I agree with Becker that the candidates that he discusses for retarding U.S. economic growth in coming years, such as a technological slowdown or reduced private investment, are implausible; also that there are useful reforms, notably tax reform, that can stimulate economic growth. But like Larry Summers, whose recent paper “The 2013 Martin Feldstein Lecture: Economic Possibilities for Our Children,” NBER Reporter, 2013, no. 4, www.nber.org/reporter, has influenced my thinking, there’s a risk, not of a technological slowdown but of a technological speed-up, that could—implausible, futuristic, and alarmist as this may seem—retard U.S. economic growth in the coming decade and beyond.
Suppose that very cheap and very capable machines, which require very little labor to design, produce, or maintain (mainly just a few computer geeks to assist the computers to design new computers and other new machines), provide an attractive substitute for labor in some industry. An example mentioned by Summers is the driverless motor vehicle, which has been developed by Google, and works, and will be in mass production in just a few years, and which is just an ordinary vehicle plus a small though very capable computer. The driverless vehicle could displace most truck drivers, taxi drivers, bus drivers, drivers of ambulances and other emergency vehicles, chauffeurs, and drivers of military vehicles. There are some 3.5 million truck drivers in the United States and 900,000 bus and taxi drivers and chauffeurs: in total, a nontrivial fraction of the total U.S. work force. Where are all these people to go when their jobs are replaced by a computer lodged in a driverless vehicle? Many other jobs will be disappearing because of automation at the same time. Similar technology is likely to be displacing in approximately the same time frame airplane pilots, warehouse workers, workers in factories that produce vehicles, oil-field workers, and many other types of worker as well. Historically, it is true, new technologies created jobs for a lot of workers, but this seems unlikely to happen in the case of computer technology. Driverless cars are a good example. They just require the building and installing of computers that have already been designed. The manpower needs are trivial.
Of course for the driverless vehicle to catch on, it will have to offer economies, as doubtless it will. And, with no truck drivers to pay, the cost of truck transportation will fall, and price will follow. So goods shipped by truck will cost less, and people will buy more of them, or if they buy the same quantity will use the money they save to buy more of other goods or services. But the production of many of the other goods or services that they buy will be automated, just like truck transportation, and the demand for labor by the providers of those goods or services may therefore be declining simultaneously with the decline in the demand for drivers. It is difficult to imagine productive activities that cannot be automated—mining, construction, many medical services, house cleaning: the list goes on and on.
A decline in the demand for labor, caused by automation, will result in lower wages, without necessarily producing an increase in employment (work effort will not grow if there is no work), let alone an increase in national income. With the federal minimum wage near the poverty level (though some states have higher minimum wages, as permitted by federal law), a fall in wages to or near that level may result in a decline in the percentage of the population that is employed, and in turn to a reduction in consumption (assuming the profits from automation are largely limited to a small upper class) and investment. One likely response will be an expansion of the social safety net, which will further discourage employment, as well as requiring higher taxes, which will further limit consumption.
These consequences of technological progress are highly speculative, but cannot be ruled out categorically. Like climate change, they are a shadow falling on the future.