Industry contributions to productivity growth in U.S. manufacturing: an application of alternative output concepts

The manufacturing sector is an important part of the U.S. economy. Yet, looking at the manufacturing sector as a whole can mask the varied performance of the many diverse industries that make up this goods-producing sector. An industry’s performance along with its relative size will determine how the performance of an individual industry contributes to manufacturing sector performance. Evaluating industry contributions to manufacturing sector performance reveals which industries are dragging down total factor productivity (TFP) growth in the sector and which are enhancing it.

TFP compares growth in the production of goods with changes in the inputs used in production. By capturing the growth in output that is not a result of using more labor, capital, energy, materials, and purchased services, TFP is often used as an indicator of technological progress and performance. In the manufacturing sector, TFP grew at an average compound rate of 0.60 percent per year from 2000 to 2021.¹ Yet, across industries within manufacturing, TFP growth ranged from 3.87 percent in the computer and electronic products industry to a negative 0.47 percent in the chemical products industry.

Looking across the last three business cycles (2000–07, 2007–19, and 2019–21), we present U.S. Bureau of Labor Statistics (BLS) published TFP measures for the manufacturing sector and for 19 industries within manufacturing and estimate the amount contributed by each industry to overall manufacturing sector performance.² We discuss changes in industry average shares, or relative size, and industry TFP growth. We also discuss the considerable variation in TFP across industries and in the amount contributed by each industry to manufacturing TFP across business cycles.

Next, we explore whether industry contributions are affected when TFP is measured by using three alternative output concepts (value-added output, sectoral output, and gross output). We use an experimental production account to investigate how output choice affects analysis of industry contributions to manufacturing sector TFP. BLS uses a sectoral-output concept to measure TFP in manufacturing. Sectoral output represents the value of output leaving the sector or industry, by excluding the value of intermediate inputs that are produced and consumed within that same industry or sector (i.e., intrasectoral transactions). By contrast, value-added output excludes all intermediate inputs, while gross output does not exclude any intermediate inputs. As discussed in the companion article, “The importance of output choice: implications for productivity measurement,” the choice of output measure (and by extension, the treatment of intermediate inputs in output) has substantive implications for TFP measurement.³

Industry TFP growth and contributions to manufacturing sector performance

TFP growth is measured as the percent change in the ratio of output to the weighted sum of inputs used in the production process. Outputs and inputs are measured either in physical quantities or in deflated values, adjusted for price change. Growth in TFP is commonly presented as⁴

where are cost-share weights for each input i. This model was developed by Robert Solow in 1957 and assumes constant returns to scale, implying that the value of output equals the total cost of all measured inputs and cost shares sum to 1.⁵

The corresponding output and input trends underlying manufacturing sector and industry TFP growth in the 2000–21 period are presented in table 1. Manufacturing sector TFP growth of 0.60 percent per year is a result of no growth in output and a decline of 0.60 percent in the use of combined inputs during this period. Although capital input increased at a 1.64-percent rate, the use of both labor and intermediate inputs declined by 1.01 percent and 1.50 percent, respectively. Among manufacturing industries, the computer and electronic products industry achieved the highest level of TFP growth because output grew at a rate of 0.76 percent annually and combined inputs declined at a 3.00-percent annual rate. This decline in inputs reflects a large 9.78-percent annual decrease in the growth of intermediate inputs accompanied by a more moderate 2.0-percent decrease in labor input and a 1.62-percent increase in capital input. The printing industry achieved the second-highest TFP growth over this period, with a decrease in output growth of 2.26 percent offset by an even greater 3.39-percent fall in the growth of combined inputs. By contrast, the chemical industry experienced the largest decline in TFP over this period, with the growth in combined inputs (0.72 percent) outpacing the growth in output (0.25 percent). Among inputs in the chemical industry, capital input grew at a rate of 3.47 percent annually, while labor experienced no growth and intermediate inputs of materials, energy, and services declined.

Chart 1 illustrates the relationships among TFP, output, and combined inputs for manufacturing industries. Of the 19 manufacturing industries, 15 experienced positive TFP growth. Among these 15 industries, only 5 had positive output growth that exceeded the growth of combined inputs. The remaining 10 industries experienced decreasing growth in output offset by an even faster decrease in growth of combined inputs. Growth in capital inputs declined in 4 of these 15 industries, while labor input growth decreased in 13 and growth of intermediate inputs slowed in all but 2 of the 15 industries. The four industries with negative TFP growth exhibited varied patterns of output and input use. The chemical industry experienced growth in output accompanied by higher growth in combined inputs. However, the fabricated metal products industry, the furniture industry, and the apparel industry experienced negative growth in combined inputs offset by even faster declines in output growth.

To determine how industries contribute to manufacturing sector productivity, we weight the TFP growth rates for each industry by the value of that specific industry’s share of sector output. The weights for each industry are the industry’s current-dollar sectoral-output share of the aggregate manufacturing sector’s sectoral output.⁶ The industry weights reflect not only the contributions of the primary inputs—capital and labor—to production but also the contributions of the intermediate inputs—energy, materials, and purchased services. These industry weights will sum to a value greater than 1 since the numerator—the value of sectoral output in each industry—will include intermediate inputs purchased from outside the industry. The denominator, on the other hand, includes only the value of intermediate inputs purchased outside the aggregate manufacturing sector. TFP growth in any one industry will augment productivity growth in other industries that use other manufactured goods in their production processes.⁷

Chart 2 shows the relative contribution of each industry to manufacturing sector TFP growth from 2000 to 2021 by using BLS published data. For this period, the computer and electronic products industry leads both in TFP growth and in contribution to manufacturing sector TFP. This result reflects not only the highest manufacturing industry TFP growth but also a relatively large average industry share of 0.10 for the 2000–21 period. By contrast, the printing and related support activities industry had a relatively strong TFP growth rate of 1.17 percent per year. Yet, it contributed only 0.03 percentage point to growth in manufacturing sector TFP because of a small 0.03 industry share. While TFP growth rates in the motor vehicles industry and the paper products industry were similar, 0.53 percent and 0.52 percent, respectively, the motor vehicle industry contributed 0.07 percentage point to manufacturing sector TFP growth over the 2000–21 period and the paper products industry contributed 0.02 percentage point. This difference in contribution occurs because the motor vehicles industry’s share of sector output (0.12) is 3 times larger than the paper products industry’s share (0.04). Combining industry TFP measures with information on the industry’s share of manufacturing sector final demand enables those industries driving or detracting from sector-level TFP growth to be identified in any given period.

Evolution in manufacturing across business cycles

Looking at changes in industry contributions, industry shares, and TFP growth across business cycles provides insights into how the manufacturing sector has evolved over the past 20 years. The 2000–21 period includes three business cycles: 2000–07, 2007–19, and 2019–21. Table 2 presents each industry’s contribution to manufacturing sector TFP growth, industry average share, and industry TFP growth, across the last three business cycles.⁸

TFP in the manufacturing sector grew at a 1.74-percent annual rate from 2000 to 2007. During this period, the growth in industry-level TFP ranged from a positive 6.59 percent per year in computer and electronic products to a negative 1.42 percent in the apparel industry. Output in the computer industry grew considerably, at 2.59 percent, while labor input declined at a 4.01-percent rate and intermediate inputs fell by 5.27 percent annually. (See table 3.) The apparel industry suffered an extreme decrease in output growth of 14.18 along with a decline in use of combined inputs of 12.94, which reflects a 17.64-percent decline in intermediate inputs. Following the Great Recession (2007–09), manufacturing sector productivity declined on average 0.18 percent per year from 2007 to 2019. During this business cycle, computer and electronic products maintained the fastest TFP growth, at 2.48 percent, while the largest negative growth was in chemical products with a decline of 1.69 percent. The growth in TFP in the computer industry resulted from an annual decline in output of 0.48 percent and a faster decline in combined inputs of 2.89 percent annual growth. The fall in combined inputs reflects a 1.17-percen increase in capital input, offset by a 0.99-percent decline in labor input and a 12.15-percent decline in intermediate inputs. With the largest decline in TFP in this period, the chemical industry output growth rate fell 1.51 percent, while growth in combined inputs increased slightly at a 0.18-percent rate. This small increase in combined inputs reflects a large 3.28 growth in capital inputs, accompanied by a small 0.38-percent growth in labor input and offset by a 1.89-percent decline in intermediate inputs.

In the most recent business cycle, which began at the onset of the COVID-19 pandemic, manufacturing TFP grew at a positive rate of 1.32 percent per year, on average through 2021. From 2019 to 2021, annual industry TFP growth has ranged from 2.92 percent in food and beverage to a negative 3.90 percent in primary metals. Output in the food industry had a small positive increase in growth of 0.80 and a larger decrease of 2.06 in growth of combined inputs in this period. Although both capital and labor inputs increased at 1.13 and 1.76 rates, respectively, intermediate inputs declined at a 3.83-percent rate. The decrease in TFP growth in primary metals reflects large declines in both output and combined inputs. Output declined at an 8.28-percent rate while combined inputs fell at a rate of 4.56. Only capital input increased slightly, at 0.05 percent, during this period, whereas labor input and intermediate inputs both declined sharply, at 5.79 percent and 5.35 percent, respectively.

Chart 3 illustrates the variation in TFP growth in manufacturing industries over the 2000–07, 2007–19, and 2019–21 business cycles, arranged from fastest to slowest industry TFP growth in the 2000–07 period. As just noted, the computer, printing, and motor vehicles industries exhibited the largest TFP growth in the 2000–07 period. From 2007 to 2019, TFP growth was highest in the computer, apparel, and printing industries. During the recent cycle, 2019–21, TFP growth was the strongest in the food, plastics and rubber products, computer, and chemical products industries.

Changes in industry shares

In reviewing industry average shares in table 2, we notice that in 13 industries, the sectoral-output share of manufacturing sector output remained relatively stable over the three business cycles, varying by one or two hundredths of a percentage point or less. The output shares of the remaining six industries varied more widely over time, because the industries' relative importance to the manufacturing sector has increased or diminished. These six industries include the computer, petroleum, chemical, food, motor vehicles, and apparel industries. Of these six industries, the computer industry and the petroleum industry have seen the largest changes in industry shares over the three business cycles.

The computer industry’s output share, at 0.13, placed it as the fourth-highest output share in the 2000–07 business cycle. This output share declined to 0.09 in the next business cycle, from 2007 to 2019, and has fallen to 0.08 in the most recent, still incomplete, 2019–21 business cycle. The computer industry's share of manufacturing output peaked in the late 1990s. From 2001 forward, the industry’s share of manufacturing output began to decline steadily, as shown in chart 4.

The petroleum industry’s share of manufacturing sector output jumped from 0.10 in the 2000–07 cycle to 0.16 during the 2007–19 cycle, when U.S. production of petroleum products increased because of the shale oil boom, and then declined to 0.12 in the more recent 2019–21 period.⁹ Over the 2000–21 period, the petroleum and coal industry experienced the second-largest gain in industry share of 0.04. The fortunes of the petroleum and coal products industry stemming from the fracking revolution are readily apparent by observing the soaring shares of this industry from 2005 to 2019.

The chemical industry had the largest share gain among manufacturing industries, with a 0.04-percentage-point increase in share of manufacturing output over the 2000–21 period. The average industry share increased from 0.15 in the 2000–07 business cycle to 0.17 during the 2007–19 and 2019–21 business cycles. The annual share peaked in 2011 before declining marginally for several years and then increasing to 0.18 in 2021.

The food industry’s share of manufacturing sector output increased 0.05 percentage point from 2000 to 2021. The industry’s average share increased from 0.17 in the 2000–07 business cycle to 0.19 from 2007 to 2019 and 0.20 from 2019 to 2021. Shares in the food industry peaked in 2010 before declining through 2014 and increasing slightly thereafter.

Shares in the motor vehicles industry decreased overall by 0.02 over the 2000–21 period. The industry share declined from 2004 to 2009, illustrating a small but steady decrease in industry output relative to manufacturing sector output, beginning in 2004. This slow decline in output share was followed by a rapid drop in the share because of the severe duress that affected U.S. auto production following the Great Recession of 2007–09 and a strong recovery brought about by the subsequent automobile bailout in which funds from the “Troubled Assets Relief Program” or TARP were used.¹⁰ During the Great Recession, the motor vehicle industry’s share declined from 0.12 in 2007 to 0.09 in 2009, before recovering to a high of 0.14 in 2016. Industry share declined again during the 2019–21 pandemic business cycle, from 0.13 in 2019 to 0.12 in 2021.

Finally, the apparel industry has been declining for several decades, and this decline continued over the 2000–21 period. As seen in chart 4, the apparel industry’s share of manufacturing sector output declined steadily over the 2000–21 period, losing nearly 0.02 percentage point as it declined from a share of 0.023 to less than 0.004. Thus, as chart 4 clearly shows, economic restructuring is important in the evolution of industries’ contributions to manufacturing sector TFP growth.

Industry contributions over time

Industry contributions over the three business cycles were more varied and reflect changes in individual industry TFP growth rates and shifts in industry shares. The computer industry was the largest contributor to manufacturing sector growth during the 2000–07 and 2007–19 business cycles, with the highest TFP growth of any manufacturing industry. However, this industry’s average share of sector output was only the fourth highest in 2000–07 and sixth highest in 2007–19. During the 2019–21 period, including the COVID-19 pandemic years, the food and chemical industries’ contributions to manufacturing final demand exceeded those of the computer industry. These higher contributions were primarily because of large increases in TFP growth in both industries, compared with those of the previous cycle. Previously, the food and chemical industries ranked as the 3rd- and 5th-highest contributors to manufacturing TFP in the 2000–07 cycle and the 18th and 19th highest, respectively, during the 2007–19 cycle. With shares consistently among the top three industries since 2000, this variation in industry contribution by the food and chemical industries was driven by fluctuation in these industries’ TFP growth across business cycles.

Table 4 highlights the top five contributing industries for each of the three business cycles. As just noted, the computer and electronic products industry was by far the largest contributor to manufacturing sector TFP growth in the 2000–07 cycle. Its contribution was nearly 3 times as large as the second-highest contributing industry during this period, motor vehicles. Although TFP growth in the computer industry fell by half in the 2007–19 period, this industry continued as the top contributor with a smaller yet still strong industry average share. The petroleum and coal products industry moved from the least contributing industry over the 2000–07 period to the second-highest contributor in the 2007–19 period. This higher contribution reflects an output share that jumped from 0.10 during 2000–07 to 0.16 in the 2007–19 cycle and TFP growth that correspondingly increased from a declining rate of 0.12 percent per year to 0.21 percent. The industry’s increased output share and higher TFP growth mirror the historic increase in U.S. oil and gas production from 2010 to 2020, as a result of the “shale boom” and increased fracking.¹¹The dynamics of the 2019–21 postpandemic recovery period led to a redistribution of the top contributing industries. In this period, the food and beverage and tobacco products industry became the top contributor, with the strongest industry share of 0.20 and the highest industry TFP growth of 2.92 percent. The chemical products industry moved to the second-highest contributing industry position, with its typically high industry share and rapid TFP growth of 2.22 percent. The computer industry, although still a large contributor to sector-level TFP during the 2019–21 period, fell to the third highest. The plastics and rubber products industry moved to the fifth-highest contributing industry, reflecting its position as the industry with the second-fastest TFP growth in this period.

Different output concepts and industry contributions

The choice of output measure has implications for the measurement and interpretation of productivity.¹² Here, we compare industry contributions constructed using three alternative manufacturing sector TFP measures. These alternative TFP measures are developed by using experimental value-added-, sectoral-, and gross-output measures from our research production account for the manufacturing sector and industries.¹³ Recall that the BLS TFP data for manufacturing rely on a sectoral-output definition. We analyze differences among industry contributions on the basis of alternative output concepts and discuss the underlying causes of differences in these measures.

The three most common output concepts used are value-added output, sectoral output, and gross output. The differences among these output concepts are a result of which intermediate inputs, if any, are included in the output measure. Value-added output is a narrowly defined concept of output including only primary inputs of capital and labor and reflects only the additional value of transforming intermediate inputs into outputs. Gross output, on the other hand, is the broadest measure of output and includes capital, labor, and all intermediate inputs (energy, materials, and services). Gross output represents the total value of goods and services produced by all firms in an industry or sector, regardless of whether they are sold directly to consumers or sold to other firms to become an input for further production. As a result, an output is counted when it is sold and counted again in the value of the product it is used to produce. Sectoral output lies between the extremes of value-added output and gross output, by including capital, labor, and intermediate inputs purchased from outside the industry or sector being measured. Sectoral output is greater than value-added output by including the value of intermediate inputs produced outside the industry but is less than gross output by excluding the value of intermediate inputs produced within the industry, intrasectoral transactions.

In the TFP growth accounting model, assuming constant returns to scale ensures that measured nominal output is equal to the sum of the costs of all inputs included in the output concept being measured. Nominal output under all three output concepts (where VA = value-added output, SO = sectoral output, and GO = gross output) will include the costs of capital and labor inputs. However, the nominal value of sectoral and gross output will also include part or all of the costs of intermediate inputs of energy, materials, and services. Because of this commonality, TFP measures constructed with the use of these three different output concepts are mathematically interrelated.¹⁴ These relationships are

                                    TFP_SO growth = (  ) × TFP_VA growth,                                   (2)

                                    TFP_GO growth = ( ) × TFP_VA growth, and                           (3)

                                    TFP_GO growth = ( ) × TFP_SO growth.                                          (4)

As equation (2) shows, measures of TFP growth in the sectoral model are proportionally smaller than measures of TFP growth in the value-added model, by the ratio of value-added output to sectoral output for that industry. Similarly, equation (3) reveals that gross-output TFP will have proportionally smaller growth than value-added TFP, by the ratio of nominal value-added output to nominal gross output. Gross-output TFP will also have proportionally smaller growth than sectoral-output TFP, by the ratio of sectoral output to gross output, as described in equation (4).¹⁵

To estimate how an industry j contributes to manufacturing sector (MN) TFP, we construct weights for each industry as the industry’s current-dollar output share of the manufacturing sector on the basis of the concept being used. For example, when using the value-added framework to estimate industry contributions to manufacturing sector TFP, we construct industry weights as the industry’s current-dollar value-added-output share of value-added output for the manufacturing sector, VA_j/VA_MN.¹⁶ In the gross-output framework, weights for each industry are the industry’s current-dollar gross-output share of manufacturing sector gross output, GO_j/GO_MN.¹⁷ The gross-output model double counts the value of intermediate inputs produced and sold within an industry, both when sold to another producer for use in production and when sold to be consumed as a final product. For this reason, the gross-output model is seldom recommended for industry contribution analysis. In the value-added-output and gross-output models, industry weights will sum to 1.

Within the sectoral framework, weights for each industry are the industry’s current-dollar sectoral-output shares of sectoral output for the manufacturing sector, SO_j/SO_MN. As just noted in our BLS published measures, the industry weights in the sectoral-output model will sum to a value greater than 1 because TFP growth in any one industry will augment productivity growth in other industries.¹⁸

Chart 5 compares industry shares by using the three different output concepts, for 2019. Note that because sectoral-output shares will sum to a value greater than 1, sectoral-output shares tend to be larger than the value-added and gross-output-based industry shares for most industries. This result is true throughout the 2000–21 period. In addition, because gross output for the manufacturing sector double counts the value of intermediate inputs and is quite large, gross-output industry shares are uniformly and substantially less than sectoral-output shares.¹⁹ As evident in chart 5, sectoral-output shares in 2019 were higher than value-added shares for 16 of the 19 industries. For example, the sectoral-output share in food and beverage was much greater than the value-added-output share in 2019. This difference can be explained by noting that food and beverage had the highest share of sectoral intermediate inputs among all manufacturing industries, at 30 percent. However, it only had the third-highest share of capital costs, at 12 percent, and the second-highest share of labor costs, at 11 percent, among manufacturing industries in 2019.

Chart 5 also shows, by comparison, the value-added share for the computer industry was larger than the sectoral-output share in 2019. In this year, the computer industry had the third-lowest share of sectoral intermediate inputs among manufacturing industries and the first- and second-highest shares of labor and capital costs, respectively, among manufacturing industries. Also note that in an industry, as the relative use of capital, labor, and sectoral intermediate inputs changes over time, the relationship between sectoral and value-added-output shares in the industry may vary. In the computer industry, for instance, sectoral shares over the 1988–2003 period initially exceeded value-added shares. Then, as a result of a continuous, dramatic decline in purchases of intermediate inputs from outside the sector, accompanied by more moderate variation in capital and labor costs, value-added shares over the 2004–21 period became larger than sectoral shares. Gross-output shares similarly may be larger or smaller than value-added-output shares, depending on the relative shares of primary and intermediate inputs among the manufacturing industries. For 2019, gross-output shares were larger than value-added-output shares in 8 industries and smaller in 11 industries.

Impact of output concepts, 2000–21

We estimate the contribution of individual industries to overall manufacturing sector TFP growth by using each of the output concepts: value-added output, sectoral output, and gross output. Chart 6 summarizes the relative contribution of each industry to manufacturing sector TFP growth for the 2000–21 period as measured by using each output concept. In this longer period, in general, the rankings of industry contributions were similar.

However, for a few industries, the choice of output measure still had a large impact on the estimated contribution to manufacturing sector TFP growth. Table 5 compares manufacturing industry contributions, shares, and TFP across the three alternative output measures. Under the value-added-output framework, the motor vehicles, bodies and trailers, and parts industry was a drag on manufacturing sector TFP, with a negative industry contribution of 0.105 percentage point. Conversely, using either the sectoral or gross-output frameworks moved the motor vehicle industry from 18th- to 2nd-highest contributing industry, with positive contributions of 0.058 and 0.038 percentage point, respectively. Thus, the impact of including purchases of intermediate inputs on industry contribution measures was especially evident for the motor vehicle industry. This movement reflects both higher average industry share values and small but positive growth in TFP when intermediate inputs are included, to some degree, in the output measure. This industry was the second-largest user of intermediate inputs, the third largest user of within sector intermediate inputs, and the second-largest user of intermediate inputs outside the sector in 2021.

As illustrated in chart 6, under the value-added-output framework, food and beverage and tobacco products was the second most important contributing industry to TFP growth, with a contribution of 0.096 percentage point. When sectoral output was used to measure TFP, the industry fell to the third most important contributing industry and the industry contribution to overall manufacturing sector TFP declined to 0.053. This result reflects the inclusion of intermediate inputs that are purchased from outside the industry in the calculation. Under the gross-output model that adds intermediate inputs produced within and outside the food industry, the industry remained the third-highest contributing industry with a smaller positive contribution of 0.032 to manufacturing sector TFP growth.

Similarly, the plastics and rubber products industry was the third highest to contribute to the manufacturing sector TFP growth, contributing 0.081 percentage point, under the value-added model. When intermediate inputs are considered by using either sectoral or gross output to measure TFP, this industry’s rank changed to seventh-highest contributing industry, with corresponding contributions of 0.031 and 0.022.

The degree to which industry contributions to manufacturing sector TFP growth vary by the choice of output concept is somewhat muted over an extended period such as 2000–21. This dampening of variation reflects the averaging of cyclical changes over a longer period. We next examine the effect of output choice on variation in TFP growth, industry shares, and industry contributions across business cycles.

Impact of output concepts across business cycles

Table 6 presents TFP growth rates by output measure for the 2000–07, 2007–19, and 2019–21 business cycles. At the industry level, sectoral and gross TFP measures are similar because they differ only by the intermediate inputs produced and consumed within the industry. However, when TFP is measured using the value-added model, only capital and labor are included in the model. Because all intermediate inputs are removed from the value-added model, value-added TFP may differ widely from sectoral and gross TFP. For the manufacturing sector, TFP grew in all three models during the 2000–07 and 2019–21 business cycles and declined during the 2007–09 business cycle. However, this pattern is not present for all industries. For example, the petroleum industry had slower or negative TFP growth in the 2000–07 and 2019–21 business cycles than in the 2007–19 period.

Industry shares of aggregate sector output under the value-added-, sectoral-, and gross-output measurement frameworks, respectively, are presented in table 7 for the three business cycle periods.²⁰ Industry shares change over time as the value of the industry’s output relative to the sector increases or decreases. For example, the computer industry’s sectoral and gross-output shares of manufacturing sector output averaged 0.13 and 0.10, respectively, in the 2000–07 business cycle, before declining rapidly, arriving at values of 0.08 and 0.06 in the 2019–21 business cycle. With the use of the value-added model, the share of the computer industry was relatively stable over time. Similarly, the petroleum industry’s output shares had greater volatility across business cycles when the sectoral and gross-output models were used as compared with the value-added model.

Table 8 and chart 7 compare industry contributions with manufacturing sector TFP across business cycles by using the three output concepts. The data in chart 7 are ranked by using the 2000–07 contribution values from the value-added model.

In the 2000–07 business cycle, the computer industry was the highest contributor to manufacturing sector TFP growth while the motor vehicles industry was the second-highest contributor in all three output models. The computer industry retained its position as the greatest contributor to manufacturing sector TFP growth in the next business cycle period, 2007–19. However, the contribution of the motor vehicle industry fell to among the lowest four contributing industries. In addition, the petroleum and coal industry rose from the least contributing industry in the 2000–07 cycle to the second-highest contributor, in all three models, during the 2007–19 period. Industry TFP growth in all output models was typically slower in the 2007–19 cycle that included the Great Recession (2007–09) than in the 2000–07 period. Numerous industries that previously contributed positively to manufacturing TFP growth instead had a negative or diminishing effect in this second business cycle period, including the chemical industry and the food industry. In the still incomplete 2019–21 business cycle, the choice of output model more greatly affected industry contribution than it did in the prior two cycles. Under the value-added-output model, the chemical products industry ranked as the greatest contributor to manufacturing sector TFP, with the food industry as the second-highest contributor. These two rankings, although quite close, were reversed under the sectoral- and gross-output models. In addition, during this period, the ranking of industry contributions by output measure exhibited much more variation among the top 12 industries.

The values of industry contributions to manufacturing sector TFP growth derived from the value-added model are markedly different from industry contributions derived from the sectoral and gross-output models in several industries. This pattern reflects the variation in TFP growth for each output model, across industries, and the differences in industry shares among the value-added-, sectoral-, and gross-output models.

Summary

To understand the performance of the U.S. manufacturing sector, we must look beyond sector-level data and explore the performance of individual industries. Using BLS published data, we have shown that any one industry’s influence on sector-level performance is a function of both its size in the sector and its TFP growth. Both the industry share of output and industry TFP growth will vary during different periods, reflecting the underlying dynamics of production in any given industry. For example, over the entire period from 2000 to 2021, the computer and electronic products industry contributed over half of the growth in manufacturing sector TFP. However, in the recent 2019–21 period, the food, beverage, and tobacco products industry contributed the most.

By tracing the contribution of industries to manufacturing sector TFP over three different business cycles, we clearly show the rise and fall of industries’ importance to overall manufacturing sector performance and their link to technological innovations, changing economic needs, and global economic patterns. The dramatic increase in U.S. oil and natural gas production because of the development of revolutionary hydraulic fracturing and horizontal drilling techniques increased the average industry share of manufacturing output produced by the petroleum industry by 60 percent, from 0.10 in 2000–07 to 0.16 in the 2007–19 period. This rise in industry share, combined with increased, positive TFP growth of 0.21 percent per year in the 2007–19 period, resulted in a petroleum industry contribution of 0.053 percentage point to manufacturing sector TFP growth. This industry’s positive contribution compares with its earlier impact as a drag on manufacturing sector TFP growth, with an industry contribution of negative 0.111 percentage point in the 2000–07 period.

Similarly, innovations in the chemical and food industries resulted in large increases in TFP growth in the recent 2019–21 period compared with the 2007–19 period, repositioning these industries as the top two contributing industries to manufacturing sector TFP growth in 2019–21 versus the two lowest industries with negative contributions to overall sector TFP in the 2007–19 period. This dramatic shift in the ranking of these two industries occurred despite a slight increase in the food industry’s average share (0.19 to 0.20) and a larger decrease in the petroleum industry’s average share (0.16 to 0.12) from the 2007–19 period to the most recent period.

By using three different output concepts (value-added output, sectoral output, and gross output), our comparison of industry contributions to manufacturing sector TFP growth shows that the choice of output measure directly affects productivity analysis both through implicit differences in the resulting empirical measures of TFP and through the estimation of industry contribution to TFP growth in any given sector.

First, the choice of output measure affects the measurement of TFP growth. Second, it affects the estimate of an industry’s contribution to aggregate TFP growth via both the output-based share weighting scheme, which varies depending on the output concept selected, and the measure of TFP growth. A TFP measure based on value-added output reflects only the effect of primary inputs—capital and labor—on the production of output. TFP measures based on sectoral output reflect the effect of intermediate inputs purchased outside an industry on the production of output, while TFP measures based on gross output reflect the effect of intermediate inputs purchased both outside and within an industry on output production. Each of the respective TFP measures—value-added output, sectoral output, and gross output—can be interpreted differently because of the inherent constraints embedded in the related growth accounting model. In addition, the selection of the output concept will affect the value of the industry’s share of sector output. Even over a longer term such as 2000–21, the choice of output model had a notable effect. The motor vehicles industry moved from the second-lowest negative contributor to manufacturing TFP growth under a value-added model to the second-highest contributing industry under either a sectoral or gross-output model. The use of alternative output measures across shorter business cycle periods revealed more pronounced variability in industry ranking by contribution to sector TFPs.

By expanding on the differences among the three output measures and how these differences affect the related TFP growth rates, industry share weights, and industry contribution measures, we show that careful deliberation is warranted before selecting a value-added-, sectoral-, or gross-output framework for TFP and contribution analysis. Using a sectoral output framework not only avoids the double-counting issues that render the gross-output framework difficult to interpret and of little analytical value but also reflects the effect on productivity of capital, labor, and intermediate inputs purchased outside a given industry. Although using the value-added-output framework also avoids the double counting inherent in the gross-output model, the value-added output captures only the increase in TFP resulting from the use of capital and labor inputs in production. Industry shares in the value-added framework reflect only differences in the use of capital and labor inputs by industry, whereas sectoral shares reflect changes in the relative use of capital, labor, and intermediate inputs. For these several reasons, BLS selected the sectoral-output framework as the most informative and least compromised output framework in analyzing productivity of the manufacturing sector and industries.