Were the Luddites Right?
Smart machines and the prospect of technological unemployment
In 1948 Norbert Wiener, the father of cybernetics, wrote an urgent letter to Walter Reuther, the president of the Union of Automobile Workers. Wiener warned Reuther that technologies that combined computing machines with production machinery would soon yield an "apparatus [that] is extremely flexible, and susceptible to mass production, and will undoubtedly lead to the factory without employees; as for example, the automatic automobile assembly line." Wiener ominously concluded, "In the hands of the present industrial set-up, the unemployment produced by such plants can only be disastrous."
The mass unemployment that Wiener predicted did not occur. As technology advanced, the number of employed workers in the United States increased from 59 million in 1950 to a peak of 146 million in 2007, and GDP grew from $2 trillion to $13.6 trillion (in 2005 dollars) between 1950 and 2012.
Now, two centuries after Luddites smashed then-newfangled weaving frames in northern England, predictions of permanent technological unemployment are being revived. In a December working paper for the National Bureau of Economic Research, called "Smart Machines and Long-Term Misery," the Columbia economist Jeffrey Sachs and the Boston University economist Laurence Kotlikoff pose the question, "What if machines are getting so smart, thanks to their microprocessor brains, that they no longer need unskilled labor to operate?" After all, they point out, "Smart machines now collect our highway tolls, check us out at stores, take our blood pressure, massage our backs, give us directions, answer our phones, print our documents, transmit our messages, rock our babies, read our books, turn on our lights, shine our shoes, guard our homes, fly our planes, write our wills, teach our children, kill our enemies, and the list goes on."
Sachs and Kotlikoff are not alone in worrying how technological progress will affect employment. Last year, Erik Brynjolfsson and Andrew McAfee of MIT's Center for Digital Business published a small book, Race Against the Machine, that looks at trends in U.S. employment and wages. It concludes that the pace of progress "has sped up so much that it's left a lot of people behind. Many workers, in short, are losing the race against the machine." And in a 2011 article for the McKinsey Quarterly, the Santa Fe Institute economist Brian Arthur describes automation as "a second economy that's vast, automatic, and invisible." In Arthur's view, "The primary cause of all of the downsizing we've had since the mid-1990s is that a lot of human jobs are disappearing into the second economy. Not to reappear."
As evidence that American workers are losing to the machines, Brynjolfsson and McAfee point to falling real wages for unskilled workers in the United States. The Employment Policy Institute's 12th State of Working America report reveals that in constant 2011 dollars, the hourly wage for men with less than a high school education fell from $17.50 per hour in 1973 to $12.70 in 2011; wages for men with a high school diploma fell from $20.70 to $17.50 per hour; and even hourly pay for those with some college dropped from $20.20 to $19.50. In the same period, real wages for college-educated men rose from $28.50 to $31.80. Men with graduate degrees saw a gain from $31.70 to $41.30. In addition, Americans are leaving the labor force. Adult workforce participation peaked at 67.3 percent in 2000 and has now fallen to 63.6 percent. This suggests that had the labor force participation rate remained the same as it was in 2000 that the current unemployment rate would be around 13 percent instead of 7.9 percent.
Other researchers are less worried, pointing out that many of the alleged labor force dropouts are actually taking time out to upgrade their skills in school. And when you're comparing how workers fared in 1973 and 2011, you should bear in mind that the percentage of male workers without a high school degree dropped from 30 to 10 percent in the same period, while those with four years or more of college rose from 16 percent to 31 percent.
According to a recent study by the Cleveland Federal Reserve Bank, labor's share of our gross national income has fallen from 65 percent in the 1980s to 58 percent now, with the result that a larger percentage of national income is going to the owners of capital, e.g., the owners of machines. The increasing rewards to capital particularly bother Sachs and Kotlikoff. "Machines, after all, are a form of capital, and the higher income they earn based on better machine brains may show up as a return to capital, not labor income," they worry. Sachs and Kotlikoff devise an admittedly simple economic model in which the current generation of skilled owners of smart machines reaps most of the benefits of increased productivity and economic growth. This competition with the smart machines depresses the wages of young unskilled workers that, in turn, limits their ability to save and invest in skill acquisition and smart machines. Thus arises a vicious circle in which each subsequent generation of young unskilled workers faces an economy in which ever less human and physical capital is available to them, which further hammers down their wages and so forth.
If it really is different this time, what should be done? To prevent "immiserizing" young unskilled workers, Sachs and Kotlikoff argue that the government should tax away some of the "windfall" that the owners of capital gain. The government would transfer some of the taxes to the younger generation, presumably so that they could invest in skills acquisition and smart machines. In addition, the government would itself invest a portion of the taxes in smart machines and then transfer the income from the government machines to the younger generations. "With the right choice of tax-and-transfer policies," declare Sachs and Kotlikoff, "all generations can benefit from the advance of technology, while under laissez faire, only today's older generation benefits, and at the expense of all other generations." That's the theory, anyway.
Brynjolfsson and McAfee make more concrete recommendations. First, they suggest more investment in education. That is a bit puzzling, since earlier in their book they note that the education sector "lags as an adopter of information technologies." Even more oddly they don't wonder why that might be. (Two words: government monopoly.) They do, however, recognize that the rise of online schooling could have a big beneficial impact on upgrading labor force skills. They also advocate reforms such as aggressively lowering the barriers to business creation, resisting efforts to regulate hiring and firing, decreasing payroll taxes, decoupling benefits from jobs, not rushing to regulate new network businesses, streamlining the patent system, and shortening copyright terms. Such sensible reforms should be adopted whether or not technological unemployment is a problem.
Brian Arthur looks at the longer-term implications of the second economy. Smart machines will boost economic and prosperity indefinitely, but such an economy may not provide jobs. Until now, compensation for labor is how people gained access to the growing prosperity that increasing productivity made possible. "The second economy will produce wealth no matter what we do; distributing that wealth has become the main problem," argues Arthur. Perhaps we will work less. After all, in 1900 Americans worked an average of 2,300 hours per year; now it's 1,800.
Or, as has happened so far, entirely new economic sectors could come into existence, providing work for future generations. In 1985, there were just 340,000 mobile phone subscribers in the United States; today there are more than 321 million. More broadly, increasing productivity lowers the prices of goods and services, leaving consumers more disposable income that they can instead spend on other goods and services. In 1950, American families spent 18 percent of their food budgets dining out. Today they spend 40 percent. In 1972, the U.S. had one restaurant for every 430 Americans. It's 1 for every 320 today.
Brynjolfsson and McAfee argue that instead of racing against the machines, we should race with them. Of course, as rising productivity shows, that's exactly what we have been doing. In the industrial era, machines largely complemented and substituted for human brawn; now they are complementing and substituting for human brains. To better race with increasingly smart machines, we will become more intimate with them by incorporating them into our bodies and brains.
Already polls suggest that a majority of us suffer from nomophobia, fear of being without our mobile phones. Babak Parviz, head of the Google Glass project, suggests in Wired that digital displays will some day be incorporated in contact lenses. Such lenses would provide the wearer access to augmented reality in which information is overlaid on whatever she is looking at while also eliminating the need for displays on phones, computers, and televisions. Even stuttering first steps toward brain/computer interfaces are being taken today.
Our ever more productive machines will continue to reduce scarcities, thus lowering the relative prices of goods and services. If governments take the good advice offered by Brynjolfsson and McAfee, we can have a future in which tens of millions of micro-entrepreneurs provide ever more specialized goods and services to ever more discerning consumers. The resulting prosperity could well free people to employ themselves in tackling scarcities in other facets of life, such as liberty, love, and time. I wouldn't want to break the machines that would make that possible.