A couple of years ago the editor of Business Week had a problem with his car: Whenever he went too fast--whenever the needle on his speedometer went above 40--the car developed a dangerous shimmy. So he carefully drove to the repair shop, never letting the needle go past 39. Alas, after looking at the car, the mechanic declared that he couldn't fix the shimmy. Moreover, he had found another problem: The speedometer was defective. In fact, when the needle was pointing to 40, the car was actually going 55. And he couldn't fix that problem, either.
To the mechanic's surprise, the editor was pleased with this news. "So what you're telling me is that the car doesn't shimmy until I go 55 miles per hour. That means I can drive home 15 miles an hour faster than I drove here!"
OK, OK, I made that story up. I have never met Stephen Shepard, editor in chief of Business Week, but I'm sure that he would never make that kind of mistake in ordinary life. It would not be necessary for the mechanic to explain pedantically that, while it was true that the news about the speedometer implied that the car could go faster than previously thought, it did not change the speedometer reading at which the car shimmied.
But he is apparently not so clearheaded when it comes to economics. Indeed, the whole "New Economy" doctrine--a doctrine relentlessly espoused by his magazine for the last few years and vociferously defended in a recent signed essay by Shepard himself--is based on a misunderstanding of the relationship between measurement and reality that is conceptually identical to the garbled thinking of the imaginary editor retrieving his car.
The New Economy doctrine, sometimes called the New Economic Paradigm, may be summarized as the view that globalization and information technology have led to a surge in the productivity of U.S. workers. This, in turn, has produced a sharp increase in the rate of growth that the U.S. economy can achieve without running up against capacity limits. "Forget 2% real growth," urges Shepard. "We're talking 3%, or even 4%." This increase in the potential growth rate, in turn, is supposed to explain why the United States has managed to drive unemployment to a 25-year low without inflation.
The conventional view that the economy has a "speed limit" of around 2-percent to 2.5-percent growth does not come out of thin air. It is based on the real-life observation that when the output of the U.S. economy--as measured by real gross domestic product--is growing rapidly, the unemployment rate falls; when the output is growing slowly or is shrinking, the unemployment rate rises. Over the last 20 years, the break point--the growth rate at which unemployment neither rises nor falls--has been between 2 percent and 2.5 percent. And this break point does not seem to have changed much in recent years: Since mid-1994, GDP has grown at about a 2.7-percent annual rate, while unemployment has fallen at a steady rate, implying that the no-change-in-unemployment growth rate is closer to 2 percent than to 3 percent. (Click to see a chart that illustrates the break point.)
So what? Don't we want unemployment to fall? Yes, of course, but the unemployment rate can fall only so far. Obviously it can't go below zero; and in reality, the limits to growth are reached long before the economy gets to that point. Both logic and history tell us that when workers are very scarce and jobs very abundant, employers will start bidding against each other to attract workers, wages will begin rising rapidly, and real growth will give way to inflation. That means that while the economy can grow faster than 2-point-whatever percent for a while if it starts from a high rate of unemployment (like the 7.5-percent unemployment rate that prevailed in late 1992), in the long run, that growth rate cannot remain higher than the rate that keeps unemployment constant. And that is where the infamous "speed limit" comes from.
Behind that speed limit, in turn, lies another bit of arithmetic: The rate of growth of output, by definition, is the sum of the rate of growth of employment (which is limited by the size of the potential labor force) and that of productivity, a k a output per worker.
A ha! say the New Economy advocates--that's exactly our point. Productivity growth has accelerated, which means that the old speed limit has been repealed. It's true, they concede, that official productivity statistics do not show any dramatic acceleration--in fact, measured productivity growth in the '90s has been about 1 percent per year, an unimpressive performance similar to that of the two previous decades. (It has gone up more than 2 percent in the last year, but this is probably just a statistical blip.) But they insist that the official statistics miss the reality, understating true productivity growth because, as Shepard insists, "we don't know how to measure output in a high-tech service economy."
He's probably right about that. What he may not realize is that we really didn't know how to measure output in a medium-tech industrial economy, either. How could productivity indexes--which basically measure the ability of workers to produce a given set of goods--properly take account of such revolutionary innovations as automobiles, antibiotics, air conditioning, and long-playing records? Just about every economic historian who has looked at the issue believes that standard measures of productivity have consistently understated the true improvement in living standards for at least the past 140 years. It's anybody's guess whether unmeasured productivity growth in the last few years is greater or less than in the past. (My personal guess is that the hidden improvements are less important than they were in the 1950s and 1960s: For example, direct-dial long-distance calling and television made more real difference to our lives than the Internet and DVD.)
But anyway, that is all beside the point. After all, what is this number we call "productivity"? It is measured real GDP per worker--nothing more, nothing less. Suppose Shepard is right that we are understating productivity growth by, say, 1 percent. Since nobody thinks we are overstating employment growth, he must believe that the official statistics understate true output growth by exactly the same amount. Now Shepard is quite right that if true productivity growth is 2 percent, not the 1-percent measured rate, and if the labor force is growing at 1 percent, then the economy's true speed limit is 3 percent, not 2 percent. But when the economy is actually growing at 3 percent, the statistics will say that it is growing at 2 percent--and yet it cannot grow any faster.
Still, Shepard thinks that it can. "Perhaps the 4% rate of the past 12 months is too high. ... But the 2%-to-2 ½% speed limit is probably obsolete. In an era of stronger productivity growth, which may just now be showing up in statistics, the speed limit is probably 3% to 3 ½% a year." In short, now that he knows (or, anyway, prefers to believe) that the speedometer has been understating his speed, and that the shimmy therefore doesn't start until he is really going 55, he thinks that he can drive 55 as measured using that same speedometer. Uh-uh.
B ut doesn't the happy combination of low unemployment and low inflation show the payoff from hidden productivity growth? Well, higher productivity growth would mean lower inflation for any given rate of wage increase. And if official productivity statistics understate the real rate of progress by 1 percent, official price statistics also overstate inflation by exactly the same amount. This is a cheerful thought, but it also means that invoking covert productivity increases doesn't help explain why even measured inflation remains quiescent. The low rate of inflation in the U.S. economy is indeed a surprise: But the puzzle is why wages have not risen more rapidly despite very tight labor markets, not why prices have remained stable given very moderate wage increases. And productivity us with that one.
Shepard's essay was pretty obviously intended as a response to economists--including Princeton's Alan Blinder, Morgan Stanley's Steve Roach, and me--who have recently been critical of the New Economy doctrine, among other things pointing out (though apparently to little effect) the dependence of that doctrine on the speedometer fallacy. It seems clear that he is baffled by the reluctance of "Old Economists" to join the party, and that he can only explain it by their unwillingness to accept the idea that the world has changed and that their pet theories are no longer valid. Well, I can't speak for the others, but I have no particular aversion to admitting that the economy can change and that old rules sometimes don't apply. In fact, as anyone who makes much of his income from book royalties and speeches can tell you, the incentives are all the other way: People would much rather hear about how everything has changed than about why most of the usual rules still apply. (And feel-good optimism sells much better than dismal realism.)
No, the reason I can't buy into the New Economy is actually very simple: Despite all the incentives, I can't bring myself to endorse a doctrine that I know to be just plain dumb.