I still remember the first time I walked into a working factory. In the foreground, innumerable machines whirred and clacked away in precise, interlocking dances. A massive vat shaped like a 50-foot-tall Campbell’s soup can loomed in the background. It was encased in a protective sheath of refractory bricks that glowed dusky pink with trapped heat. A crane arm dumped heavy sand continuously into the top at (literally) industrial volumes. Steaming, liquid glass gushed out of the business end at the bottom in a matching stream. I couldn’t see the heating element, but it was in there somewhere, and it was working.
It’s embarrassing to admit this, but at that moment I felt the way I was always told I was supposed to feel when I saw a mountain range. I laughed involuntarily, and then had a thought that I never could have had about a mountain: Men made this. I was looking at concretized human ingenuity. In the auto industry, “car guy” is a slang term for an executive who doesn’t just view the business of a car company as making money, but loves the cars themselves. I’m a factory guy.
I spent the first few years of my career in the 1980s as one small part of a self-conscious movement to rescue American manufacturing from its projected obsolescence. I’ve worked in glass plants, assembly plants, oil refineries, and textile plants from Florida to Canada, and many points in between. I’ve carried a union card and walked a picket line. I loved being in the plant, and using the old-fashioned accounting term “burden” for head-office costs. I enjoyed making fun of the guys working in finance in New York.
Reform of manufacturing was focused on the need to change business practices. There was no one huge insight; it was painstaking effort conducted one work-rules negotiation, quality-improvement session, and production-planning algorithm at a time. As a general rule, people resist change in business. There are good reasons for this: It is easy to theorize about improvements, but the law of unintended consequences applies to all complicated organizations, not just governments. But there are also less-good reasons for it: predominantly, that change is hard, human beings are lazy, and it’s hard to get incentives aligned properly so that the necessary people are willing to act.
The power of capital markets to force change was revealed to me when a large manufacturing company where I was consulting was confronted by a leveraged-buyout offer. The financial wolf was at the door, and I was placed on a small team tasked with figuring out the changes that would have to be made to meet the new financial targets and avoid losing the company to the raiders. Suddenly, in the face of crisis, much of the previous resistance melted away. In the end, the company did pretty much what the buyout group would have done. It was sloppy and painful, but the company emerged healthier and growing. It turns out those finance guys in New York were doing something important to prevent the decline of American manufacturing into mediocrity and irrelevance after all.
Repeated application of the process of financial pressure followed by operational response has paid off. The U.S. has a very productive manufacturing sector. It just looks much different from what most of us imagined it would be. One of the most important differences is that it doesn’t employ many more people than it did in 1947. Here is a chart produced by the Federal Reserve Bank of Cleveland that summarizes the situation:
At the end of World War II, manufacturing accounted for about one-third of the U.S. workforce. Today it is about one-tenth. In terms of employment, we are no longer transitioning to a services economy; we are there. Over that same period, manufacturing has consistently represented about 15 percent of rapidly growing U.S. economic output. Manufacturing has become ever more productive; as any industry matures, this becomes increasingly necessary for survival. While continuous improvement is essential for any relatively mature business, the process of improvement itself is fairly routinized, and almost inexorably eliminates labor. Productivity gains are beneficial to consumers, but not so much to employees who have trouble moving or learning new skills.
The ongoing stay of execution for the Detroit automakers represents an attempt to keep this process from reaching the last bastion of resistance. In spite of its glorious past, the Detroit ecosystem — Ford, GM, Chrysler, their suppliers, and their dealers — is going to have to go through the functional equivalent of bankruptcy, or else become a ward of the state. Large economic transformations are always painful, and it is almost inevitable that industries, communities, and individuals will attempt to use the political process to protect their current status. In the end, though, a healthy political process will prevent this dynamic from choking off economic progress.
It is often easier for us to see this from a greater distance. Consider the evolution of the U.S. agricultural sector, which presaged what we have seen in manufacturing. In 1800, America was a nation of farmers. About three-quarters of the labor force worked in agriculture. The proportion has been in almost continuous decline since then. By the eve of the Civil War, it was a little over half; by 1900, it was about one-third. Today, agriculture represents less than 3 percent of the workforce. This has been great for consumers. Farming is incredibly efficient, and food is cheaper and more plentiful in real terms than ever before in human history — to the point that obesity is a serious public-health problem. American agriculture today is also a very successful industry. In 2007, the U.S. exported over $75 billion of agricultural products, and it has maintained a trade surplus in food for decades. It’s just not an industry that can provide employment for very many people anymore.
This seems obvious now, but resistance to these changes was a key basis of the populist movement in the late 19th and early 20th centuries, as farming’s long-term decline in employment became undeniable. William Jennings Bryan is often seen as something of a laughingstock in contemporary culture because of his buffoonish treatment as the lawyer attacking the teaching of evolution in the 1960 movie Inherit the Wind, but he was a very influential political figure who won the Democratic nomination for president in 1896, 1900, and 1908. He, more than any other person, incorporated the key themes of the populist movement into the two-party system.
Bryan was unabashedly attempting to stand athwart economic history yelling “Stop!” In one of the most arresting passages in his famous “Cross of Gold” speech, he said bitterly: “I tell you that the great cities rest upon these broad and fertile prairies. Burn down your cities and leave our farms, and your cities will spring up again as if by magic. But destroy our farms and the grass will grow in the streets of every city in the country.”
America became a manufacturing nation within Bryan’s lifetime. This sweeping technological and economic change produced enormous flux in social, political, and family relationships, and his search for permanence was emotionally understandable. One of the most painful things about markets, one that tends to be made more extreme by periods of rapid change, is that they often make fools of our fathers. Sharp operators with an eye for the main chance often outperform those who carefully learn a trade and continue a tradition. The Industrial Revolution combined material deprivation for people who had known only farming with radical uncertainty about the future.
Bryan lost three presidential elections, so the other guys got to write the history books. But of course losing three times means that he won the nomination three times, which is excellent evidence that he spoke for a good fraction of the American people of his era. In the end, though, accepting his broad program would have meant opting out of the modern world. Though sentimentally painful, rejecting Bryan was wise. The alternative would have been to attempt to prop up emotionally resonant family farms and retard the development of the industrial economy.
This same dynamic has been playing out for manufacturing over the past 60 years. It has been exacerbated by international competition to a degree not seen in agriculture. Countries with much lower wages can produce goods cheaply and export them to the U.S. market, especially goods with labor-intensive manufacturing processes. Consequently — unlike the case with agriculture — the U.S. runs a consistent merchandise trade deficit. Those American manufacturing industries that are not in secular decline typically respond to this situation with one or more of the following strategies: (1) automate production to reduce the labor content, (2) manufacture overseas and become a research, design, and marketing company, (3) employ low-wage/low-skill manufacturing labor (very often illegal-immigrant labor with pre–New Deal economic protections), and (4) produce complex goods that require frequent manufacturing innovation and provide high-wage jobs. As Jack Welch, the legendary CEO of General Electric who probably did as much as any individual to drive manufacturing reform, put it in the 1980s: American factories must “automate, emigrate, or evaporate.”
Most of these alternatives are bad for traditional-production workers, though some can work out well for manufacturing employees who can be flexible and develop advanced capabilities. As a result, while there is always a wage premium for those with better skills, manufacturing-sector jobs are increasingly re-segregating into high-skill/high-wage and low-skill/low-wage ghettos in a pattern very reminiscent of early-20th-century America. According to research by Claudia Goldin, an economic historian at Harvard, the ratio of pay for a starting engineer to that of the average production worker declined from about 1.4 in 1904 to about parity in 1956, at the peak of the post-war economic high tide. In 2007, the average starting industrial engineer made about $55,000 per year, or about 1.5 times the $37,000 that the average non-supervisory production worker made. And more broadly, wage inequality and the economy-wide skills premium have been rising in the U.S. for the past 30 years.
The days of getting out of high school, working in a factory, and having a middle-class life are pretty much gone, because the economic world of 1955 is gone. The jobs that provided this opportunity have been automated out of existence, and our international position no longer allows us to protect them at feasible cost. I take no joy in the need for restructuring the auto industry. I wish that old world still existed, but it does not.
I realize now that my attempts to resist this change were like William Jennings Bryan’s attempts to resist the coming of the economic order that I was trying to preserve. I slowly came to understand through experience that my original vision of saving manufacturing would have destroyed it. Theories for how to revive American manufacturing abounded in the 1980s, and it’s hard to exaggerate how difficult it is to understand which alternatives are feasible and which are not in the face of an economic transformation. It is almost impossible not to be guided by our sentiments in such a situation, and this happened to me. I fought the direction in which market price signals were pushing manufacturing, but in the end, they were the only reliable guide to what might work.
As a crude generalization, new economic sectors that rely on innovation are the ones that produce lots of high-wage jobs for an economy. It is often said that services are less productive than manufacturing, but in a certain way, this is the good thing about services: They provide jobs. Eventually, what we now call services will presumably go through the same transformation as farming and factories have, and we’ll have to find a new sector to provide employment.
At each of these stages, we don’t abandon the maturing industries. We still need food and manufactured goods, and it would be foolish to become completely dependent on foreign supplies for either. In the event of a real shooting war we couldn’t book enough conference rooms to protect ourselves from a determined adversary. But, as we’ve seen, we are well able to feed ourselves, and we have an extremely robust manufacturing economy. In fact, had we tried to freeze in place either family farms or employee-intensive factories, we would probably have a far worse defense capability because we would have less domestic agricultural output and antiquated factories.
Which new sectors will actually be productive, and how they will ultimately develop, is highly unpredictable. This is why the free play of markets with limited intrusion by the government is so essential. Almost all industrial policy ends up protecting existing institutions: This is a function of human nature and is not fixable with clever program design. In practice, industrial policy normally means maintaining jobs that a ruthless market would eliminate, and subsidizing technological developments that can be exploited by existing large firms. But these are rarely the sources of new high-wage jobs. Ironically, these attempts to protect ourselves end up creating a sclerotic economy that in the long run puts everyone at greater risk. The painful reality of economic growth is creative destruction, and in a globalized economy, to lose out in this race is ultimately to put ourselves at the mercy of those who may or may not share our interests.
There is, however, one thing that seems to be predictable about these changes. Historically, each new growth sector, from agriculture to manufacturing to services, has tended to be more abstract than its predecessor. This often leads them to be labeled as “not real work,” and they seem to be a flimsy basis for an economy — just as, I assume, the first farmers who started to scratch out little garden plots were mocked by the hunters on one side and the gatherers on the other, those who did the “real work” in their clan.
I’ve seen this dynamic in fast-forward in the software industry. From directly manipulating physical circuits to assembler code to early programming languages to modern object-oriented programming languages, software has been on a 50-year journey of increasing abstraction from the physical movement of electrons. At each step, some experts in the prior art disparage practitioners of the new one as being not “real engineers.” What they fail to realize is that the skills that the old guard mastered have become routinized, then automated, and are no longer a source of high-wage employment.
This ever-increasing requirement for abstract-reasoning capabilities is why the terrible performance of American students, as compared with international competitors, is so worrisome. While the culture is a significant source of this problem, we need better schools. New methods of organization are typically pioneered in the new industries that drive each economic transformation, and then propagate through other parts of the economy. As is widely known, our basic model of public schools — accepting raw material in the form of five-year-olds, and then adding value through a series of processing steps to produce educated graduates 12, 16, or more years later — is an application of the ideas of the industrial economy. Our progress in improving the quality of this model has stalled. Meanwhile, technology continues to automate away high-wage jobs, and our international competitors rapidly catch up with us, in some cases passing us.
We need a new vision for schools that looks a lot more like Silicon Valley than like Detroit: decentralized, entrepreneurial, and flexible. This will not mean abandoning traditional, disciplined learning, but rather incorporating it as a baseline and doing other things as well. The traditional response is that there are only so many hours in the day. True, but getting more done per hour of labor is just another way of saying “increasing productivity,” and we can’t let education exempt itself from the ongoing productivity revolution any more than we can the auto industry.
Most great teachers, software engineers, and manufacturing employees have intrinsic motivations for their work. This is healthy, and usually necessary for excellence. But the systems in which they work must be unsentimental about channeling these motivations in a progressive direction for society as a whole, when nobody really knows the long-term effects of their actions. This is what markets do, and why we need them. Our relevant political duty, then, is to defend markets. So while I remain emotionally a factory guy, I recognize that it is counterproductive to use the political process to prevent factories from becoming something almost unrecognizable to me.
– Mr. Manzi is a senior fellow at the Manhattan Institute and chairman of an applied-artificial-intelligence software company.