The Third Pillar

by Raghuram Rajan

  Apple, for example, has produced internationally from its early days, but had a manufacturing presence in the United States until 2004, when it closed its last US manufacturing facility. It then proceeded to exit manufacturing entirely. Apple is one of the most profitable companies in the world despite manufacturing virtually nothing. The reason for its success, quite simply, is that it holds on to everything outside manufacturing, including research and development, design, content (including its profitable iTunes store and the apps that are made for its products), marketing, and finance. Let me repeat an oft-cited example here, oft-cited because it makes the point so clearly: The Apple iPhone XS Max costs Apple about $390 to make, and is sold to retail buyers for $1,250, a price that is more than three times its manufacturing costs.12 Most of the final manufacturing is done in China by firms like Foxconn, but much of the profit is retained by Apple as compensation for the intellectual property and software platform it has created. More generally, even as they outsource the low-value-added manufacturing segments of the production chain, developed countries retain the high-value-added and profitable premanufacturing segments like R&D and the equally profitable post-manufacturing segments like marketing and finance.13

  Such a division of labor is not unattractive to emerging markets that are trying to move up the complexity chain to build more technology-intensive products. Since the early 1990s, the possibility of participating in global supply chains has convinced a number of these countries to lower their tariffs, improve their business environment, sign treaties protecting foreign investment, and enhance their protection of intellectual property. This has eased the way for more segments of the value chain to be moved to emerging markets.

  Even while developed-country firms are outsourcing manufacturing to the emerging markets, emerging-market firms have been relying on developed-country firms for R&D and design. Many manufacturers of generic drugs in the emerging markets reverse engineer drugs that go off patent after they are initially made by developed-country firms. Indeed, the Indian pharmaceutical firm Cipla developed an anti-HIV three-drug cocktail, which it offered to poor African countries and AIDS groups at a dollar a day in 2001, which was one-thirtieth of its then price.14 Pharmaceutical firms around the world were forced to lower prices, and AIDS treatment has become affordable everywhere. Having built their business imitating, emerging-market firms in China and India are now contributing to original drug research.

  Who wins and who loses in this process? Highly educated and creative designers, scientists, and engineers, as well as advertising and marketing mavens now have a world market. Initially, this has favored the highly educated and skilled in developed countries, much as has technological development. As emerging markets train their own people well (with students often finishing with an advanced degree from a developed country), capabilities are migrating to the rest of the world. The highly educated and skilled everywhere now compete for business from global supply chains. In some countries, their wage differential relative to others, while high, is plateauing or even falling.15 The losers are clearer: the moderately educated workers in developed countries. When the supply chain was entirely in the developed country, they benefited from the competitive edge that developed-country design or R&D gave them. Their jobs were safe, protected by the indivisibility of the production process, which allowed them to bargain for higher pay, lower and more predictable work hours, and more safeguards at work. As the production process fragments, though, they have been exposed to the full force of competition from cheaper, more flexible, but equally competent labor elsewhere. Of course, consumers everywhere, like the AIDS patients who now have access to cheap drugs, benefit from more competitive and efficient production.

  As communications and information technology improves, more and more service value chains will be subject to the same competitive scrutiny as manufacturing value chains. Providers will reexamine what can be outsourced and what ought to be retained. As economist Alan Blinder has argued, all impersonal services that can be delivered electronically at a distance, with little or no degradation in quality, are potentially vulnerable.16 What will be harder to replace are human creativity, customization, and human empathy.

  JOB LOSSES DUE TO TRADE AND THE EFFECTS ON THE COMMUNITY

  Let us look at job losses in a developed country, specifically the United States, more carefully. Job losses can be a sign of a dynamic free enterprise economy, not necessarily evidence of an economy in decline. Around 40 percent of all US workers were in agriculture at the beginning of the twentieth century, while only 2 percent were thus occupied at the end of the century, but the 2 percent produced significantly more than did the earlier 40 percent. Similarly, with all the talk of the United States losing competitiveness, few realize that employment in US manufacturing peaked in 1944 at 39 percent of the labor force and has been on a steady decline since then to 8.5 percent of the labor force in 2017 (this probably overstates the decline since aspects of manufacturing ranging from product design to factory cleaning are now done by specialized outside firms and relabeled manufacturing-related services). Nevertheless, the share of real US GDP coming from manufacturing has not fallen in the last fifty years.

  This means that manufacturing activity has not migrated away from the United States, at least not in the aggregate, but it has become more productive and high tech. Certainly, smokestack industries like iron and steel have moved to developing countries like China, as has furniture making, but in their place the United States has cleaner factories in more technologically advanced fields such as aircraft or communications equipment. However, fewer workers are classified as being strictly in manufacturing—5.8 million fewer from 1999 to 2011 alone. Those who remain are typically more skilled or qualified workers, because more of the output now comes from technologically advanced sectors.17

  How many of these job losses come from increased imports, especially from China, which increased its share of world manufacturing value added from 4.1 percent in 1991 to 24 percent in 2012? A recent study by economists Daron Acemoglu, David Autor, David Dorn, Gordon Hanson, and Brendan Price estimates that the direct effects of Chinese imports on job losses—such as a furniture factory closing down in North Carolina because the firm now imports from China—is only about 10 percent of US manufacturing job losses over the period 1999 to 2011, which spans China’s great export boom.18 When the losses in output of US firms that buy from, or sell to, the now-closed factory are added in, imports from China can account for about 18 percent of job losses in manufacturing. While estimates of the sources of job losses are by their very nature imprecise, the evidence suggests that increases in productivity through automation and computerization, which account for the majority of the remaining job losses, have been the larger source of job losses. So why is there so much more public anxiety about trade?

  The job losses due to trade have been more centered in low-tech manufacturing with well-paying unionized jobs. Such establishments have typically been located near smaller towns, such as the US Steel plant in Granite City, Illinois, and rural areas in the interior of countries, where the cost of living and thus of labor has been low, and regulation light. These establishments have dominated the local community, providing the incomes that keep the local hairdressers, laundries, and shops in business. If unable to keep up with competition from imports, the establishments close down local operations, and may move machinery to a country where labor is cheaper. Since manufacturers in an industry cluster together, they are likely to decide to lay off workers or close down at similar times, compounding the magnitude and impact of the job losses. Of the 1,250 workers represented by the steel workers union in Granite City, only 375 were working at the end of 2016.19 As described by Amy Goldstein in her book Janesville, which follows the Janesville community after General Motors closed a large plant there, the effects on the community can be devastating.

  In contrast, the job losses due to greater automation and computerization have bee
n spread across manufacturing and services, and typically have hit firms that are more likely to be located near urban areas. Moreover, instead of the whole factory or office closing, a few workers doing routine jobs that can be automated are let go periodically. The remaining workers doing nonroutine work continue to be employed, and typically now are more productive. Higher productivity allows their employer to lower prices, sell more, and hire more workers in nonroutine jobs to meet the increased demand. The demand of all these workers for local goods and services such as haircuts and dry cleaning increases, creating new local service jobs that offset the lost jobs.

  Indeed, a study that separately examines the effects of trade competition and technological progress in the United States finds stark differences in their effects on jobs in the local community.20 Over the period 1990 to 2007, manufacturing-dependent communities that are hit by trade competition see a significant fall in employment across all jobs—routine and nonroutine, skilled and non-skilled. They also experienced an increase in the share of local unemployed, and an increase in working-age workers who leave the labor force permanently. In contrast, manufacturing-based communities that are most prone to automation saw little overall job losses in the local labor market over the period.

  This does not mean there were no effects on jobs due to automation. As one might expect, the study finds automation reduced the number of jobs in routine activities like production, assembly, clerical and administrative support. However, the study finds the loss of routine jobs was offset by an increase in nonroutine skilled jobs such as management, professional services, and technical personnel, as well as an increase in nonroutine jobs for the moderately skilled, such as automobile repair and hairdressing.

  Given the twin threats of automation and trade, laid-off workers have two clear options. One is to go back to college to acquire or refresh their managerial, professional, or technical capabilities. This requires investment of time and money, but pays off eventually in higher salaries and greater job security. The second is to move down the pay scale to service jobs like that of a security guard, waiter, or a driver that require fewer educational credentials but are immune for the time being to automation.

  In cities, automation has forced many moderately educated manufacturing or service workers into minimum-wage service jobs such as that of a “fulfillment” associate, pulling products off shelves to make up online orders in a large warehouse, while a digital assistant whispers instructions in an ear. The fall from comfortable middle-class unionized jobs into the struggle to make ends meet is extremely painful, but at least there are jobs.

  Moderately educated workers whose firms close because of trade competition typically have few palatable alternatives. With few new jobs near the small towns or semirural areas where these firms are located, and most such jobs to be found in firms in the same industry beset by the same competitive woes, workers have bleak prospects if they stay put. Nevertheless, a US study tracking the careers of these workers finds that while better-educated management workers move quickly elsewhere to jobs in new industries, many of the moderately educated seem to have clung on to nearby, and progressively less-well-paid, manufacturing jobs as long as they could.21 When these were no longer available, many quit the labor force altogether, to go on Social Security disability insurance and Medicaid—the only supports available to working-age workers who no longer feel they can find work.22

  Why does the moderately skilled manufacturing worker hit by trade competition not retrain to get a better-paid skilled job or take a low-paid manual-service job, just as those hit by automation do? For the worker affected by trade, both choices typically require a physical move away from the community, for the local economy is usually devastated. Retraining is not easy, especially for manufacturing workers who went to work after high school many years ago, and who really have not used computers at work or at home.23 Moreover, a job as a medical technician requires a few years’ training, significant expenditure on courses, and no income in the meantime. The US government’s Trade Adjustment Assistance program, the primary program for workers affected by trade competition, meets only a fraction of the likely expenditures, and has onerous qualification requirements. For males, a job as a nurse also requires an acceptance that old stereotypes—for example, that nursing is for women—no longer apply.

  A well-paying job at the end of this odyssey is not assured. Many workers would deem such investment both costly and risky. Easier perhaps would be to move to take a service job that does not require additional training in a flourishing city. Here too, as we will see shortly, options are narrower because of occupational licensing. Moreover, after paying the higher rents in a strange costly city, there is little left over to support a family. For many, perhaps it is easiest to stay put and hope past jobs return, even while the local economy and the community decline further—after all, there are still friends and family. The behavior of these workers is not dissimilar to that of the handloom workers during the Industrial Revolution who kept entering the industry even as its impending demise was clear. It may not change unless the community or the state come up with more viable options.

  THE EFFECTS OF TECHNOLOGY AND TRADE ON INCOMES

  Let us now turn to incomes. If technology and trade affect the nature and number of jobs, they must also affect wages and incomes. A wage is a price for labor services, so it will be affected not just by the jobs available (crudely speaking, demand) but also by the available applicants for those jobs (the supply). There are two important patterns in the data. First, there has been an increase in incomes of the college educated (those with an undergraduate degree or higher) relative to the moderately educated (those with a high school diploma). Second, incomes for the very top earners (say the top 1 percent of incomes) have been running away from the incomes for the rest.

  THE STAGNANT MEDIAN WAGE DEBATE

  A number of studies have documented rising income differentials since the late 1970s in the United States between workers at the ninetieth percentile of the income distribution (typically college educated) and workers at the fiftieth percentile of the income distribution (typically only high school educated). This has two aspects to it. First, except for a brief period of growth between 1996 and 2004, the real hourly wage in the United States (the money wage deflated by the Consumer Price Index [CPI]) for the worker in the middle of the wage distribution has stagnated, so that in 2014, it was about the same as in 1980.24 These facts are not disputed, but they need not mean that the worker in the middle is no better off. For one, the worker takes home the median wage after taxes are subtracted and government transfers are added. Taxes at the middle have fallen and transfers have increased since the late 1970s so worker income, after tax and transfers, has grown. Second, actual inflation experienced by households has been lower than the CPI, which means real wages have grown more. Finally, Americans have smaller families, with many living alone, so there are fewer people dependent on any single income. Correcting for these factors, the Congressional Budget Office concludes that median household incomes have increased by about 50 percent since the late 1970s.25 This is not spectacular over a period of about forty years, but it is much better than stagnation.

  Less disputed is that the wage premium for those who have been to college has increased steadily since 1980. The wages of the college educated have been running away from the wages of the merely high school educated, though even for the college educated the job market has moderated in recent years.26 Across the world, the highly educated earn a premium over the moderately educated, who in turn earn more than the uneducated. On average, across the Organization for Economic Co-operation and Development (OECD— informally, the club of rich countries) in 2015, those who had not completed high school earned only 79 percent of the wages earned by high school graduates, those with bachelor’s degrees earned 46 percent more, while those with master’s degrees earned 98 percent more.27 The educated are also much more likely to be part of the labor
force—either working, or actively looking for work—and also less likely to be unemployed.

  In the United States, the premium on education is higher than the OECD averages. The few who have not completed high school earned only 68 percent of the wages earned by high school graduates, those with bachelor’s degrees earned 66 percent more, while those with master’s degrees earned 132 percent more. The average wage premium for the educated does not mean that everyone with an undergraduate or graduate degree has a well-paying job. These are averages, and the averages conceal a number of highly educated individuals in low-paying jobs who had other priorities than income, were unfortunate in their choice of college or field of study, or who were just plain unfortunate.

  Only a handful of countries have a greater fraction of the labor force aged twenty-five to sixty-four with tertiary degrees, or higher average years of schooling in their population than the United States, yet its tertiary education premium resembles that in emerging markets like Turkey or the Czech Republic, which have a far lower fraction of the labor force with a tertiary degree.28 Indeed, the OECD estimates that the total benefits to getting a tertiary education (relative to staying with high school education) for a man in the United States in 2013 to be $569,600, second only to Chile at $576,900.