Why One Economist Predicts Slow Growth

Robert J. Gordon of Northwestern University belongs to the committee of distinguished economists who officially declared on Sept. 20 that the U.S. recession ended way back in June 2009. Don't mistake that pronouncement for optimism. According to Gordon's research into the long-term determinants of growth, America's next two decades are going to be disappointing. He predicts that between 2007 and 2027, gross domestic product per capita will grow at the slowest pace of any 20-year period in U.S. history going back to George Washington's Presidency. Although the data he examined closely go back only to 1891, he says that based on his knowledge of early American economic history, he thinks it is fairly safe to predict that the period will witness the slowest growth ever in GDP per capita and, therefore, American living standards.

Gordon isn't out on a limb. His prediction is based on several strands of existing research on workforce demographics, educational attainment, and technological change. His contribution has been to pull the strands together and draw the logical conclusion. According to data Gordon prepared for Bloomberg Businessweek, based on research he published earlier this year, GDP per capita in the U.S. grew at a robust 2.44 percent annual rate from 1928 to 1972. That slowed to 1.93 percent from 1972 to 2007 and is likely to slow further, to 1.5 percent from 2007 to 2027. At that rate, GDP per capita would increase by a total of 35 percent by 2027. That's far short of the 62 percent that it would grow if the 1928-1972 pace of growth had continued, or the 47 percent increase if the 1972-2007 pace had continued.

What's behind this forecasted stretch of slow growth? One major reason, Gordon says, is the coming retirement of the baby boomers, which will decrease productive wage-earners as a share of the population. That demographic negative would be tolerable if the workers who remain on the job rapidly increased their productivity, or output per hour. Gordon doesn't see that happening. For one thing, he says, education levels in the U.S. have stopped increasing: It's usually hard to raise productivity in the long run without better-educated workers.

For another, there's no reason to predict another technological revolution like the Internet boom, which briefly boosted productivity in the early 2000s. Gordon says the past decade has disproved economists who believe that the Net had a "magic quality" that would speed up productivity growth for an extended period. That was a common theme for investors at the time of the tech boom of the late 1990s, when some "New Economy" believers insisted that productivity growth had reached a new, higher plateau. Not so, says Gordon: "What we're going through now is a transformation to smaller devices without a change in the basic idea of computers communicating with each other," he says.

Labor productivity grew a strong 3.5 percent in 2009 and continued at a 3.9 percent annual rate in the first quarter of 2010. Those big gains in output per hour seemed to presage healthy gains in standards of living over time. But Gordon considers the big productivity jump a fluke, noting that employers cut jobs more drastically than in previous cycles. "It was a general panic to cut costs" that wasn't sustainable, he says. One sign that employers cut too much is that they've had to backtrack: Productivity declined at a 1.8 percent annual rate in the second quarter, although that decline isn't expected to continue for the whole year.

If living standards do rise more slowly over the next couple of decades, it will become even harder for the U.S. to fix costly problems, from carbon emissions to poverty. It's always easier to share a pie that's growing. More education and innovation would help. For Gordon, that's for someone else to worry about. His job is to run the numbers.

The bottom line: Economist Robert Gordon predicts the U.S. will soon be generating the slowest GDP growth per capita in its history.

    Before it's here, it's on the Bloomberg Terminal.