In May 2013 shareholders voted to break up the Timken Company—a $5 billion Ohio manufacturer of tapered bearings, power transmissions, gears, and specialty steel—into two separate businesses. Their goal was to raise stock prices. The company, which makes complex and difficult products that cannot be easily outsourced, employs 20,000 people in the United States, China, and Romania. Ward “Tim” Timken, Jr., the Timken chairman whose family founded the business more than a hundred years ago, and James Griffith, Timken’s CEO, opposed the move.

The shareholders who supported the breakup hardly looked like the “barbarians at the gate” who forced the 1988 leveraged buyout of RJR Nabisco. This time the attack came from the California State Teachers Retirement System pension fund, the second-largest public pension fund in the United States, together with Relational Investors LLC, an asset management firm. And Tim Timken was not, like the RJR Nabisco CEO, eagerly pursuing the breakup to raise his own take. But beneath these differences are the same financial pressures that have shaped corporate structure for thirty years.

Urging Timken shareholders to vote for the split, Relational Investors argued that they should want “pure-play” companies, focused on a single industrial activity. Investors would then be free to balance their portfolios by selecting businesses in industrial sectors with varying degrees of risk and sensitivity to different phases of economic cycles. A firm such as Timken—about one-third a steel company (a materials play) and about two-thirds a bearings and power transmission business (an industrial components play)—would lock investors into a mix that, Relational Investors claimed, leads to a discount on share price.

Timken management argued that making both materials and products enabled them to bring to market higher-quality goods that met customers’ needs: for example, their ultra-large bearings for windmill towers, which measure two meters in diameter, weigh four tons, and have to stand up to extreme wind and temperature conditions. Controlling the entire value chain, they said, allowed them to fine-tune the attributes of the steel in order to make superior products. Nonetheless, the financial calculation about how to maximize quarterly returns won out.

Timken’s story is not only about stock prices and product quality. Since the 1980s financial market pressures have transformed U.S. corporate structure itself. The system was once dominated by a few dozen very large, vertically integrated firms in which most or all of the functions needed to take a new idea about a product or a service to market—from R&D, design, manufacturing, testing, and logistics through sales and after-market services—were contained within the four walls of a single corporation. Now even the big firms are smaller, leaner, and centered on “core competencies,” with much of their production outsourced and overseas. Those pressures have driven companies such as Timken to hive off activities that involve heavy capital outlays, require large workforces, or promise less profitability in the short term.

The contribution of this decades-long trend to the rapid decline in American manufacturing has not been fully acknowledged. But understanding its role is essential to a revival of American manufacturing that will create jobs and promote long-term economic health. Both private and public sectors are taking constructive steps. Will American finance get in the way?

Timken was one of hundreds of manufacturing companies in a sample of firms interviewed by an MIT research team about their experiences in bringing novel ideas, products, and processes to market. I was a principal investigator in that project. The aim of the project—Production in the Innovation Economy—was to discover whether we really need manufacturing to gain the benefits of innovation: economic growth, new companies, new profits, and good new jobs in this country. After all, Apple and other companies like it—which do R&D, design, and distribution but little or no production in the United States—reap the lion’s share of their profits here. To explore these issues, the MIT researchers collected data on the efforts to scale innovation up to market by startup firms, Main Street small and mid-sized manufacturers, and Fortune 500 companies. When we learned about the Timken shareholders’ vote, we realized that we were seeing up close and in real time the forces that over the past thirty years have transformed and shrunken manufacturing in the United States.

In the radical downsizing of American manufacturing, changes in corporate structures since the 1980s have been a powerful driver, though not one that is generally recognized. Over the first decade of the twenty-first century, about 5.8 million U.S. manufacturing jobs disappeared. The most frequent explanations for this decline are productivity gains and increased trade with low-wage economies. Both of these factors have been important, but they explain far less of the picture than is usually claimed.

Since the 1980s, financial market pressures have driven companies to hive off activities that sustained manufacturing.

In the case of productivity gains, there have clearly been major advances over the long term. The periods of advance, however, do not correlate neatly with the years of greatest job losses. Some periods of rapid productivity growth also saw employment growth, or at least of stability in the manufacturing workforce. Research by economist Susan Houseman and her colleagues on the past decade—when manufacturing employment has fallen off the cliff in the United States—suggests productivity growth in manufacturing was actually modest, but national statistics fail to account for the rising volume and value of imported components and thus systematically overstate productivity. The corrected numbers indicate that productivity growth over the past decade took place primarily within one manufacturing industry: computers and electronics. So productivity gains alone can hardly explain the shrinking of U.S. manufacturing employment.

Nor can trade and outsourcing to low-wage countries account for manufacturing declines that began well before China became a factor. Today there is overwhelming evidence for the powerful effects of globalization and trade on manufacturing employment, particularly in those regions home to labor-intensive industries such as apparel and furniture, which were heavily exposed to import competition. But the big impact of low-wage imports hit in the United States only after the entry of China into the World Trade Organization in 2001. Before then the proportion of imports coming from low-wage economies relative to imports from high-wage economies was very low. Economists generally found it too small to explain manufacturing job losses. Even taking into account the outsourcing of existing jobs as well as import competition, it was implausible a decade ago to attribute a significant share of job destruction in manufacturing to these causes. Department of Labor surveys found that outsourcing and offshoring jobs represented less than one percent of all layoffs in 2003, and two percent in 2004. These surveys did not catch the total volume of jobs being transferred overseas, but in all estimations those jobs amounted to a small fraction of total annual job turnover in the American labor market.

To better understand the decline of American manufacturing, we need to go back well before the last decade to see how changes in corporate structures made it more difficult to scale up innovation through production to market.

In the 1980s about two-dozen large, vertically integrated companies such as Motorola, DuPont, and IBM dominated the American scene. With some notable exceptions (for example, GE), large vertically integrated companies today have pared off activities and become not only smaller but also more narrowly focused on core competencies. Under pressure from financial markets, they have shed activities that investors deemed peripheral—such as Timken’s steel.

This process has been fostered by great technological advances in digitization, which have allowed companies to outsource and offshore many of the functions they previously had to carry out themselves. In the 1970s a Hewlett-Packard engineer who designed circuits for a new semiconductor chip had to work together with a technician with a razor blade to cut a mask to place on silicon. Now the engineer can send a complete file of digital instructions over the Internet to a cutting machine. The mask and the chip fabrication can take place in different companies, anywhere in the world. A senior executive of Cisco told MIT researchers:

The separation of R&D and manufacturing has today become possible at a level not even conceivable five years ago. Progress in technology allows us to have people working anywhere collaborating. We no longer need to have them located in clusters or centers of excellence. We now have the ability to sense and monitor what’s going on in our suppliers at any place and any time. Most of this is based on sensors deployed locally, distributed control systems, and new middleware and encryption schemes that allow this to be done securely over the open Internet. . . . In other words, not only do we monitor and control what’s happening inside a factory, but we’re also deeply into the supply chain feeding in and feeding out of the factory.

Digitization and the Internet continue in multiple ways to enable the fragmentation of corporate structures that financial markets demand.

The breakup of vertically integrated corporations and their recomposition into globally linked value chains of designers, researchers, manufacturers, and distributors has had some enormous benefits both for the United States and for developing economies. It has meant lower costs for consumers, new pathways for building businesses, and a chance for poor countries to create new industries and raise incomes.

But the changes in corporate structures that brought about these new opportunities also left big holes in the American industrial ecosystem. These holes are market failures. Functions once performed by big companies are now carried out by no one.

Vertically integrated companies used to provide semi-public goods through apprenticeships, basic research, funding to bring innovation to scale, and diffusion of new technologies to suppliers. The resulting spill-overs into the economy were enormously important: they subsidized community college education, provided job training, and more generally created an industry-wide ecology that fed job creation and growth. Companies such as Alcoa, AT&T, DuPont, and Xerox used to support long-term R&D in facilities such as Bell Labs. Over the past twenty years, those laboratories have been shuttered or greatly reduced in size and scope. As companies downsized, they could no longer, or would no longer, keep these activities in house or pay for them.

Today companies harness R&D to specific business divisions and to near-term product development. Most basic and precompetitive research starts out in public and private laboratories isolated from manufacturers. When cutting-edge innovations come out of such laboratories, it is not clear where to find capital to bring their new products and processes to market. Who will handle prototyping, pilot production, testing, and large-scale commercialization? When DuPont brought nylon into mass production in the 1930s and 1940s, it could draw on its own retained earnings as well as established relations with investors, bankers, factories, and suppliers. Today’s innovative small company lacks virtually all of these resources. Investors excel in providing venture capital funding for startup companies, but once these companies reach the stage of commercialization and venture capital is no longer available, they find few financial backers. Now that investors have curbed their appetite for startups going public, acquisition by big companies and recourse to foreign capital seem to be the main avenues for bringing to market the innovations that begin life in university and public laboratories. Both of these routes have troubling implications for American innovation and jobs. When big companies acquire startups, the MIT researchers found, much of the dynamism and promise of the new technology can be lost in the process of integration. When commercialization takes place outside the United States, opportunities to learn about scaling new technologies are foregone. Over time, it becomes more likely that innovation will shift to places where companies have more experience with scale-up and commercialization.

The loss of apprenticeships and job training are similarly problematic. Today’s companies do not intend to keep employees in lifetime careers, so they no longer invest in the skills they nonetheless need from employees. This not only creates challenges for companies that once could rely on workers trained in house; it has a negative effect throughout the economy. Even in the days of long job tenure, significant numbers of workers who started in large firms moved around, so big-company investments in training benefited the broader industrial community.

More generally, with the shrinking of publicly available resources once generated within big companies and with the disappearance of many suppliers as production moved offshore, it has become more difficult for all manufacturers to move new products and processes to market. It is not only startups that face the hurdle of finding resources to substitute for the capabilities and coordination once provided within the four walls of the vertically integrated corporation. For established small and mid-sized manufacturers, too, the process of commercialization has become more challenging. To bring a new product or process to life, a firm needs to find additional capital, skills, suppliers, and possibly expertise. Today small and medium-sized companies find little that they might combine with their own resources. They drip in money from retained earnings and develop their products in small increments. Even promising new projects move ahead haltingly, and few jobs are created.

The depletion of the industrial ecosystem is hardly inevitable in advanced high-wage economies. Consider Germany. In 2010 manufacturing employed 22 percent of Germany’s workforce and contributed 21 percent of value-added to GDP. In the same year in the United States, just under 11 percent of the workforce was employed in manufacturing, which contributed 13 percent of value-added to GDP. Many factors explain the difference, including the role of government in supporting innovation and Germany’s education system. But, importantly, German companies operate in ecosystems rich in research consortia, Fraunhofer Institutes, specialist suppliers, technical universities, apprenticeship training, and local and regional banks.

This ecosystem supports innovation and commercial production through diverse mechanisms. Through their connections to larger firms and to each other, small and medium-sized companies are able to obtain information that would otherwise be difficult to access. In research consortia, companies come together to discuss road maps for new technologies and to work collaboratively on them. The consortia build equipment and facilities to be used by multiple firms that could not afford to buy these themselves. They distribute public funds through competitions.

In such an environment, German companies find valuable complements to their own legacy strengths when they move into new domains. In contrast, small and mid-sized U.S. manufacturers are home alone when they try to bring innovation to market.

There is no returning to the corporate structures of fifty years ago. The question is how to fix the holes in the U.S. industrial ecosystem, given that we now have companies with smaller domestic footprints who face powerful and agile competitors around the world. In a globally distributed economy, how can we generate the coordination, knowledge diffusion, resources for scaling-up, and leadership that could accelerate and deepen the flow of innovations into the market? The MIT researchers looked for initiatives promising complementary capabilities—skills, expertise, finance, services, technologies—that can be drawn on by multiple firms in diverse sectors. The cases we identified are experiments in rebuilding the American industrial ecosystem. No one of these models would fit every situation, though they do have common features of working to build trust among partners, reduce and spread the financial risks of innovation, and convene private and public actors.

The depletion of the industrial ecosystem is hardly inevitable in advanced high-wage economies.

A number of these initiatives have been spearheaded by state and federal government. The Obama administration’s new manufacturing innovation institutes form one such initiative. The federal government announced a competition in which it would invest $30 million for a center to work on additive manufacturing (3d printing) technologies. A dozen groups made up of universities, companies of all sizes, and economic-development organizations competed for the center, which eventually was granted to a Northeast Ohio/Southwest Pennsylvania team and was set up in 2012 in Youngstown, Ohio. The private sector has now contributed more than $60 million to this project. Fifteen more manufacturing innovation institutes have been announced. On the state side, the Massachusetts Clean Energy Center, for example, acts as a convener and coordinator in the renewable energy sector, bringing universities, turbine makers, energy storage device makers, the Massachusetts Maritime Academy, and others together in new projects. None of these particular parties had worked together before. MassCEC also invested $18.2 million in state funding to secure an additional $26.7 million in federal Department of Energy funds to build a world-class wind-blade testing facility.

While public-driven efforts to rebuild the industrial ecosystem are the most visible, the MIT team also recognized that private sector companies were taking steps to create resources that could be used in combination with others. Indeed in the sample of firms we interviewed, Timken stood out as one of the companies most engaged in such initiatives. Before the shareholder-driven breakup, Timken and Stark State College of Technology in North Canton, Ohio were building a test center for ultra-large bearings for use in wind turbines. Students would get hands-on experience in the center, which would also be used by Timken for its own product testing. At the University of Akron, Timken took the lead in expanding research and degree programs on engineered surfaces. Timken transferred its coating laboratory equipment and staff to the university and initiated an industrial consortium that other firms could join—on condition that they too contribute to the programs. Although coatings are vital to Timken’s products, they were essentially a “stranded technology” in a company whose business lines are bearings and special steels. At the new industrial consortium at the University of Akron, coatings research can be developed into startups; one has already emerged. Timken also became one of the most active promoters of the coalition that eventually won the first federal manufacturing innovation institute grant in additive manufacturing.

Timken obviously has a business interest in these initiatives: sharing costs for activities that were once borne entirely in-house, educating students who can be recruited for employment, gaining eligibility for new state and federal grants. But in fulfilling its own goals, Timken was also acting as a convener for industry and education. It placed its own resources on the table in order to attract others to do the same. As Timken prepares to split into two smaller publicly listed companies, how likely is it that either of them will be so active in strengthening the Ohio industrial ecosystem? Judging by the records of other companies that have gone through similar restructuring, I am not optimistic.

With domestic shale gas and oil exploitation pushing energy prices down, rising wages in China, and new corporate realism about the costs involved in outsourcing and offshoring, there is a more favorable climate for manufacturing in the United States today than there has been in decades. Yet these changes are unlikely to have durable effects if the basic weaknesses of the system are not repaired.

The new public-private partnerships to rebuild capabilities in the industrial ecosystem seem to have enormous promise. But, as the Timken case illustrates, the financial pressures that broke up American companies in the ’80s persist.

The solution may be out of reach. Today California teachers need to protect their pensions by dismantling Ohio manufacturers. The structures of U.S. capital markets and fiscal policy reward investors whose decisions are based on maximizing returns over the short-term. While the Dodd-Frank financial reforms may cut down on some of the riskiest securitization-based investment strategies, new regulations have not created real incentives for the more patient investment that growing production in America requires.

Photograph: Timken Company.