US manufacturing employment has fallen from over a quarter to less than 10% of total employment over the last five decades. This decline has garnered significant policy attention, with both Republicans and Democrats proposing a range of policies to boost domestic manufacturing.
Research in both international trade and macroeconomics offers valuable insights into this transformation. In trade, which generally focuses on manufacturing, the decline in manufacturing employment has been linked to import competition, most recently from China (Autor et al. 2013, Acemoglu et al. 2014, Pierce and Schott 2016). In macro, the relative growth of non-manufacturing versus manufacturing employment is generally explained in terms of unbalanced productivity growth (Baumol 1967, Ngai and Pissarides 2007, Duernecker et al. 2018), rising incomes and non-homothetic demand (Comin et al. 2018), or both (Matsuyama 2019).
In a new paper (Ding et al. 2022), we take a firm-level view of structural change that captures essential elements of both perspectives while highlighting the role of firms’ development of proprietary knowledge. Our approach is motivated by a detailed decomposition of US employment growth over the last four decades, which reveals that a small set of continuing manufacturing firms accounts for a substantial share of the US shift towards services. Non-manufacturing employment growth among these firms, as for the US as a whole, is concentrated in Business Services, particularly high-skill, technologically advanced Management and Professional, Scientific, and Technical Services (PSTS) that are used as inputs across a wide range of sectors, and which relate to their past manufacturing activities.
In the first part of the paper, we provide a comprehensive, firm-level summary of the US transition towards services by constructing a 40-year panel of the universe of firms and the detailed industries in which their establishments operate. In line with our interest in studying the full range of US manufacturing, we broadly define a manufacturing firm as any firm that contains a manufacturing establishment during our sample period. In line with our emphasis on how firms accumulate knowledge, we summarise two conceptualisations of a firm. The first definition uses conventional longitudinal firm identifiers for collections of establishments with a common owner. The second, broader definition allows for the possibility that firm knowledge may continue after a firm exits as long as some of its plants survive, for example, due to merger and acquisition activity. Using these two definitions, we find that manufacturing firms account for 16% to 32% of the rise in aggregate US non-manufacturing employment between 1977 and 2019, as illustrated in Table 1. As indicated by the breakdown of employment growth by intensive and extensive margins, a very small number of continuing manufacturing firms account for virtually all of this growth.
Table 1 Employment growth in manufacturing and non-manufacturing from 1977 to 2019, by firm type and margin
Manufacturing firms' non-manufacturing growth does not simply reflect aggregate trends (see Figure 1). Rather, it is concentrated in Business Services and Wholesale and Retail, while non-manufacturing firms’ growth is strong not only in Business Services, but also in Health Care. Within Business Services, manufacturing and non-manufacturing firms also differ, with manufacturing firms’ growth concentrated in Computer Systems Design, Research and Development (R&D), and Architectural and Engineering Services. These activities relate to manufacturing, both because they constitute key inputs (e.g. R&D) and because they rely on related physical outputs (e.g. Computer Systems Design services uses computers). Moreover, we find that continuing manufacturing firms increasingly pivot towards these services over the period. For example, continuing firms with Computer Systems Design establishments at the end of our sample period exhibited an average decline in their manufacturing employment share from 40% to 15% from 1977 to 2016. These transitions suggest a mix of functional and sectoral structural change within firms, i.e. greater provision of input services to produce the same final goods (functional) and transition from sales of goods to sales of services (sectoral).
Figure 1 US employment and payroll by sector and firm type
Source: LBD and authors' calculations.
Notes: The figure displays US employment (top panel) and payroll (bottom panel) by broad NAICS sector and firm type from 1977 to 2019. Other includes Educational (61), Arts, Entertainment and Recreation (71), Other (81) and Public Administration (92). Payroll data for 1988 are missing.
We posit that proprietary knowledge provides firms with a competitive advantage related to their successful transition to services. Our findings contribute to a growing body of research on intangible knowledge (e.g. Haskel and Westlake 2017, 2018) and emphasis the role of firm boundaries in influencing incentives to accumulate intangible knowledge. For example, Target's CIO describes shifting away from outsourcing to the in-house provision of services to avoid:
“...a third-party provider sending [an advantage through shorter lead times] to Retailer B down the road.” Target had been outsourcing significant parts of its application development and back-end systems to India and domestic companies, including Infosys and IBM. As he told the Wall Street Journal, “We got to a stage where almost half the team is in third parties. It's unhealthy. By keeping the intellectual property generated by the in-house software engineers, the company can preserve competitive advantage.”
To assess this hypothesis, we develop a novel measure of proprietary firm knowledge using data on auxiliary establishments. Auxiliaries, such as R&D labs and warehouses, are service establishments that primarily serve other establishments of their firm and are therefore plausible creators and receptacles of firm knowledge. Auxiliaries are present in input service sectors and consistent with the premise that they focus on in-house provision of these services using high-skill labour; we find that they are smaller but pay higher wages relative to comparable establishments in the same sector that sell to external customers. In line with our hypothesis that these plants facilitate functional and sectoral structural change, we find that firm growth and reallocation across sectors increase with auxiliary employment shares, even after controlling for differences in firm size, age, and industry composition.
In the second part of the paper, we rationalise the relationship between within-firm structural change and firms' in-house provision of Professional, Scientific, and Technical Services in a model of heterogeneous, multi-product firms similar in spirit to Melitz (2003) and Bernard et al. (2010). Our framework features two key distinctions from past work. First, firms produce by combining two types of inputs – manufacturing and knowledge – which we assume are complementary. As a result, firms increase the level and share of knowledge inputs in response to a fall in their manufactured input costs. Second, firms that produce this knowledge in-house gain a sector-neutral productivity advantage. Although the model is static, one interpretation of firms' in-house knowledge investment is that it represents a stock of intangible capital that is non-rival within the firm and does not fully depreciate in response to changes in final-good demand or input prices.
The model shows how shocks to firms' output and input prices can induce within-firm structural change. An increase in competition in a firm's output markets represents a negative demand shock for firms' manufacturing output and thus decreases their manufacturing sales and their use of both manufactured and knowledge inputs. For firms with in-house knowledge workers, the output shock will thus decrease not only manufacturing, but also non-manufacturing employment. By contrast, an increase in competition in a firm's physical inputs lowers input prices and thus the cost of production. Since knowledge and manufactured inputs are complements, firms will increase both the level and share of knowledge inputs, though only the subset of firms with in-house knowledge workers increases their non-manufacturing employment. As these firms also have a relatively higher sector-neutral productivity, firms with in-house knowledge workers prior to the shock are also more likely to expand into new sectors that also use these non-rival inputs.
In the final part of the paper, we provide empirical support for our framework by examining US manufacturers' responses to plausibly exogenous variation in firms' output and input prices driven by China's global integration from 1997 to 2007. Our results confirm prior findings of large, negative effects of Chinese import competition in a firm's output markets on its employment and sales. These results are evident regardless of firms' auxiliary status and are strongly increasing in firm size. Not surprisingly, the output shock has a larger impact on firms' manufacturing activities. In fact, it only affects non-manufacturing employment at firms with auxiliaries and has no statistically significant relationship with non-manufacturing sales. This asymmetry supports the premise that firms with in-house knowledge workers use workers in their non-manufacturing establishments to support their manufacturing sales.
Results for the input shock are quite different. We see no statistically significant impact on firms' total or manufacturing employment. However, in line with our assumed complementarity between manufactured and knowledge inputs, we find a strong, positive relationship between the input shock and non-manufacturing employment for firms with auxiliaries: a ten percentage point increase in China's market share in the EU in a 500 employee firm's inputs raises its non-manufacturing employment by 19%. Unlike the output shock, we also find that the input shock increases firms' non-manufacturing sales, suggesting that functional structural change may induce sectoral structural change as firms that produce services in-house to support their manufacturing products begin selling these services to other firms. This result highlights a role for firm boundaries in aggregate structural change and in developing and preserving knowledge that cannot readily be sold in the marketplace.
Our results reignite past work that emphasises the role of firm boundaries in the production of knowledge. By following firms' transitions over time and demonstrating the importance of in-house input provision in response to shocks, we provide new support for the premise that firms play a key role in developing knowledge and using it to adapt to changing economic circumstances. We hope our findings will spur future work investigating what knowledge really is, how it is developed, and how it influences firm behaviour. Given the transition we document in the US towards ‘knowledge-related’ industries, work in these areas is likely to be increasingly valuable.
Our results relate to a number of policy debates. First, our finding that manufacturing firms account for a large share of non-manufacturing employment growth indicates that manufacturing increasingly encompasses activities that transcend traditional production worker occupations. The current policy focus on jobs in manufacturing establishments is therefore missing the considerable growth in manufacturing firms’ support activities, such as R&D. Second, the growth of manufacturing firms’ non-manufacturing activities suggests that US manufacturers contribute to US comparative advantage in skill-intensive occupations, both in terms of production and service activities. Third, the scope of manufacturing firms’ activities highlights the challenges associated with increased globalisation, in which US firms may be providing input services, such as R&D, that are particularly valuable when they can be combined with low-wage foreign labour. Our findings thus raise new opportunities and challenges for how to measure, regulate, and tax the increasingly important contribution of these intangible activities to the global economy.
Acemoğlu, D, D Autor, D Dorn, G Hanson and B Price (2014), “The Rise of China and the Future of US Manufacturing,” VoxEU.org, 28 September.
Autor, D H, D Dorn and G H Hanson (2013), “The China Syndrome: Local Labor Market Effects of Import Competition in the United States,” American Economic Review 103 (6): 2121–68.
Baumol, W J (1967), “Macroeconomics of Unbalanced Growth: The Anatomy of Urban Crisis,” American Economic Review 57: 415-26.
Bernard, A B, S J Redding and P K Schott (2010), “Multi-product Firms and Product Switching,” American Economic Review 100(1): 70-97.
Comin, D, D Lashkari and M Mestieri, (2021) “Structural Change with Long-run Income and Price Effects,” Econometrica 89(1): 311-374.
Ding, X, T Fort, S Redding and P Schott (2022), “Structural Change Within Versus Across Firms: Evidence from the United States,” NBER Working Paper 30127.
Duernecker, G, B Herrendorf and A Valentinyi (2018), “Structural Change and the Productivity Slowdown,” VoxEU.org, 16 May.
Haskel, J and S Westlake (2017), Capitalism without capital, Princeton: Princeton University Press.
Haskel, J and S Westlake (2018), “Productivity and Secular Stagnation in the Intangible Economy,” VoxEU.org, 31 May.
Matsuyama, K (2019), “Engel’s law in the global economy: Demand-induced patterns of structural change, innovation, and trade,” Econometrica 87 (2): 497-528.
Melitz, M J (2003), “The Impact of Trade on Intra-Industry Reallocations and Aggregate Industry Productivity,” Econometrica 71(6): 1695-1725.
Ngai, L R and C A Pissarides (2007), “Structural Change in a Multisector Model of Growth,” American Economic Review 97(1): 429-443.
Pierce, J R and P K Schott (2016), “Trade Liberalization and Mortality: Evidence from US Counties,” NBER Working Paper 22849.