98 years of economic insights for Indiana

The IBR is a publication of the Indiana Business Research Center at IU's Kelley School of Business.

Executive Editor, Carol O. Rogers
Managing Editor, Brittany L. Hotchkiss

Automation and offshoring in durable goods manufacturing: An Indiana case study

Director of Economic Analysis, Indiana Business Research Center, Indiana University Kelley School of Business


What accounts for the massive loss of employment in manufacturing over the last 15 years or so? Even while the U.S. economy recovered from the 2002 recession and grew through the mid-aughts, manufacturing employment continued to fall both in the U.S. and Indiana, but mildly bounced back after the Great Recession. Figure 1 provides visual context.

Figure 1: Manufacturing employment in the U.S. and Indiana


Source: U.S. Bureau of Labor Statistics QCEW data for all private employers

To help answer this question, we conducted something of a case study focusing on durable goods manufacturing in Indiana to ascertain whether manufacturing job losses in the state can be attributed to offshoring of labor (shifting production and employment to another country) or automation. Having suitable and sufficient data somewhat limits making iron-clad claims, but the available data and our analysis does point to whether offshoring and/or automation influenced (the mostly negative) employment outcomes in manufacturing in the state.

As the average Hoosier will be able to tell you, Indiana is the most manufacturing-intensive state in the United States. Employment concentration in manufacturing as a percent of the workforce is greater in Indiana than any other state. As with many of its Midwestern manufacturing-intensive neighbors, the steady rate of manufacturing job losses in the last decade and a half across the country is well documented. Indeed, announcements of foreign direct investment by Indiana-based companies that would create jobs in countries like China and Mexico add to the evidence and general unease about manufacturing jobs being offshored from Indiana to other countries. Table 1 presents Indiana-based company job announcements from the fDiMarkets data source.

Table 1: Announced jobs created offshore due to greenfield foreign direct investment by Indiana-based companies, 2010 to 2016

China India South Korea Mexico Japan Brazil Malaysia United Kingdom Canada Ireland Italy Singapore Total by Industry
Total by Country 5,380 3,892 3,715 3,205 2,917 2,017 1,725 988 720 670 510 304  
Commercial & institutional building construction (Simon Property) 1,356 3,295 2,712 1,481 1,648 474 10,966
Engines & turbines (Cummins) 1,620 1,312 420 416 205 427 380 21 5,514
Motor vehicle gasoline engines & engine parts (mostly Cummins) 600 676 263 410 2,716
Motor vehicle electrical & electronic equipment (mostly Cummins) 259 1,500 1,867
Medical equipment & supplies (Zimmer, Hill-Rom, Cook) 225 237 39 334 264 1,228
Pharmaceutical preparations (Lilly) 350 80 236 969
Paints, coatings, additives & adhesives (Uniseal) 125 711 918
General purpose machinery (Franklin Electric, Hillenbrand) 186 268 3 13 16 777
Biological products (except diagnostic) (Lilly mostly) 704 69 773
Other motor vehicle parts 758 758
Basic chemicals (Vertellus Specialties) 214 528 752
Medicinal & botanical (Lilly) 130 100 510 740
Software publishers, except video games (Interactive Intelligence)   100       26 77 16 230     19 580

Row and column totals include all countries (row) and all industries (column). Difference between row totals and the sum of industries is all other investment not reported due to small values.
Source: fDiMarkets.com

Table 2 presents the durable goods manufacturing industries in Indiana that suffered the greatest job losses from 1998 to 2015.

Table 2: Largest employment losses (on a numeric basis) in Indiana durable goods manufacturing, 1998 to 2015

NAICS Industry Change in total employment at annual average rate Change in industry headcount Employment level in 2015
3311 Iron and steel mills and ferroalloy manufacturing -2.3% -9,997 19,418
3315 Foundries -3.6% -9,172 9,921
3322 Cutlery and hand tool manufacturing -6.0% -1,047 537
3344 Semiconductor and other electronic component manufacturing -5.6% -6,911 3,948
3352 Household appliance manufacturing -14.2% -6,050 508
3353 Electrical equipment manufacturing -5.3% -4,983 3,119
3359 Other electrical equipment and component manufacturing -5.7% -4,331 2,454
3363 Motor vehicle parts manufacturing -2.4% -33,082 60,447

Note: These QCEW-based figures include all employment, both production and non-production.
Source: U.S. Bureau of Labor Statistics QCEW data

This article tees-up the empirical results of a publication that will soon appear in the journal Economic Development Quarterly, available online November 2018. Lest we “scoop ourselves,” the detailed findings are not presented here, but we can present the state of the discussion about offshoring versus automation, provide theoretical background and, we hope, whet the appetite of the reader to read the full article.

State of the discussion

The change in the geography of manufacturing employment has been a source of concern among many economists, policymakers and those in traditionally middle-class jobs who have found themselves displaced. Pierce and Schott (2016) link the stunningly swift decline of U.S. manufacturing jobs to the U.S. establishing permanent normal trade relations with China. By the later aughts, the concept of offshoring, and its empirical validation, had gone mainstream, with researchers attempting to assess the effects of offshoring on wage and income inequality, as well as determining how the increase in trade has affected productivity growth (Houseman, 2007).

The research literature generally points to the increasing practice of offshoring for the decline in manufacturing production and employment (e.g., Olsen, 2006). But service jobs were not immune from offshoring either (Bradford & Kletzer, 2005). Houseman and colleagues have a considerable opus related to offshoring, the importation of intermediate inputs, import prices and measures of productivity (e.g., Houseman et al., 2011). Acemoglu and colleagues (2016) also argue that even before the Great Recession, the U.S. was in an employment sag and suggest job losses due to Chinese import competition to be in the range of 2.0 million to 2.4 million.

With such job losses, the gains from trade seemed illusive, or at least those gains were inequitably distributed. States like Indiana were in turmoil as they tended to bear the disproportionate impact of jobs being relocated.

There has also been emphasis on the types of jobs lost, as well as the potential for certain occupations to be offshored or automated away. The focus of this research is the types and level of skills associated with at-risk occupations. While manufacturing jobs moving to lower-wage countries was typical and had higher visibility in the opening decade of the century, there has been an increasing focus on routine service jobs moving to lower-skill countries, or even high-tech jobs that can be moved to lower-cost locations (for example, tech support provided by workers in India). Indeed, there is a growing literature on how to score a job’s vulnerability to offshoring, see Autor and colleagues (2003), Blinder (2009) and Blinder and Krueger (2013). For a follow-on to this discussion about a job’s vulnerability to being offshored shifting to the vulnerability of jobs being replaced by automation and computerization, see, for example, Frey and Osborne (2017). Just last year, our fellow Hoosier researchers at Ball State University published a report and supporting data—How Vulnerable Are American Communities to Automation, Trade, & Urbanization?—that synthesizes the scholarly work mentioned above and makes some rather grime predictions about the vulnerability of American jobs (Devaraj et al., 2017).

The basic takeaway from the Ball State work is that both forces—offshoring and automation—are in play in the massive occupational realignment due to hit the U.S. economy, but automation is dominant (Wells, 2017). This assessment, that automation is the greater threat, is consistent with their earlier work placing the loss of manufacturing employment on increases in productivity (Hicks & Devaraj, 2015). In this study, they disaggregate industries to their three-digit NAICS detail and show the differences in the productivity levels in the years of 2000 and 2010 to explain the job losses. Their conclusion attempts to reconcile large increases in productivity with, in their view, the relatively low effects that trade exerted upon those productivity gains. Trade, however, can influence the demand for domestically produced final goods, as well as the purchase of imported intermediate inputs—the latter of which they rightly note is not well captured in the data.1 One is left with the sense that productivity increased, but the source of those gains is still something of a mystery.

It becomes an even greater mystery when one considers the argument that those gains arise not from intermediate input production offshoring, but from automation and increased application of capital to replace labor in order to make labor more productive. The question then becomes, where did those machines come from? Investment in private fixed capital has not been impressive. According to the U.S. Bureau of Economic Analysis, from 1998 to 2016, current-dollar fixed private investment in nonresidential equipment, information processing equipment and industrial equipment increased at an average annual rate of 2.6 percent, 1.3 percent and 2.3 percent, respectively. In contrast to the period of 1988 through 1998, the current-dollar increases were 7.1, 7.6 and 6.2 percent, respectively.2 In short, there wasn’t much of a capital investment boom through most of the recent period.3

Moreover, the boom in robotics investment has only begun in earnest since 2014.4 And the location of those investments may be in dispute. The Brookings Institution assessed the cities with the greatest exposure to robots. Three of the top five cities were in Indiana. In contrast, Leigh and Kraft (2017), using a different methodology to assess a region’s concentration of robotics firms and systems integrators, found Chicago and Indianapolis as having relatively low automation concentrations. At this point, it is difficult to find a unanimity of opinion as to the precise location of robotics investment other than anecdotal evidence found in news articles.

There is also the issue of differentiating between technology and automation. Are they the same? Are they complements? How then, is labor replaced by capital? The question is a little more nuanced as automated machines, like labor, are bound by their location. Should we not be able to observe the effects of the substitution of capital/automation for labor in economic data/statistics, especially geographically defined data? Waldman (2016), using survey-related data on automation investment, suggests that the productivity-enhancing results of automation investment are slow and that those results, perhaps much like the productivity gains that eventually appeared with the application of computers, would become evident in the future. In short, Waldman asserts that there is not much to see here—at least not yet.

Given the modest increases in investment in capital stock over the period, how does one make the argument that productivity gains are mostly attributed to automation, robots and technology?

The state of the discussion, or argument, could be summarized this way. The effects of trade and, more specifically, the effects of offshoring as evidenced by the closure of manufacturing plants in the U.S. and the movement of operations elsewhere shifted the political and economic discussion from the benefits of trade (cheaper goods) to the detriments of trade (the loss of jobs). The political and economic welfare discussion then shifted. Automation emerged as the possible, if not probable, culprit for the loss of jobs. Economic researchers began to explain the decline in manufacturing employment, not as a negative outcome of trade, but as linked to automation. Jobs were being lost, but it wasn’t freer trade that moved production activity (and jobs) from one country to another. Rather, it was the productivity-enhancing benefits of capital investment in automation and technology on the factory floor that was identified as the driver to worker redundancy.

Given the high stakes for the average American worker, and the concerns of policymakers, politicians and business owners, there is scarce empirical and peer-reviewed proof of whether offshoring or automation is the leading cause of manufacturing job losses. Very recently, even as we were concluding our analysis for the forthcoming EDQ publication, Houseman (2018) made some sobering conclusions:

The prevailing view that automation largely caused the swift relative and absolute declines in U.S. manufacturing employment in the 2000s reflects a misinterpretation of the numbers. Moreover, the automation view is not backed by rigorous research … [W]hile industrial robots may have the potential to displace many workers in the future, any effects on manufacturing employment to date are small.

A large and growing body of research has also examined the effects of trade on domestic manufacturing in the 2000s. No study captures all aspects of globalization and its effects on manufacturing and aggregate employment, and the limitations of any individual study need to be recognized. Collectively, however, the research points to sizable adverse effects from trade on employment, output, and investment. (p. 4)

Houseman’s national scope, data and method differed significantly from our own. It was our hope that our approach highlighting one state would add clarity to the question of offshoring versus automation and begin to bring the source of manufacturing productivity gains into greater focus, especially for Indiana.

Our hypotheses

While often reported as a broad, monolithic sector, manufacturing is comprised of many industries that are categorically different in terms of production technologies, material inputs, and capital and workforce requirements. These industries are subject to different market forces, demand shocks, regulations, supply considerations and management concerns. Offshoring and/or automation will affect industries within the broad manufacturing sector differently.

Our study investigates the changes in an industry’s production function—the combination of labor, capital and intermediate inputs. The change in the mix of these inputs and factors of production can provide some evidence of whether changes in industry employment is driven by automation (i.e., substituting capital for labor) or by offshoring.

There are at least two types of offshoring. Type 1 offshoring moves production activities of final goods offshore, for example, producing plastic toys in China. The analysis of type 1 offshoring is not within the scope of our forthcoming paper at it relates more to final demand and the last stage of production. Type 2 offshoring, in contrast, is to increasingly source intermediate inputs and parts from other countries for latter-stage production activities. For example, rather than an automobile company producing its own tie rods, it procures them from suppliers in Canada.

Our first hypothesis is: Type 2 offshoring is indicated if the value of shipments increases, and the proportion of intermediate inputs to shipments increases but employment decreases.

Our second hypothesis is: Automation is in evidence if the value of shipments increases, along with an increase in the ratio of capital to labor and employment decreases.

There is also something of a third hypothesis related to workers that were not offshored (or those left behind). It relates to the so-called Stolper-Samuelson theorem that states that the competitive pressures associated with trade and offshoring depress wages. However, we expect average wages to rise as a result of offshoring. As firms make workers redundant, either by offshoring or by automation, the average wage will rise for the remaining workers. As firms relocate production activities and reduce their domestic labor force, the productivity of remaining workers—their output per worker—is expected to increase. A recent example of this was cited as a reason clothing manufacturing has experienced a larger-than-average uptick in annual earnings.5

The third hypothesis may be answered using administrative record data from the Indiana Department of Workforce Development. These highly confidential data are at the individual plant, or establishment, level, so one can determine whether those workers who remain after a large layoff event are, on average, earning more. (These data are anonymized before we analyze them, in keeping with data and employer confidentiality.)

A brief excursion on offshoring versus nearshoring

It may be suggested that rather than intermediate input production being offshored to other countries, the manufacturing of those inputs may have simply moved to another state, so-called nearshoring. If intermediate input production was shifted to other states, in contrast to being imported, then one would expect that Indiana’s experience of increasing material consumption associated with declining production employment would differ from the experience for a particular industry in other states. In other words, for a particular industry, was Indiana’s experience different from other states with a dominant presence in a particular industry?

We selected several industries that may have served as a counterfactual for the Indiana case study example. The results for those industries and states were largely similar to Indiana. There were no stand-out industries that would lead one to think that Indiana’s experience ran counter to the national trends. Granted, the comparisons, industry by industry and state by state, were not comprehensive, but the comparisons showed that Indiana was not a curious outlier. In the three selected industries, Indiana’s experience—increasing share of the value of intermediate inputs and decreasing share of production labor—generally matched other states.

In the case of motor vehicle parts manufacturing, Tennessee, Kentucky, Alabama and Georgia gained employment, which could be evidence of nearshoring. That said, it is difficult to build an argument for nearshoring accounting for the majority of job losses in the other 16 of the top 20 states (by 2016 production employment) given the four-state job gains totaled about 30,000 in contrast to the 220,000 jobs lost in states like Indiana, Michigan and Ohio. In summary, there may have been some nearshoring, but that employment shift is relatively trivial compared to offshoring.


As stated from the start, this article is to tee-up and promote the upcoming article in Economic Development Quarterly. The question about whether the hemorrhage of manufacturing jobs in the U.S. over the last 15 years is primarily due to offshoring or automation is not settled. It is our hope that this presentation has fairly and briefly discussed the literature and the arguments for attributing the job losses to either automation or offshoring. At the time of this writing, there has not been watertight arguments or peer-reviewed analysis either way. A watertight case is a very high bar, but we hope that the forthcoming journal article makes a sufficiently robust case that for Indiana, offshoring contributed mightily to the loss of manufacturing employment over the last 15 years.


  1. See also Houseman et al. (2011) for a thorough discussion of the issues regarding the importation of parts and intermediate material consumed and how exchange rates and price adjustments make the effects of lower-cost inputs difficult to measure.
  2. Table 5.5.5. Private Fixed Investment in Equipment by Type and Table 5.5.6. Real Private Fixed Investment in Equipment by Type, Chained Dollars (BEA: https://www.bea.gov/national/nipaweb/DownSS2.asp)
  3. Chain-dollar rates are significantly different, especially in information equipment, due to deflator adjustments attributed to quality changes, so-called hedonic measures.
  4. See Robots R’ Us: Funding and deal activity to robotics see new highs in 2015. (2016, March 23). CB Insights Research Briefs. Retrieved from https://www.cbinsights.com/research/robotics-startups-funding/. See also IRF International Federation of Robotics. (2017, September 27). How robots conquer industry worldwide [Presentation]. Retrieved from https://ifr.org/downloads/press/Presentation_PC_27_Sept_2017.pdf
  5. Isidore, C. (2018, February 2). Here’s who’s getting a raise these days. CNN Money. Retrieved from http://money.cnn.com/2018/02/02/news/economy/jobs-report-wage-increases-raises/index.html


  • Acemoglu, D., Autor, D., Dorn, D., Hanson, G. H., & Price, B. (2016). Import competition and the great U.S. employment sag of the 2000s. Journal of Labor Economics34(S1), S141-S198.
  • Autor, D. H., Levy, F., & Murnane, R. J. (2003). The skill content of recent technological change: An empirical exploration. The Quarterly Journal of Economics118(4), 1279-1333.
  • Blinder, A. S. (2009). How many US jobs might be offshorable? World Economics10(2), 41.
  • Blinder, A. S., & Krueger, A. B. (2013). Alternative measures of offshorability: A survey approach. Journal of Labor Economics31(S1), S97-S128.
  • Devaraj, S., Hicks, M. J., Wornell, E. J., & Faulk, D. (2017). How vulnerable are American communities to automation, trade & urbanization? Center for Business and Economic Research, Ball State University. Retrieved from https://projects.cberdata.org/123/how-vulnerable-are-american-communities-to-automation-trade-urbanization
  • Frey, C. B., & Osborne, M. A. (2017). The future of employment: How susceptible are jobs to computerisation? Technological Forecasting and Social Change114, 254-280.
  • Hicks, M. J., & Devaraj, S. (2015). The myth and the reality of manufacturing in America. Center for Business and Economic Research, Ball State University.
  • Houseman, S. (2007). Outsourcing, offshoring and productivity measurement in United States manufacturing. International Labour Review146(1‐2), 61-80.
  • Houseman, S., Kurz, C., Lengermann, P., & Mandel, B. (2011). Offshoring bias in U.S. manufacturing. The Journal of Economic Perspectives25(2), 111-132.
  • Houseman, S. (2018). The decline of U.S. manufacturing employment—Automation and trade. Employment Research 25(2): 1-4. https://doi.org/10.17848/1075-8445.25(2)-1
  • Jensen, J.B., & Kletzer, L. G. (2005). Tradable services: Understanding the scope and impact of services outsourcing. Peterson Institute for International Economics Working Paper, 5-9.
  • Leigh, N. G., & Kraft, B. R. (2017). Emerging robotic regions in the United States: Insights for regional economic evolution. Regional Studies, 1-13.
  • Olsen, K. B. (2006). Productivity impacts of offshoring and outsourcing. STI Working Paper 2006/1. Retrieved from http://www.oecd.org/science/sci-tech/36231337.pdf
  • Pierce, J. R., & Schott, P. K. (2016). The surprisingly swift decline of US manufacturing employment. The American Economic Review106(7), 1632-1662.
  • Waldman, C. (2016). The evolving contours of productivity performance and automation investment in U.S. manufacturing. Business Economics51(4), 213-238.
  • Wells, N. (2017, July 19). Half of American jobs are at risk from automation, new study suggests. CNBC. Retrieved from https://www.cnbc.com/2017/07/19/half-of-american-jobs-are-at-risk-from-automation-new-study-suggests.html