Recent developments in industrial capacity and utilization.

Recent Developments in Industrial Capacity and Utilization

Richard D. Raddock, of the Board's Industrial Output Section in the Division of Research and Statistics, prepared this article. The author and Charles E. Gilbert, assisted by Tory M. Wolff, are responsible for the revised estimates.

The Board of Governors of the Federal Reserve System has revised and restructured its estimates of industrial capacity and capacity utilization from January 1967 to the present. Like the earlier estimates, the revised ones show that utilization peaked in late 1988--early 1989 and that pressures on capacity have diminished since then because output has leveled off while capacity has continued to expand.

The new estimates, made in conjunction with a revision of the index of industrial production, cover manufacturing, mining, and utilities in greater industry detail than before. The separate set of estimates for industrial materials has been eliminated, and many of its former components have been included in the new estimates for mining and the primary-processing component of manufacturing.

Each utilization rate is the ratio of a seasonally adjusted industrial production (output) index to a related capacity index (chart 1). The capacity indexes are developed from a variety of capacity, utilization, and related data and, in regard to coverage, weights, and other aspects, are designed specifically to be used with the industrial production indexes. Indeed, when production indexes are substantially revised or rebased, the related capacity index must be revised or rebased as well so that appropriate utilization rates can be calculated.(1) Both the capacity and production indexes are now expressed as percentages of output in 1987.

The capacity index for an individual industry represents a realistically sustainable maximum level of output for that industry, rather than some higher, unsustainable, short-term maximum. Thus, utilization rates of 100 percent have been reached or even temporarily exceeded in specific industries, particularly in materials industries. The assumption for an individual establishment is that plant and equipment are fixed but that labor and materials are available. At the macro level, however, all industries do not simultaneously operate their fixed plant and equipment (capital) at 100 percent of capacity because peak demands for all goods are not synchronous and because shortages of skilled labor and materials create bottlenecks. Before 1967, the overall manufacturing rate slightly exceeded 90 percent at peaks during the Korean and Vietnam wars; since 1967, the utilization for total industry has remained under 90 percent.

THE NEW SERIES STRUCTURE

Central to the revision was a restructuring of the capacity estimates for industries within manufacturing. Previously, manufacturing capacity had been estimated largely at the level of major two-digit industry groups of the Standard Industrial Classification (SIC). Thus, for example, only one capacity series existed for non-electrical machinery (SIC 35), which is a large and heterogeneous industry group. In this revision, SIC 35 comprises three series. Total manufacturing, which previously comprised twenty-four individual series, now has fifty-four (table A.1).

The greater number of series improves the estimation and extrapolation of the capacity indexes in two major ways. First, the nonelectrical machinery group and several large two-digit industry groups have been split into component series that display different cyclical patterns and growth trends of output. Splitting the major groups in this way allows the capture of more information for deriving capacity estimates. Second, the detailed physical volume measures of capacity previously used to estimate a number of major materials have now been fully and explicitly incorporated in the new manufacturing estimates. This incorporation of measurable physical data improves the quality of the estimates.

The disaggregation of the manufacturing estimates marks another step in their evolution, which began in the 1960s with only two manufacturing components--primary and advanced processing. In the 1970s, manufacturing was expanded to fifteen components that conformed to the industry detail on capacity utilization contained in the McGraw-Hill Spring Survey of Business Plans for New Plants and Equipment.(2) Additional estimates at the two-digit SIC level of detail, based on the Census Survey of Plant Capacity, were added in 1985. In their present expanded form, the manufacturing estimates depend on both broad surveys of manufacturing utilization rates and on narrower surveys for particular industries conducted by trade associations and governmental agencies.(3) Particular care has been taken to ensure consistency over time when new data sources have been added so that the estimated utilization rates will indicate the relative pressure on capacity over a cycle and among cycles appropriately.

Besides the individual materials series that are now in manufacturing, sixteen former materials series are incorporated in the category of mining. Because manufacturing and mining now include so many of the materials series, the Federal Reserve determined that the separate and independently estimated set of materials measures could be eliminated without significant loss of information. The new structure integrates the detailed information on the capacity to produce materials into the measures for total industry in a consistent manner that was previously lacking.

Whereas the general public has focused on the overall utilization rate, specialists have often focused on those primary-processing or materials-producing industries for which physical quantity measures of capacity, output, and utilization are available, including raw steel, aluminum, paper and pulp, plastics resins, and others. For many of these industries, high or rising rates of utilization during an expansion, such as in 1988, have been associated with notable increases in prices or profit margins.(4) Users of the new Federal Reserve Statistical Release G.17 (419) "Industrial Production and Capacity Utilization," will find explicit representation of more of these industries in the tables on capacity and utilization.

REVISIONS OF UTILIZATION AND CAPACITY

Overall, the revisions of utilization and of growth of industrial capacity are relatively small and, except for showing a shallower recession in 1982, do not change the basic cyclical story indicated by the previous data (chart 2). On average over 1967-89, the operating rate has been revised up 1/2 percentage point to 82.2 percent. The recent cyclical high of 85 percent in April 1989 also is higher than previously shown, but it remains noticeably below the upward-revised peaks of the 1970s--89 percent in the fall of 1973 and 87 percent in early 1979 (table 1). After the April 1989 peak, the new production and utilization figures weaken a bit more than the old estimates; nevertheless, the revised utilization rate for all industry in late 1989 and early 1990 averages about 83.2 percent, 1/2 percentage point higher than was previously shown and still about a percentage point above its historical average.

For recent years, the new utilization estimates for both mines and utilities are higher than the old estimates but remain well below historical highs. The new manufacturing rate, on average, is a little higher as well. During 1989, however, the manufacturing rate now appears to have declined a bit more from a slightly higher peak. As a result, by late 1989--early 1990, the new operating rate for manufacturing, like the old rate, was nearly 82 1/2 percent. The main difference in cyclical patterns in manufacturing occurs at the 1982 low, now 70 percent compared with 68 percent previously shown. The smaller decline in the revised index of manufacturing output during that recession largely accounts for this change.

Growth in industrial capacity averaged 3 percent a year from 1967 to 1989, about the same as previously shown, with growth having slowed progressively after the energy crises and recessions of the early 1970s and 1980s from more than 4 percent a year in the late 1960s to only about 2 percent in 1986 and 1987 (table 2). After 1987, yearly capacity growth accelerated, to about 2 1/2 percent recently--less of an acceleration than the old estimates showed. A larger decline in the capacity to extract crude oil, a marked slowing (albeit from high rates) in growth in some high-technology industries, and a smaller rebound in capacity growth in primary-processing industries within manufacturing largely account for the more modest response to the recent high utilization rates.

Capital Stock and Capacity

Investment spending and related capital stock estimates are sometimes used to make inferences about capacity trends. Estimates of capital stock by the Bureau of Economic Analysis (BEA) for total manufacturing, mining, and electric and gas utilities show even more slowing during the 1980s than do the capacity estimates, in part because of the capital intensity of the declining mining industry (chart 3).

Capital stock is only one of the factors that determines capacity output, however, and the relationship between capital input and capacity output is variable over time and across industries. New technology and capital expenditures to promote a safer and cleaner environment, the closing of plants and the restructuring of industries in the 1980s, the lengthening workweek of capital, and the shift in the composition of capital to shorter-lived equipment have caused the growth rates of capital stock and the Federal Reserve measures of capacity output to differ.(5)

For manufacturing, capital stock grew more rapidly than capacity from 1967 to 1980 and more slowly after 1982. In some industries, the growth of capital stock was much slower than that of capacity during the 1980s. In durable manufactures as a whole, especially for nonelectrical machinery, the capacity and production trends reflected the strong gains in computers. In other industries, such as lumber and products, apparel, and textiles (in which the new looms and spindles are much more productive than the old), capital stock declined while production and capacity rose. For tobacco products and fabricated metal products, however, capital stock has grown much faster than capacity or production. In contrast, the capital stock declined less than did capacity for petroleum refining and primary metals in the 1980s. In these industries, the capacity figures incorporate available data on permanent plant closings, whereas the capital stock figures incorporate a formula for discards that does not allow for the bunching of closings that occurred.

Because of the important theoretical and empirical differences between capacity output and capital input, the Federal Reserve uses the capital stock figures only to refine year-to-year movements in its capacity estimates and only when physical capacity data have not been compiled. In the absence of such capacity data, it seems reasonable to assume that capacity growth in an industry is positively correlated with investment spending and growth in the capital stock in that industry. The over-all levels and the long-term movements of the Federal Reserve capacity indexes, however, are determined primarily by production indexes divided by utilization rate data as reported in various surveys.

REVIEW OF DEVELOPMENTS BY SECTOR

In late 1988 and the first half of 1989, the nation's factories, mines, and utilities together operated above 84 1/2 percent of capacity, the highest rate since the start of the decade. Although the pressure on industrial capacity has lessened during the past year because of the stagnation in industrial production, utilization remains above the relatively low rates that prevailed from mid-1980 until early 1987. An acceleration of economic expansion after 1986 and a low rate of growth of capacity contributed to the higher utilization rates.

For much of the time between 1980 and 1986, however, an extended double-troughed recession and adverse international terms of trade yielded severe financial stress and restructuring in key segments of American industry. Manufacturing utilization averaged only 77 1/2 percent from 1981 to 1986, a low level for so long a period. Overall industrial capacity grew slowly, many mines and factories closed, and capacity in some basic industries shrank sharply.

The 45 percent depreciation of the dollar during 1985-87 helped restore the competitiveness of U.S. products, and the growth of production picked up in 1987-88. Utilization rates rose, and reports of shortages of metals, chemicals, and other products of primary-processing industries began to appear. As a result, prices, profits, and investment in capacity in these industries increased.

In 1989 and early 1990, the level of industrial investment, an important determinant of capacity growth, generally rose, even though manufacturing output plateaued. Lower sales and production of motor vehicles contributed to lower production of steel and other related materials. This weakness was aggravated at times by limited inventory corrections by producers and distributors, who kept their stocks in line with less robust sales. Although many commodity prices fell back, continued growth of U.S. exports to a generally strong world economy supported sales and production levels in several industries. As a result, by the first quarter of 1990, the rate of capacity utilization in manufacturing had fallen only a couple of percentage points from its high a year earlier; and at present, the moderate pace of industry capacity expansion appears to be continuing.

Mining

Capacity utilization in mining rose in 1988-89 to a high of 87 percent, a moderate level of utilization for this series (chart 4). It was held down by a low level of oil well drilling. Mining capacity, which had dropped 6 percent from 1984 to 1987, declined another 6 1/2 percent in 1988 and 1989, despite a rebound in metal mining capacity.

Oil and gas extraction, which dominates mining production, is in a lengthy decline. Domestic capacity to drill oil wells and extract crude oil from those wells is declining even though the price of oil has recovered to about $20 per barrel. According to a survey by the Reed Tool Company, the available rotary drilling rigs in the United States dropped 7.6 percent in 1989 to 2,542 rigs in continuation of a decline from a peak of 5,644 rigs in 1982; even so, only 57 percent of the available rigs were reported to be actively drilling within a month of the survey. This continued low utilization of the rig fleet points to a further reduction in the available fleet in 1990 and to a continued decline in the U.S. capacity to pump crude oil. Such capacity is declining not only in the older oil fields of Texas and Louisiana but also in the newer fields of Alaska.

According to the American Petroleum Institute, most of the record 6.8 percent decline in 1989 crude oil production was not caused by the Exxon Valdez oil spill or any other temporary disruptions. The institute indicates that the underlying annual rate of decline in production may have reached 100,000 barrels per day (bpd) in Alaska and more than 300,000 bpd in the other forty-eight continental states, compared with national production of 7,600,000 bpd in 1989.

Most analysts agree that the production of domestic crude oil is close to effective capacity and that its downtrend is likely to continue. With demand growing, imports jumped 8 percent in 1989 and supplied 46 percent of domestic deliveries. Also, the American Gas Association expects the excess natural gas production capability--the gas "bubble"--essentially to disappear this year as demand continues to expand. The association estimates that the new gas wells will again fail to offset the declining production capability of the old wells, although the proportion of drilling activity directed toward gas wells, rather than oil wells, has increased.

Mining capacity outside the oil and gas industry has been increasing. In particular, the former decline in metal mining capacity reversed in 1987-89 as strong worldwide demand for metals led to the reopening of mines such as the huge Bingham Canyon, Utah, copper-silver mine, which had closed for extensive modernization during the period of weak demand. Shuttered iron mines were purchased from bankrupt companies and restarted. The large Red Dog zinc mine in Northern Alaska opened in late 1989 and may double U.S. zinc mining capacity by 1991. Gold production at new and expanded mines has soared as the mining technique of heap leaching reportedly has cut costs at some domestic sites to less than $200 an ounce, far below the recent market price of gold.

Manufacturing

The manufacturing sector--85 percent of total industrial production--covers a wide range of industries that have significantly different patterns of utilization and capacity growth. For analytical purposes, the Federal Reserve has divided manufacturing into two broad groups: primary-processing industries, which produce mostly materials and supplies, and advanced-processing industries, which produce mostly finished consumer or capital goods. Primary-processing industries. These industries include textile mill products; paper and products; industrial chemicals; petroleum products; rubber and plastics products; lumber and products; primary metals; fabricated metal products; and stone, clay, glass, and concrete products.

High levels of capacity utilization in these basic manufacturing industries often occur late in cyclical expansions as inventory accumulation and rising demand for consumer and capital goods converge on materials industries. This pattern appeared in late 1987, when rapid growth in production raised the utilization rate of primary-processing industries above 87 percent, its highest level since 1979. Robust production growth continued to outpace the 2 percent growth in capacity in 1988, and utilization in primary processing peaked in January 1989 at 89 percent, about as tight as it was when it peaked in 1978-79 but less tight than it was in 1973, when an extraordinary range of materials was reported in short supply (chart 5). From late 1987 to early 1989, the pressure of rising production on capacity in primary metals, petroleum refining and industrial chemicals, paper, textile, and lumber industries contributed to rising commodity prices and profit margins and to an increase in capital expenditures and the rate of capacity growth (chart 6). Since early 1989, utilization has declined, and the prices of many basic commodities have fallen--some, like aluminum ingot, rather sharply.

Data based on physical volumes for output and capacity are available for many primary-processing industries; and the quality of the data, the strategic importance of the industries, and the continuous operation (twenty-four hours a day, seven days a week) of several industries give them analytical interest greater than their weight in total production. In the revision, a particular effort was made to incorporate these data explicitly in the manufacturing estimates. The review of industry that follows discusses these data.

In 1988 and early 1989, after a wrenching restructuring during the 1980s, the smaller, more efficient primary metals industry operated near 90 percent of its capacity and prospered more than it had for a decade (chart 5). Several factors had caused the restructuring: Besides the severe recessions and the overvalued dollar, which had affected so many domestic industries earlier in the decade, the substitution of plastics, energy conservation with its emphasis on lighter vehicles, a lower investment in heavy industry, the development of lighter, stronger steels, and so on, reduced the need for tonnages of metal. Cutbacks and modernization eliminated a substantial amount of outmoded capacity and cut costs. By 1989, output had risen strongly from low levels; however, it was still nearly 20 percent below its level a decade earlier, and employment remained down nearly 40 percent.

Utilization in the steel industry, which had languished below 70 percent of capacity for 5 1/2 consecutive years, soared in 1987 and held near 90 percent during 1988 and early 1989. Some analysts said that this rate was close to effective capacity. Purchasing managers reported that steel was short in supply and higher in price; by the summer of 1988, the producer price index for iron and steel had risen 13 percent from a year earlier.

Although increases in production were a major factor behind the rise in utilization, the level of production remained far below the highs of a decade earlier. In 1989, for example, raw steel production of 97 1/2 million tons was nearly 30 percent below its level in 1979. Steelmakers, however, had slashed their capability to produce raw steel 28 percent, from 155 million tons to 112 million tons, between 1979 and 1987. Since then, the reopening of one "permanently closed" mill and numerous small improvements have added back 4 million tons of capacity.

By the first quarter of 1990, declining iron and steel production had pushed the operating rate down to 81 1/2 percent, about 8 percentage points from its recent peak. Little change is expected in raw steel capacity; nevertheless, ongoing investment in continuous casters and other new technology will gradually increase the yield of steel mill products.(6)

Among producers of nonferrous metals, primarily those of aluminum and copper, rising production and lower capacity similarly led to sharp increases in utilization and to rising prices in 1987 and 1988. The producer price index for primary nonferrous metals in the first half of 1988 was up about 37 percent from a year earlier. Over the twelve months ending in February 1990, however, as production and utilization eased, the same price index declined more than 20 percent.

The primary aluminum industry was able to operate at or above rated capacity of 4 million metric tons for much of 1988 and 1989 because the maintenance of potlines was delayed. But just as it had in steel, a significant restructuring had occurred earlier in the 1980s, and output in 1988-89 remained well below earlier peaks. Low aluminum prices and the higher cost of electricity, which accounts for as much as 35 percent of the product cost, aggravated the problems of low sales volume, excess inventory, and overcapacity. As a result, the industry renegotiated labor contracts and permanently closed eleven of thirty-two primary aluminum plants to cut costs. Producers cut capacity to produce aluminum ingot from 5 million tons in 1983 to 4 million tons in 1986, a year when annual production dropped to only 3 million tons, a third below the 1979-81 level. Capacity has changed little since 1986, and no new smelters are planned.

In late 1989, utilization in petroleum refining recovered to nearly 90 percent of capacity, a rate that before the energy crisis had been typical of this industry but that had not been seen for many years.

The soaring price of crude oil, which tripled to $35 a barrel over the three years ending in 1981, forced a sharp reduction in the use of gasoline and other petroleum products. Production plunged. Even though oil prices receded over the next few years, they remained high through 1985, and demand recovered only a little. As a result, in 1985 production of petroleum products remained 15 percent below its level in 1978, and gross inputs to domestic refiners' distillation units were down nearly 20 percent. By early 1986, more than 100 refineries were closed permanently, and the distillation capacity of operable refineries was slashed by 17 percent to 15 1/2 million barrels per day.

Since then, distillation capacity has not increased much, and little change is expected in 1990. Supplies of high-octane, lead-free gasoline and of heating oil have occasionally been tight. As a result, investment in refining has been directed toward downstream processing units that upgrade the mix of products to fit demand, rather than toward an increase in the total volume of crude input that can be processed.

Like many of the materials-producing industries, industrial and synthetic chemicals manufacturing maintained high utilization rates in 1988 and peaked in late 1988-early 1989. During this period, purchasing managers reported that ethylene and its derivatives, plastics resins and products, synthetic fibers and rubber, benzene, and caustic soda and soda ash were in short supply. Prices and profit margins for many chemicals rose rapidly, and numerous reports appeared about plant reopenings and capacity expansions based on "debottlenecking," technological improvements, new equipment, and increased investment in new plants, especially in the petrochemical industry around Houston, Texas.

Projects extending to 1993 have been announced for plastics and their basic organic chemical feedstocks, ethylene and propylene. By mid-1989, fifteen ethylene producers had announced planned capacity increases totaling almost 14 billion pounds; these projects, if fully implemented, will raise ethylene capacity more than a third by 1993 and may lead to excess capacity. Similarly, capacity to produce plastics resins expanded 5 1/2 percent in 1987 and almost 10 percent in 1988, after relatively slow growth in 1982-86. Numerous projects in various stages of completion indicate that the expansion of plastics resins capacity has continued.

The capacity to produce other synthetic chemicals has grown less rapidly. Fiber Organon, in its November 1989 survey of capacity, reported that operable capacity of facilities producing manufactured fiber (cellulosic and synthetic), which declined nearly a billion pounds (9 percent) from 1979 to 1986, is expected to increase 5 1/2 percent in 1990 to 11 billion pounds, about the same level that it was in 1979.

The basic pulp, paper, and paperboard industry operated essentially at 100 percent of capacity in late 1987 and early 1988 after two years of substantial growth in output. The producer price index for paper rose at an 11 percent annual rate from mid-1987 to the end of 1988. Since then, output has flattened while capacity has grown at about a 3 percent rate, and utilization and price have declined although the industry continues to operate at a relatively high rate.

This industry, which had increased capacity at a 2 percent annual rate from 1982 to 1987, stepped up its rate of expansion and now plans to increase paper and paperboard capacity 8 million tons to more than 90 million tons during 1989-91. These figures translate into a 3.2 percent annual rate of growth, which is about 1/2 percentage point above the production trend. The industry rapidly increased its capital expenditures in recent years, almost entirely for new equipment that increases capacity and enables the industry to adjust to changing demands for different types of paper, to meet environmental requirements, to conserve energy, and to cut costs. Advanced-processing industries. The advanced-processing sector--about two-thirds of manufacturing--comprises most manufacturers of finished consumer goods and equipment. It includes food and kindred products, tobacco products, apparel products, printing and publishing, chemical products, such as drugs and toiletries, leather and products, furniture and fixtures, electrical and nonelectrical machinery, transportation equipment, instruments, miscellaneous manufacturers, and government-owned, government-operated ordnance facilities.

Among these industries are most of the fast growers: computers, communications equipment, semiconductors, instruments, and printing and publishing. Consequently, even though the utilization rate has not reached an exceptionally high level in recent years, output and capacity growth in advanced processing have exceeded output and capacity growth in primary processing (chart 7).

From 1977 to 1987, annual capacity growth in advanced-processing industries averaged 4 1/2 percent, but since then it has decelerated to about 3 1/4 percent. The slowing is evident particularly in high-technology machinery and in transportation equipment. For instruments, furniture, printing, miscellaneous manufacturing, and drugs and other chemical products, the growth of capacity has accelerated.

Utilization in advanced processing reached a peak in early 1989. However, its peak was only 83.4 percent, about 3 percentage points below the cyclical highs of the 1970s because the rates for machinery, instruments, and motor vehicles remained well below their earlier highs. Since then, utilization in advanced processing has fallen and, at the start of 1990, was not far from its 1967-89 average.

Within advanced processing, only the motor vehicles industry provides detailed physical data pertaining to capacity.(7) This industry has not been a fast grower. Indeed, closings of automobile and truck assembly plants have been widely reported for a decade, and some localities have been hard hit. However, the opening of new assembly plants, often by foreign companies or joint ventures, has largely offset the effect of the closings on national production capacity. Demand continues to be insufficient to utilize fully the capacity to assemble cars and trucks.

The first wave of plant closings occurred in the early 1980s, when sales plunged in the wake of soaring gasoline prices and the recessions of 1980 and 1982. Consequently, between 1979 and 1982, the industry slashed domestic capacity to assemble autos and light trucks from 14 1/2 million to 12 3/4 million units, with especially sharp percentage cuts at Ford and Chrysler (chart 8, left panel). From 1982 to 1987, however, assembly capacity rebounded to 14 3/4 million units, with a strong gain in General Motors's light truck capacity and the buildup of capacity by the first Japanese transplants: Honda, Nissan, and Nummi Motors (the General Motors-Toyota joint venture). During this period, General Motors also built new automobile plants to replace older facilities.

Since 1987, a second wave of closings has occurred. Chrysler has shut the large plant in Kenosha, Wisconsin (formerly owned by American Motors), and General Motors has permanently closed capacity to produce 900,000 vehicles and indefinitely idled plants that can produce another 700,000 units.

On balance since 1979, domestic assembly capacity, including the expanding transplants, has dropped only about half a million units (1 1/4 million if the indefinitely idled General Motors plants are included) to approximately 14 million units. Moreover, the big three automakers and the Japanese transplants have expanded capacity in Canada and Mexico enough so that overall North American capacity to produce motor vehicles for the U.S. market has increased. In 1989, however, U.S. output was only 10.6 million units, well below earlier highs; as a result, utilization of domestic capacity was about 75 percent, with a lower rate for automobiles than that for light trucks (including vans), which have grown in popularity. With sales sluggish and transplants scheduled to increase North American capacity significantly over the next few years, excess capacity threatens a further shakeout of automobile plants in the United States.

The production of medium and heavy trucks declined even more severely in the early 1980s than did the output of lighter vehicles. Producers cut capacity one-fourth by 1984 and have not added to capacity since then. Even so, utilization of heavy truck capacity remains relatively low because production has remained a third or more below its 1979 peak. In 1989, production eased to 250,000 units, 40 percent below its peak a decade before.

Utilities

Consumers' reactions to soaring energy prices in the 1970s and early 1980s caused the slower growth and the low operating rate (compared with its own past peak) that characterized the utilities for more than a decade. Capacity growth averaged only 1.4 percent per year from 1977 to 1989. Demand perked up only after 1986, when crude oil prices fell back. In late 1989, utilization jumped to 87 1/2 percent, its highest level in more than a decade.

The operating rate of electric utilities has been relatively low, and the growth of electric generating capacity has been slowing since late 1973, when fuel prices soared and users began to conserve energy (chart 9). From 1967 to 1974, generating capacity increased about 8 percent per year; from 1976 to 1978, capacity growth slowed to a 4 1/2 percent rate, and since then, it has averaged about 2 percent a year. Utilization has risen since its low in 1980 but remains well below the rates that prevailed before 1973.(8) According to the North American Electric Reliability Council, the nation's utilities intend to increase capacity at an annual rate of 1.2 percent between 1989 and 1998 compared with a projected annual growth in summer peak demand of 2 percent. The more rapid rise in demand is expected to reduce the margin of excess capacity about 5 percentage points.

Although growth in demand for electricity has slowed, the level of generation was 50 percent higher in 1989 than it was in 1973. In contrast, gas utilities had to cut production significantly until a low was reached in 1986. Despite increases since then, capacity utilization probably remains low. Data on underground storage of natural gas, a factor in deliverability at times of peak demand, and on withdrawals and reserves of natural gas indicate little change in capacity in recent years. Explicit capacity data for the gas utilities are not available.

CONCLUSION

Economic expansion combined with a restructuring of some basic industries during the 1980s led to higher utilization rates and a marked rise in commodity prices in 1988 and early 1989. In the slowdown that has occurred since then, utilization in some basic industries has declined, and commodity prices have eased. Nevertheless, as projects undertaken in the past year or two develop, the rate of growth of industrial capacity should be maintained in the near future in the same 2 to 3 percent range that has characterized the past two years. Although the growth of investment has slowed, its level should be high enough to expand capacity.

APPENDIX

This appendix discusses (1) the evolution of the Federal Reserve Board's capacity and utilization estimates, (2) the six basic steps in the derivation of the estimates, (3) the new structure as shown in table A.1, (4) the effect of changing weights and linking indexes on utilization, and (5) concepts and levels of capacity. At the end of the appendix are historical data, 1967-89, for total industry (table A.2) and formulas for aggregating capacity and calculating utilization.

Background

From its earliest days, the Federal Reserve has been interested in timely measures of economic activity. In 1919, it began to publish monthly indexes of the "physical volume" of domestic trade, which in the 1920s evolved into the monthly index of industrial production. Later, it began to study data on capacity and capacity utilization in conjunction with production to analyze inflationary pressures and the demand for capital goods.

In the booming 1960s, interest in capacity contraints was great, and the Federal Reserve maintained estimates of output, capacity, and utilization for selected industrial materials and for manufacturing. For the most part, these estimates were used internally until 1966, when the U.S. Council of Economic Advisers published the manufacturing estimates in its Annual Report. Later in the year, Frank de Leeuw described them in the Federal Reserve Bulletin, and then the Federal Reserve began to publish three quarterly series--total manufacturing, primary-processing industries, and advanced-processing industries.(9)

Pressures on capacity again became a concern in 1973 and 1974, when many metals and other industrial materials were reported in short supply and prices were soaring. At this time, the Federal Reserve began to publish capacity utilization rates for major materials (based on physical volume data), first in the Federal Reserve Bulletin and then in the monthly statistical release, "Industrial Production."

In 1976, capacity measures for all ninety-six materials series in the index of industrial production were estimated, and the manufacturing series were reestimated in more detail (fifteen components instead of two) and substantially revised back to 1948. The new monthly Federal Reserve statistical release, "Capacity Utilization," included monthly utilization rates, as well as quarterly data for output, capacity, and utilization, for manufacturing and industrial materials (and their major component series). In 1983, estimates of capacity for mining and utilities and total industry were added for the period back to 1967; and in 1985, the manufacturing components were increased to twenty-four.

The aim of the 1990 revision was to create an integrated, more detailed system of measures for total industry. The materials system has been discontinued as a separate entity, and those components of the materials system that are based on physical unit counts have been incorporated into the expanded system of capacity measures for manufacturing, mining, and utilities. These series are now published monthly in the redesigned Federal Reserve statistical release, "Industrial Production and Capacity Utilization." The new capacity measures cover the period since 1967. The fifty-four individual series in manufacturing have been designed to produce subtotals, primarily at the two-digit SIC level, that are consistent with the old manufacturing measures for 1948-66.

Concepts and Levels of Capacity

Capacity is an ambiguous concept, and capacity and utilization estimates are often rough approximations. No census of plant capacity exists.

Capacity and capacity utilization figures may be used for very different purposes; thus, different definitions and assumptions may be used. One set of assumptions may be appropriate to answer questions about capacity available for mobilization for war, another for questions about inflationary pressures. Moreover, capacity is a measure of maximum output and typically is not achieved; consequently, it is difficult to measure precisely.

Even for a single plant, a range of capacity concepts applies. At the high end of the range would be an engineering maximum based on the rated speed of the machinery in place and operated with minimum downtime. At the low end, economic concepts relate a firm's capacity to a quantity of output that minimizes its unit costs or beyond which its costs rise unacceptably. The surveys incorporate responses between these extremes.

Often, respondents use the concept of practical capacity that the Census survey defines as "the maximum level of production that this establishment could reasonably expect to obtain using a realistic employee work schedule with the machinery and equipment in place" and assuming a normal product mix and downtime for maintenance, repair, and cleanup. Based on this definition, two physically identical factories with identical equipment, but with different work rules and schedules, different qualities of labor and management, and different mixes of products, could have quite different capacities. Asking what the practical capacity of a plant is may be like asking how much water a bathtub can hold without overflowing: It depends on who is in the tub at the time and what he or she is doing.

Besides the conceptual difficulties of measuring capacity at the plant level, an aggregation problem exists. Looking at the capacity of machinery, tools, furniture, and other equipment or structures in great detail, one always sees more capacity than can be used when account is taken of bottlenecks, seasonal shifts in product mix, and so on. The president of an automobile company may see the capacity of the company as less than the sum of the capacities of all the establishments within the company because shortages of key components, such as engines or automatic transmissions, may limit total production. This discrepancy may explain, in part, why the utilization rates reported in the surveys of companies are higher than those reported in the Census survey of plants.(10)

Interindustry bottlenecks may further limit aggregate capacity. In major cyclical peaks, for example, shortages of metals or other industrial materials may limit production of consumer durables and business equipment and keep down apparent utilization of capacity in those industries.

Because the levels of the Federal Reserve measures of utilization are determined by the surveys, in particular the McGraw-Hill survey, which was the first broad survey available, theoretical considerations have not played a major part in determining the level of the rates. In fact, normative considerations have been avoided in determining the overall levels. Thus, no attempt is made to select a figure, such as 100 percent, that signals accelerating inflation or to correct for the effect of interindustry bottlenecks on the broad totals.

A major task for the Federal Reserve in developing reasonable estimates of capacity and utilization is dealing with differences among the various surveys of utilization and with inconsistencies between the movements of the industrial production index and the utilization rates reported in those surveys. A major aim is that the Federal Reserve rates be consistent over time so that, for example, a rate of 85 percent means about the same degree of tightness that it meant in the past.

In its statistical release, the Federal Reserve provides cyclical highs and lows and averages with the monthly utilization rates to give users some historical perspective. For total industry a rate of 88 percent would be very high and would indicate the probability of severe strains in some sectors. For some individual industries, such as paper and pulp, however, 88 percent is moderate.

Output, Capacity, and Utilization

In the absence of monthly surveys of industrial capacity or capacity utilization, the Federal Reserve Board estimates monthly utilization by dividing a monthly industrial production index by a related capacity index. This approach of providing an integrated system of output, capacity, and utilization measures has three advantages over the recently available surveys of utilization alone. (1) It provides a more complete picture of industrial developments and allows the user to see which industries are growing rapidly and which are declining. (2) It provides more current estimates of utilization rates because the production indexes for a given month are available about fifteen days after the end of the month. (3) It provides estimates of utilization that appear to reflect the cyclical movements in production more accurately than the rates based solely on the judgments of respondents to surveys.

Although the Federal Reserve conducts no surveys of capacity or utilization, it uses data on utilization rates from various surveys in estimating its capacity indexes. The Bureau of the Census and McGraw-Hill/DRI have conducted broad surveys of utilization rates, essentially for manufacturing, while various trade associations and other sources have surveyed specific industries. As of this writing, McGraw-Hill/DRI has discontinued its survey (the last data are for the end of 1988) because of a declining response rate and a lack of profitability, and the Bureau of the Census has inadequate funding for its survey of manufacturing utilization rates for late 1989.

The Federal Reserve Method

Six basic steps are involved in calculating the utilization rates published by the Federal Reserve:(11)

Step 1. Preliminary implied end-of-year indexes of industrial capacity (IC) are calculated by dividing a production index (IP) by a utilization rate obtained from a survey ([U.sub.s]) for that end-of-year period. Thus IC = [IP.sub.t]/[U.sub.t]. These ratios are expressed, like the indexes of industrial production, as percentages of production in a base year, currently 1987, and they give the general level and trend of the capacity estimates. Each implied capacity index number is an estimate of maximum sustainable output expressed as a percent of output in 1987. Thus, if in December 1987 the production index is 100 and a related utilization rate from a survey is 80 percent, then the implied capacity index is 100/0.8 = 125. In the absence of information on utilization rates for an industry, trends through peaks in production may be used to estimate capacity output for that industry.

Typically, after a revision of the industrial production index, the capacity indexes must also be revised. For example, in the latest revision, the production index for nonelectric machinery shows much faster growth because of a change in the methodology of measuring computer output.(12) Consequently, the related capacity index (maximum output) had to be similarly revised. The implied capacity ratios (IP/[U.sub.s]) automatically incorporate the production revisions in the estimation of capacity.

Step 2. The annual movements of the preliminary capacity indexes are refined to give consideration to alternative indicators of annual capacity change; these alternatives include capacity data in physical units and estimates of capital stock. In general, the refined estimates of capacity are the fitted values from regressions that retrend the physical capacity or capital stock estimates to the trend growth path of the preliminary implied capacities ([IC.sub.t] = [IP.sub.t]/[U.sub.s,t]).(13) The resulting capacity indexes are generally procyclical--that is, they tend to rise faster in good times than in bad, following investment cycles--but they typically do not fluctuate as much as the preliminary implied capacity series. They will, of course, broadly follow the levels and trends of the preliminary implied capacity indexes.

Step 3. A continuous monthly time series is formed by interpolating between the final end-of-year capacity indexes.

Step 4. An "annual capability adjustment" is applied to estimates of capacity that appear to reflect short-term peak capacity rather than a sustainable level of maximum output. When data sources are changed, a level adjustment may also be required to maintain continuity and consistency with the historical levels based on the McGraw-Hill/DRI survey, which has been the primary determinant of the level of utilization in manufacturing. In this revision, new sources for component series with appropriate level adjustments have been introduced.

Step 5. Value-added weights are applied to the monthly capacity indexes, and the weighted indexes are summed to appropriate groups.

Step 6. Utilization rates for the individual series and groups are calculated by dividing the pertinent production index by the related capacity index. Thus, utilization rates for published groups are not aggregated directly from the utilization rates for components but are derived from aggregate production and capacity indexes for those groups.

The New Structure

The new structure of the Federal Reserve's measures of capacity in manufacturing, mining, and electric and gas utilities is shown in table A.1. Manufacturing and its twenty major (two-digit SIC) industry groups are now summed from more than fifty components for the period since the end of 1966. The eighteen mining components and the separate series for electric and gas utilities are the same as in the previous structure.

The manufacturing series now explicitly incorporate more physical-unit data on output and capacity. Nearly all of these data had been included in the old materials measures, but data for truck assembly plants were newly compiled for this revision.

The greater number of manufacturing series also promotes better estimation or extrapolation of capacity. The industry subgroups within a major group often show different trends or cyclical movements in production, which may be useful in identifying disparate capacity trends. For example, within nonelectrical machinery (SIC 35), the output of engine, farm, construction, and allied equipment declined from 1982 to 1987 whereas the output of office, computer, and accounting machines soared (chart A.1).

Table A.1 also contains information on the sources of capacity and output or utilization rates. Some major groups have multiple, and often disparate, sources. For instance, both the McGraw-Hill/DRI survey and the Census Survey of Plant Capacity provided utilization rates for chemicals and products, major industry group 28; the Census also provided estimates for components within the major group; and trade associations provided physical capacity and output data for plastics resins and other selected components.

Given disparate data and the need to create consistent time series, the Federal Reserve Board staff adopted three operating guidelines in this revision. First, we gave primacy to physical output and capacity data in estimating indexes of sustainable annual capacity. Second, we retained (aside from rebasing) pre-1967 levels of the two-digit groups in manufacturing, which have been based largely on the McGraw-Hill/DRI survey. Third, we adjusted the estimates of manufacturing capacity based on the Census survey (available from 1974 to 1988) and capital stock to levels that maintain consistency over time with the McGraw-Hill/DRI survey (available 1955 to 1988). In general, a simple level adjustment achieved this broad consistency. In some cases, both level and trend adjustments were required because the utilization rates based on the Census survey trend lower over time than those based on the McGraw-Hill/DRI survey.

Often the Census utilization rates and capital stock estimates were used to estimate a single capacity index for a combination of three-digit SIC categories within a two-digit group. In such cases, we weighted Census utilization rates at the four-digit SIC level with value-added figures from the Census of Manufactures to create a survey-based utilization rate for the new combination.(14) Then we followed the steps described earlier to estimate capacity indexes. Capital stock estimates were used to refine the annual movements of capacity for these combinations (step 2 above). The capital stock estimates were derived by the perpetual inventory method from investment data found in the Annual Survey of Manufacturers and adjusted at the two-digit level to BEA capital stock figures.

Weights, Linking, and Utilization: A Technical Note

Table A.1 includes the value-added proportions for 1977, 1982, and 1987 that are applied to the individual capacity indexes when they are summed into aggregate indexes for the period since 1977. These are the same weights that are used to combine series in the production index. As the formulas at the end of the appendix show, using such aggregates of output and capacity to calculate utilization for broad groups is equivalent to summing individual utilization rates weighted by capacity value added. Updating weights allows for using reasonably current price relationships to determine the relative importance of series in the index.

The aggregate output and capacity indexes are initially calculated in overlapping six-year segments, each segment with its own set of weights. For the years 1967-72, 1967 weights are used; for 1972-77, 1972 weights; for 1977-82, 1977 weights; and so on. If past practice continues, the 1987 weights will serve until the introduction of 1992 weights.

Separate indexes are calculated for each six-year weight period; then these segments, each segment expressed as a percentage of its own base year, are linked to form a continuous time series expressed as a percentage of 1987 output. Linking involves finding a constant, called a link factor, that shifts the level of the earlier-based index to the level of the later-based index in the overlapping link period.

Output and capacity indexes for each series are independently linked, and the link factor for each is independently calculated. In most cases, the link factors for output and capacity are nearly the same; however, in some cases the link factors for output differ from those for capacity. Such differences affect the level of utilization. We found that linking tends to raise or lower aggregate utilization rates noticeably when two conditions are present: (1) The relative weights (prices) of series, such as computers, gold ore, or crude oil, change significantly, and (2) the individual series has a utilization rate that differs from the average of the group.

For example, dramatically higher fuel prices incorporated in the 1982 weights increases the relative importance of oil and gas extraction, an industry that operated much closer to capacity in 1982 than did the remainder of mining. Thus, for 1982, utilization in total mining is higher if output and capacity are combined with 1982 weights (86.1 percent) than if they are based on 1977 weights (83.2 percent). Linking capacity at 1982 raises all the utilization rates from 1977 to 1981 by a factor of 1.035 (a ratio of 86.1 to 83.2). For 1977, utilization based on the linked series is 94 percent, 3 percentage points higher than the 1977-weighted (unlinked) utilization rate of 91 percent. This result is unacceptable because, as the formulas show, the unlinked capacity and output series combined with value-added weights from the 1977 Censuses of Manufactures and Minerals Industries provide correct aggregate utilization rates for 1977.

To minimize the distortion of mining utilization caused by linking, the level of linked capacity in 1977 is raised relative to linked output so that the final utilization rate in 1977 is restored (lowered) to the unlinked rate. The aggregate capacity series is then fit (its growth rate is slowed by a constant amount) between the adjusted 1977 level and the 1982 level so that no discontinuities result when new weights are introduced.

In this revision, such adjustments have been applied in half a dozen cases, where the distortions to the utilization rate caused by linking seemed unacceptably large (more than 2 percentage points): mining, metal mining, oil and gas extraction, stone and earth minerals, non-electrical machinery, and total industry.

Aggregation Formulas

These formulas pertain to calculations before linking and can be used to replicate results for the period from 1987 to the present. Definitions:

i--i-th industry

t--t-th period

o--base period

U--utilization rate

C--capacity output

Q--output

P--Census unit value added (price).

Given that [U.sub.i,t] = [Q.sub.i,t]/[C.sub.i,t] and conversely that [C.sub.i,t] = [Q.sub.i,t], and that ([Q.sub.i,i]/[Q.sub.i,o]) is a production index, then the derived capacity index consistent with the above production

index is as follows:

(1) [Mathematical Expression Omitted] The time series of this preliminary, implied capacity index often requires smoothing as explained in step 2 of the methodology.

The capacity indexes are then aggregated by using the same base-period value-added weights that are used in aggregating the comparable production indexes: (2) [Mathematical Expression Omitted] The assumption underlying the use of [P.sub.i,o] in equation 2 is that base-period market prices (actually unit value added) are suitable for aggregating capacities.

This results in aggregate capacity indexes equal to the following: (3) [Mathematical Expression Omitted]

Given that the final aggregate estimates of utilization are aggregates of industrial production indexes divided by the aggregate final capacity indexes and that an aggregate industrial production index equals the following ratio, (4) [Mathematical Expression Omitted] then an aggregate utilization ratio may be expressed as the equation: (5) [Mathematical Expression Omitted] Therefore, (6) [Mathematical Expression Omitted] Thus, aggregate utilization rates reflect combinations of individual output and capacity series that have been weighted with base-period unit value added ([P.sub.i,o]). These aggregate utilization rates are equivalent to capacity-weighted aggregates of utilization rates.

Given that [Mathematical Expression Omitted], and substituting into equation 6, one obtains the following: (7) [Mathematical Expression Omitted] Thus, the aggregate utilization rates are equivalent to combinations of individual utilization rates combined with proportions that reflect current capacity levels of output ([C.sub.i,t]) valued in base-period Census value-added weights ([P.sub.i,o]). [Chart 1 to 2 Omitted] [Chart 1 to 9 Omitted] [Chart A1 to A2 Omitted]

(1)The 1990 revision of the production index, which covers the period since 1977, is described in Kenneth Armitage and Dixon A. Tranum, "Industrial Production: 1989 Developments and Historical Revision," Federal Reserve Bulletin, vol. 76 (April 1990), pp. 187-204. (2)This pioneering survey, which was an important source for the Federal Reserve estimates of manufacturing capacity and utilization, was begun by Douglas Greenwald more than forty years ago. (3)Table A. 1 includes the sources of the estimates. (4)A recent study examining the relationship between manufacturing capacity utilization and producer prices found that, in several two-digit industries, movements in the real level of producer prices respond positively to changes in capacity utilization. Without exception, these industries produce materials and supplies, such as textiles; paper; chemicals; clay, stone, and glass; and primary metals. See William Kan, Reva Krieger, and P.A. Tinsley, "The Long and Short of Industrial Strength Pricing," Finance and Economics Discussion Series 99 (Board of Governors of the Federal Reserve System, Divisions of Research and Statistics and Monetary Affairs, November 1989). (5)In part,, the differing growth rates of Federal Reserve capacity indexes and measures of the capital stock reflect changes in the service flow derived from each unit of capital. Estimates of capital service flow for major sectors of the economy and for two-digit manufacturing industries have been published in recent years by the Bureau of Labor Statistics. For a discussion of the BLS's capital service estimates and their relation to the BEA's capital stocks, see Stephen D. Oliner, "The Formation of Private Business Capital: Trends, Recent Developments, and Measurement Issues," Federal Reserve Bulletin, vol. 75 (December 1989), pp. 771-83. (6)The rise of minimills, which use electric furnaces to make steel from scrap, and heavy investment in computer controls and continuous casters, which save energy and increase the yield of steel mill products from raw steel, contributed to increases in productivity. A ton of raw steel goes further than it once did. The finished steel yeild is now about 85 percent of raw steel tonnage, compared with 71 percent in 1977. Thus, 84 million tons of steel mill products were shipped in 1989, down 16 percent since 1979, compared with production of 97 1/2 million tons for raw steel in 1989, down 30 percent over the same period. (7)In this revision, the capacity index for motor vehicles and parts has been built from separate estimates for assemblies of automobiles, light trucks, medium and heavy trucks, and parts. Previously, the estimate for motor vehicles was based on utilization rates from the McGraw-Hill/DRI survey supplemented by those from the Census survey of utilization. A separate series on automobile assemblies based on physical counts was shown as an addndum item. (8)Utilities build sufficient generating facilities to safely surpass peak demands such as those that occur on a hot August afternoon. Most of the time there is a lot of spare capacity. The estimates of utilization that the Federal Reserve has constructed reflect the excess of generating capacity over the desired level of capacity, that is, a capacity level high enough to safely surpass expected noncoincident peaks in generation. The industry publishes much lower utilization rates based on the much larger excess of generating capacity over annual generation. The Federal Reserve chose an economic definition of capacity--called annual capability--because a utility must invest in more capacity when summer brownouts threaten even though the annual generation load may be only half of annual generating potential. (9)Frank de Leeuw with Frank E. Hopkins and Michael D. Sherman, "A Revised INdex of Manufacturing Capacity," Federal Reserve Bulletin, vol. 5 (November 1966), pp. 1605-15. (10)Another reason is that the company surveys, especially the McGraw-Hill survey, are relatively small samples of large companies that respond voluntarily. These firms may have above-average utilization rates. (11)This section is an abridgment of the methodological appendix found in Richard D. Raddock, "Revised Federal Reserve Rates of Capacity Utilization," Federal Reserve Bulletin, vol. 71 (October 1985), pp. 760-66. (12)The methodology for estimating the new computer series, the annual levels of which are derived by dividing current dollar figures by the BEA price index for computers, is described in Armitage and Tranum, "Industrial Production." (13)The fitted values from a regression of the equation below give an estimate of the difference in the trends of the implied capacity and the annual capacity indicator: [Mathematical Expression Omitted] where

[IC.sub.t] = implied capacity index in period t

[K.sub.t] = annual capacity indicator

[f.sub.t] (t) = specified functions of time

a, [b.sub.i] = parameters to be estimated

[e.sub.t] = error term. The refined capacity estimates are taken to be the annual capacity indicators multiplied by the antilogarithms of the fitted values from the equation. (14)The formula is [Mathematical Expression Omitted] where U is a utilization rate, Va is value added, and i is the individual four-digit industry in the group. (15)To simplify the formulas, we are expressing the production indexes in terms of base-period output equal to one. The utilization rates are expressed as decimal fractions.