suitable and thus not presented further in this chapter:
As can be seen in (Figure 6.7a), increases in each of the demand indicators occurred during the growth/growth interval of 1985 to 1988. Earlier, energy consumption also increased during this interval (Figure 6.6).
The demand indicator experiencing the largest increase in the growth/growth interval is the value-added demand indicator (34.3 percent). This indicator also is the only demand indicator that did not experience an overall reduction during the growth/recession interval of 1988 to 1991 (Figure 6.7b).
During the growth/recession interval, the growth in the various demand indicators for High-Energy Consumers was either minimal or negative (Figure 6.8a). With the exception of GPO, most of the slow or negative growth during the growth/recession interval can be attributed to the Stone, Clay and Glass Products and the Primary Metals industries. However, the Primary Metals industries posted an 8.7- percent increase in 1988 to 1991 in GPO. The Primary Metals industries also posted the largest increase in value added during the growth/growth interval, 1985 to 1988.
The High-Value Added Energy Consumers accounted for 47 percent of value added in 1985. The share of value added for this group decreased to 41 percent by 1991 because industrial machinery and electronic products both lost out to cost-competitive imported products.
The High-Value Added Energy Consumers also were the largest contributors to industrial production and GPO, accounting for 45 percent and 47 percent, respectively, in 1991. With the exception of value added, all of the other demand indicators, as shown in (Figure 6.8b), decreased during the growth/recession interval.
The Low-Energy Consumers faced reductions in each of the demand indicators, the exception being value added (Figure 6.8c). These reductions seem to be greater than reductions in the other two groups. Most of the industry groups within the Low-Energy Consumers seemed to have faired poorly during the growth/recession interval. One apparent exception was the Rubber and Miscellaneous Plastics industry, which posted increases in all of the demand indicators.
As is demonstrated in the previous discussion, the economic environment may have major impacts on the levels of the demand indicators that drive energy consumption. These effects are, however, structural effects that need to be considered when measuring changes in energy-intensity indicators in the manufacturing sector.
The key drawback for the shipments-based demand indicator that it measures the product shipped from industries, whether manufactured this year or taken from inventory. Stock changes can obscure production activity in the manufacturing sector. Industries that build up stocks will underestimate actual production. Similarly, for industries depleting their stocks, actual production will be overestimated. All manufacturers reduced inventory during the growth/growth interval and built up some inventories during the growth/recession interval (Figure 6.9). These movements in inventory need to be considered when choosing a "best" indicator of energy-intensity in the manufacturing sector.
The best demand indicator that considers inventory changes is gross output. The problem is that, with the exception of value of shipments demand indicator, gross output and the other demand indicators presented in this chapter have not been corrected for the SIC reclassification in 1987 from a 1972 base. Using these as demand indicators will not provide as accurate a trend analysis for the years before 1987. Value of shipments data in MECS have been corrected to SIC 1987 basis in all years, adjusted by MECS weights and then deflated to constant 1987 dollar.
Although the value of shipments demand indicator does not take inventory changes into account, it can be revised to reflect these changes.(82) Inventory-adjusted value of shipments, also called value of production estimates, act much like weather adjusted estimates in the residential and commercial sectors. If stocks are drawn down during the base year (1985), then the value of production will be less than the value of shipments for that year.
During the growth/growth interval, as inventories were being drawn down, the percentage increase in the value of shipments was more than the percentage increase in the value of production, which reflects the levels of production more accurately for all three energy-consuming groups (Figure 6.10a). Without inventory adjustments, decreases in energy-intensity indicators and potential increases in energy efficiency during these years might be overestimated.
During the growth/recession interval, inventories increased. The value of shipments was less than the value of production, reflecting a lower level of production than actually occurred (Figure 6.10b). Although some of the value of the production did not leave the establishments as value of shipments, the actual production of this inventory did take place and energy was used in the production process. Decreases in energy-intensity indicators and potential increases in energy efficiency might be underestimated if inventory changes are not considered.
Production levels vary in accordance with capacity levels and utilization rates. In some cases, manufacturers predetermine the utilization rate to maximize profits or minimize operating losses. The utilization rate takes into account scheduled maintenance on plant and facilities, vacation plans, and investment in new capital equipment. Other events, such as unscheduled outages, labor strikes or slowdowns, and materials supply bottlenecks, may alter utilization rates. If an establishment is not operating at full capacity, energy may still be used but this energy should not be considered when attempting to measure changes in energy-intensity. However, at the other end, if an establishment is operating at or near full capacity, it may be using all of the equipment it can, including some of the old, inefficient equipment.
Accounting for changes in both inventory and capacity utilization yields a capacity-adjusted value of production demand indicator. It is derived by comparing the rate of capacity utilization reported by the Federal Reserve Board for each year and the 26-year average (1967-1993) for all major industry groups.(83) Not surprisingly, for most of the major industry groups, the average capacity utilization rate is higher than the reported annual rate in 1985 and 1991 (when inventories were being drawn down) and the average capacity utilization rate was below the reported annual rate in 1988 (when stocks were built). This adjustment smoothes the value of production estimates by raising the output in years of low capacity utilization (e.g., 1985 and 1991) and lowering output in years of high capacity utilization (e.g., 1988). If this adjustment is not applied to years of low capacity utilization along with little reduction in the amount of energy consumed for heat and power and electricity generation, any measure of energy-intensity would cause its corresponding measurement, energy efficiency, to be underestimated. Figure 6.10 shows the resulting percent changes in capacity-adjusted value of production for all the consuming groups for both the growth/growth and growth/recession intervals. These are smaller than the percent changes in the value of production for both periods. These reflect the adjustments on the value of production estimates upward in 1985 and 1991 where capacity utilization rates were lower than the 26-year average and adjustments downward in 1988 when actual capacity utilization rates were higher than the 26-year average.
Energy consumption and the "drivers" of energy consumption, the demand indicators for the manufacturing sector, have been discussed in detail. The next step is to construct measures of energy-intensity using the two components. The ideal would be to be able to measure energy efficiency in its purest form. The best that can be accomplished is to measure changes in energy-intensity, adjust where adjustments are possible, and be aware that although reductions in energy-intensity may reflect increases in energy efficiency, these reflections also contain some structural and behavioral influences. A general indicator of energy-intensity used in the manufacturing sector is energy (thousand Btu) per demand indicator (in 1987 dollars). Box 6.4 lists the various potential indicators of energy-intensity in the manufacturing sector. There are choices, though, as to which indicator of energy-intensity to use. Table 6.3 shows the values of the different choices for energy-intensity indicators. Over all the indicators of intensity, the High-Energy Consumers are not only high users of energy, they are the most intensive per dollar of output. One exception is the Food and Kindred Products industry group. This industry group is a high user of energy, but not a very intensive user of energy.
The energy-intensity indicators using gross output, value of shipments, value of production, and adjusted value of production are comparable while the remaining three indicators measures are roughly twice the value in thousand Btu per dollar. The leading difference between gross output and value of shipments is that gross output considers inventory changes while value of shipments does not, implying that gross output and value of production are very comparable. One can see the difficulties in selecting an energy-intensity measure when measuring the impacts of a particular energy program.
Table 6.3. Energy-Intensity Indicators in the Manufacturing Sector, 1985, 1988, and 1991
Figure 6.11 adds to the choice dilemma when looking at the changes in the various energy-intensity indicators for all manufacturing and High-Energy Consumers. The outlier by far is value added. The problem with using an energy-intensity indicator based on value added, is that the changes could be a reflection that U.S. manufacturing is responsible for less finished goods production as final assembly and processing occurs out of the country. One also has to remember that only the energy-intensity indicators using value of shipments has been standardized to the 1987 SIC classification.
Figure 6.11a shows the changes in the energy-intensity indicators over the growth/growth interval. There seem to be conflicting messages, depending on which energy-intensity indicator is used. It does seem though, with the clear exception of the intensity indicator using value added, that there have been very slight changes in energy-intensity and thus energy-efficiency changes seem to have been minimal for both the manufacturing sector and the high consumers of energy. When looking at the interval of growth/recession as shown in Figure 6.11b, the intensity picture changes. It seems that for the most part, energy indicators fell implying possible increases in energy efficiency. Again, the reader must be aware that structural change could be the influential factor. A huge price increase or other market fluctuation could augment a measure such as the value of shipments and make energy-intensity appear reduced. Adjusting for capacity utilization helps strip away certain economic effects; however, a dollar-denominated energy-intensity indicator will always be susceptible to such market changes.
In addition to the effects economic influences may have on the measurement of energy-intensity, changes in the relative market share of major industry groups (in terms of constant dollar value added, value of shipments, etc.) can be used to indicate changes in industry structure. EIA has developed a method for examining structural changes by holding the relative market share constant (percent share of the value of shipments within each two-digit SIC group) and revising four-digit SIC industry data in order to build up a structurally-adjusted intensity measure at the two-digit SIC level.(84)
Examination of the components affecting intensity measures offers some insight into the structural changes of U.S. manufacturing and their impact on the efficiency of energy use.
If five of the High-Energy Consumers(85) are the most energy intensive per value of shipments or value of production, then reducing their share of demand could contribute to an overall reduction in energy-intensity for U.S. manufacturing. Likewise, the less energy-intensive categories--High-Value Added and Low-Energy Consumers--could increase their share of shipments or production and contribute to reduced energy-intensity for the manufacturing sector as a whole. For further discussion and detailed analysis using this methodology, see EIA's publication Changes in Energy-Intensity in the Manufacturing Sector 1985 - 1991 (DOE/EIA-0552(85-91)).
In this chapter, energy-intensive indicators were developed only for the manufacturing sector. Most of the energy used in the industrial sector is used in the manufacturing sector. This does not lessen the importance attached to the development of nonmanufacturing energy-intensive indicators. Future work needs to incorporate the development of these indicators as more data become available or new methodologies are developed that will allow the use of limited nonmanufacturing data.
The energy-intensity indicators developed in this chapter are varied. Each have certain strengths and limitations or weaknesses. The demand indicators used in the energy-intensity indicators are compared in Table 6.4.
These are all dollar-denominated surrogates for actual output and as such, a huge price change or other market fluctuation could augment or distort the value of shipments and make energy-intensity indicators appear higher or lower than the actual change in intensity. Such distortion may cause underestimation and overestimation of energy-efficiency improvements.
Table 6.4. Strengths and Limitations of Demand Indicators Used in the Energy-Intensity Indicators
73As an interesting note, both primary and site energy were presented in Figure 6.1 to show the effects of a change in the energy mix over time. Around 1982, a gap develops between primary and site consumption, reflecting a change in the energy mix to the use of more electricity with its inherent losses that occur in the generation and transmission of electricity.
74Agriculture, mining, and construction are represented in the Census Bureau's set of quinquennial economic censuses which provide data on expenditures for purchased energy. The 1992 Census reports are not available at the present time. In addition, The Census of Mineral Industries collects and publishes data on consumption of purchased energy and consumption of onsite-produced energy. Also, the Annual Farm Costs and Returns Survey, conducted by the U.S. Department of Agriculture, is a source of annual data on energy expenditures and respondent-estimated prices for certain fuels. These data could only be used for rudimentary analysis.
75 The Office of Management and Budget derived this hierarchical system. Wherever possible, data presented correspond to the 1987 SIC reclassification. As a result, 1985 and 1988 data may not be comparable with previously published data.
78Since the MECS does not collect annual data, EIA developed a method to derive estimates of offsite-produced energy for the missing years 1982-84, 1986-87, and 1989-1990. This methodology is developed in Derived Annual Estimates of Manufacturing Energy Consumption 1974-1988 (DOE/EIA-0555(92)/3).
79See U.S. Department of Labor, Bureau of Labor Statistics Multifactor Productivity in U.S. Manufacturing, 1949-1991. Data points in 1985 are based on 1972 SIC; 1988 and 1991 data points are based on 1987 SIC.
File Last Modified: October 17, 1999