Originally published in The Wall Street Journal
Appeared in the June 10, 2018, print edition

The Fed Can’t Save Jobs From AI and Robots

The central bank’s employment mandate can’t be squared with coming tech disruption.

By MARTIN FELDSTEIN

(PDF Version)

The day is coming, experts tell us, when artificial intelligence and robotics will massively disrupt the labor market. Autonomous vehicles will put 3.5 million truck drivers at risk of losing their jobs. Checkout machines may replace 3.4 million retail cashiers. That is only the beginning of the long list of jobs that will be destroyed by technological change.

The shift will not happen all at once, and most of the people who lose their jobs will eventually find new employment. The benefits of automation will include lower production costs, which will increase real incomes and job-creating consumer demand. But the technology will also cause individual hardship and frequent periods of increased unemployment.

These large supply shocks cannot be offset by monetary policy. They therefore will present the Federal Reserve with a new challenge. In 1978 Congress gave the Fed a “dual mandate” of price stability and maximum employment. This distinguishes the Fed from the world’s other major central banks. The European Central Bank, the Bank of England and the Bank of Japan, for example, are required to target only the rate of inflation (although they may informally pay attention to the level of employment).

The creation of the Fed’s dual mandate reflected concern in Congress that cyclical declines in the demand for goods and services would lead to increased unemployment. Economic theory implies that low interest rates and an easy monetary policy can increase the demand for output and labor. Thus, Congress charged the Fed with pursuing a monetary policy that would achieve maximum employment.

The coming challenge is different. Job losses will be caused not by low demand, but by supply shocks as artificial intelligence allows machines to replace labor. Technological unemployment has happened in the past, such as when automated looms in factories replaced hand looms. But the experts expect that AI will lead to much more widespread disruption.

If so, an easy monetary policy would be the wrong response. Policies that increase aggregate demand will not succeed in replacing the jobs that technology makes obsolete. With regard to employment, the technological change that makes a truck driverless is equivalent to destroying the truck.

Monetary policy will still be an appropriate tool the Fed can use to respond to traditional cyclical changes in demand. But technological disruption will make the unemployment rate a very noisy signal of the demand level. The Fed’s policy goal should therefore be shifted so that it focuses solely on price stability, in line with what other central banks now do. Achieving the government’s goal of maximum employment will require different policies, like increased job training and the removal of state licensing barriers.

Shifting the Fed to a single mandate of price stability would provide an opportunity to reconsider the inflation target of 2%. The Fed adopted this as its definition of price stability in 2012, but other central banks don’t pursue a single, fixed inflation rate. The European Central Bank aims at inflation “below, but close to, 2%.” Some central banks prefer a range of allowable inflation rates.

My own preference would be for the Fed to use monetary policy to prevent sharp moves in measured inflation. With this approach, the Fed would tighten if the inflation rate began to rise rapidly and would ease if there were a sharp decline.

But whatever the inflation goal, the Fed should not try to regulate the rate of unemployment. Though it may sound paradoxical, in an economy in which job losses are being driven by technological disruption, the Fed would do better to ignore employment and focus exclusively on the rate of inflation.