HOW FAST CAN THE U.S. ECONOMY GROW?
(From Harvard Business Review, July/August 1997)
Most economists believe that the US economy is currently very close to, if not actually above, its maximum sustainable level of employment and capacity utilization. If they are right, from this point onwards growth will have to come from increases either in productivity (that is, in the volume of output per worker) or in the size of the potential work force; and official statistics show both productivity and the workforce growing sluggishly. So standard economic analysis suggests that we cannot look forward to growth at a rate of much more than 2 percent over the next few years. And if we - or more precisely the Federal Reserve - try to force faster growth by keeping interest rates low, the main result will merely be a return to the bad old days of serious inflation.
However, many influential people - business leaders, journalists, and even a few reputable economists - do not accept that dreary verdict. They believe that the old speed limits on growth have been repealed, perhaps even that the whole idea of speed limits is obsolete. The conceptual basis for their optimism is sometimes referred to as the "new economy view", sometimes more grandly as the "new paradigm". Whatever one calls it, this new view of the economy has spread with a rapidity rare in the annals of economic thought - indeed, one would have to go back to the rise of supply-side economics in the 1970s to find a case in which a radical economic theory has so quickly become conventional wisdom among a large group of opinion leaders.
The essence of the new paradigm is the claim that the changes everyone can see in our economy - the rise of digital technology, the growing volume of international trade and investment - have qualitatively altered the rules of the game. Rapid technological change, the new paradigmatics claim, means that the economy can grow much faster than it used to; global competition means that we need no longer fear that an overheating economy will experience inflation. This is obviously an attractive view to anyone who would like to see more rapid growth than the disappointing 2-point-something percent offered by conventional economists; it is also a view that many businesspeople insist corresponds to what they see happening in their own industries. So we should not be surprised at the popularity of this view in the business community.
There is only one problem: when you think about it carefully, you realize that the new paradigm simply does not make sense. For some reason, the fact that there are gaping conceptual and empirical holes in the new paradigm - holes that seem obvious to many economists - has not been effectively communicated to a broader audience. Yet the issue is not really difficult or technical: all that it takes to see the trouble with the new paradigm are a few thought experiments and some simple arithmetic.
WHY DOES THE ECONOMY HAVE A "SPEED LIMIT"?
With the decline of the traditional extended family, in which relatives were available to take care of children at need, many parents in the United States have sought alternative arrangements. A popular scheme is the baby-sitting coop, in which a group of parents agree to help each other out on a reciprocal basis, with each parent serving both as baby-sitter and baby-sittee. Any such coop requires rules that ensure that all members do their fair share. One natural answer, at least to people accustomed to a market economy, is to use some kind of token or marker system: parents "earn" tokens by babysitting, then in turn hand over these tokens when their own children are minded by others. For example, a recently formed coop in Western Massachusetts uses Popsicle sticks, each representing one hour of babysitting. When a new parent enters the coop, he or she receives an initial allocation of ten sticks.
This system is self-regulating, in the sense that it automatically ensures that over any length of time a parent will put in more or less the same amount of time that he or she receives. It turns out, however, that establishing such a token system is not enough to make a coop work properly. It is also necessary to get the number of tokens per member more or less right. To see why, suppose that there were very few tokens in circulation. Parents will want on average to hold some reserve of tokens - enough to deal with the possibility that they may want to go out a few times before they have a chance to babysit themselves and earn more tokens. Any individual parent can, of course, try to accumulate more tokens by babysitting more and going out less. But what happens if almost everyone is trying to accumulate tokens - as they will be if there are very few in circulation? One parent's decision to go out is another's opportunity to babysit. So if everyone in the coop is trying to add to his or her reserve of tokens, there will be very few opportunities to babysit. This in turn will make people even more reluctant to go out, and use up their precious token reserves; and the level of activity in the coop may decline to a disappointingly low level.
The solution to this problem is, of course, simply to issue more Popsicle sticks. But not too many - because an excess of popsicle sticks can pose an equally severe problem. Suppose that almost everyone in the coop has more sticks than they need; then they will be eager to go out, but reluctant to babysit. It will therefore become hard to find babysitters - and since opportunities to use popsicle sticks will become rare, people will become even less willing to spend time and effort earning them. Too many tokens in circulation, then, can be just as destructive as too few.
What on earth does all this have to do with the new paradigm? Well, a baby-sitting coop is a kind of miniature macroeconomy: a system in which individual decisions to spend and save are crucially interdependent, because your spending is my income and vice versa. The depressed state of a babysitting coop with too few tokens in circulation is essentially the same as that of the U.S. economy as a whole when it slips into a recession. And the ability of a Paul Volcker or an Alan Greenspan to engineer a recovery from such a recession rests on their control over the money supply - which is to say, over the number of popsicle sticks.
There are, of course, some important differences between the full-scale economy and a baby-sitting coop with at most a few dozen members. One difference is that the big economy has a capital market: individuals who are short of cash can borrow from others who are cash-rich, so that the effects of an overall shortage or abundance of money get mediated through the level of interest rates. An even more important difference involves prices. In the typical baby-sitting coop, prices are fixed: one popsicle stick buys one hour of baby-sitting, and that's that. In the big economy firms are free to change their prices - to cut prices if they are having trouble selling their product, to increase them if they think it will not hurt their sales. It turns out as a practical matter that firms are quite reluctant to cut prices (and workers are very reluctant to accept wage cuts): while prolonged recessions do eventually lead to price reductions, they do so only gradually and painfully. Firms have, however, historically been less reluctant to raise prices in boom conditions. For this reason the kind of shortage situation a coop gets into when there are too many tokens in circulation is rarely severe in market economies; excessive money creation gets dissipated in inflation instead.
Still, the popsicle-stick economy may help us to dispel some commonly held misconceptions about why economists generally think that there are limits to how fast the economy can grow. First, nobody claims that the economy has a 2-point-something percent speed limit under all circumstances. When a baby-sitting coop is in a depressed state because of an insufficient popsicle-stick supply, its GBP (gross baby-sitting product) can rise very quickly when that supply is increased. Thus there is nothing puzzling about the ability of the U.S. economy to grow at a rate of more than 3 ½ percent from 1982 to 1989: thanks to expansionary monetary policy the economy was rebounding from a recession that had raised the unemployment rate to 10.7 percent, and left output probably 10 percent below capacity. The "speed limit" applies only when the economy has expanded as much as it can by taking up slack, by employing unemployed resources.
Second, the logic of the standard economic argument against over-ambitious growth targets once the economy is near full employment is also widely misunderstood; many, perhaps most, of the critics of that argument base their opposition on a misleading caricature of what economists are saying. All too often, advocates of faster growth assert that their opponents believe that "growth causes inflation"; this supposed view is then held up for ridicule. After all, isn't inflation supposed to be a matter of too much money chasing too few goods - and if so, how can growth, which means producing more goods, be a cause of inflation? But that is not what economists who think that an overambitious growth target will be inflationary are saying. Nobody claims that a baby-sitting coop would suffer from shortages if it grew by adding new members, or if current members became more efficient at child care and were therefore able to do more baby-sitting. The limits - and the risk of inflation - apply only to growth achieved by expanding demand, say by issuing more popsicle sticks.
So how much is too much? Again, return to the babysitting economy. How would you know when there were too many popsicle sticks in circulation? One useful indicator would be the frequency with which parents sought but could not find opportunities to babysit - which would essentially be the coop's unemployment rate. Another would be the frequency with which parents sought but could not find babysitters. This would more or less correspond to the U.S. economy's "vacancy rate" - the number of jobs offered by business that are still unfilled. If unemployment were very low and vacancies high, this would be an indicator that the coop was suffering from excessive demand. In the full-scale economy it turns out that the vacancy rate and the unemployment rate are very closely (inversely) correlated - but data on unemployment are collected more regularly and systematically, so we can use the more readily available unemployment rate as a pretty good indicator of labor market tightness.
How low an unemployment rate is too low? There is, to be honest, a fair amount of uncertainty on this question. Pre-1990 evidence suggested to most economists that inflation would begin to accelerate when the unemployment rate fell below about 6 percent; the failure of inflation to show any discernible increase with the rate barely above 5 has been something of a surprise. However, wage increases have begun to accelerate, and stories about labor shortages - usually rare in the U.S. economy - have become common. (In the last 6 months of 1996 such stories were about 3 times as common as they had been a year earlier). So it seems unlikely that there is much room for cutting the unemployment rate merely by increasing demand. (The emphasis here is to shortcut arguments about the possibility of cutting unemployment through worker training or other kinds of interventions. Such schemes might or might not work, but they have nothing to do with the new paradigmatics, who claim that the Fed can go ahead and adopt higher growth targets even in the absence of programs to increase the number of employable workers).
What we do know reasonably well is what growth rate is consistent with keeping the unemployment rate at roughly its current level. There is a strikingly close relationship between the economy's growth rate and the rate of change in the unemployment rate; indeed, it is one of the few things economists are willing to call a "law" (Okun's Law) with a straight face. Figure 1 illustrates that relationship in practice. On the horizontal axis is the economy's growth rate, on the vertical axis the rate of change of the unemployment rate (i.e., if the unemployment rate falls from 7 to 6 percent, the number would be -1). Each marker represents a year between 1980 and 1995. The data all lie quite close to the indicated line, which crosses zero at 2.4 percent - that is, over the period 1980-1995 the unemployment rate rose if the growth rate fell short of 2.4 percent, fell if growth exceeded that rate. (Each additional point of growth turns out to have reduced unemployment by half a point). On the evidence of these data, then, the rate of growth consistent with keeping the unemployment rate more or less where it is right now has in recent history been about 2 percent.
Nor is there any evidence that the growth rate consistent with a constant unemployment rate - which is the maximum growth rate that can be sustained once the economy has taken up all of its slack - has increased in the last few years. The unemployment rate in 1995 averaged 5.6 percent, about the same as the rate in 1990. The average growth rate over those 5 years was 1.9 percent, below our estimate of the sustainable growth rate based on the scatterplot.
Why does the sustainable growth rate seem to be so slow? There are two main reasons. First, the U.S. labor force is no longer growing as fast as it did in the years when baby boomers were growing up and women moving into paid employment; in the 90s to date the number of people working or looking for work has grown at an annual rate of only about 1 percent. Second, according to the official numbers, productivity - output per worker - has also grown at a sluggish annual rate of only 1 percent. The sum of these two numbers is 2 percent: the growth in the economy's productive potential.
All of this seems pretty ironclad. How, then, can the new paradigmatics claim that the economy is in fact capable of much faster growth? Part of their answer is that they simply do not believe the official numbers - that they believe that outdated statistics are greatly understating productivity growth. But is this true? More important, does it matter?
It is a truism that increases in productivity - that is, in real GDP per worker - are the key to long run economic growth. It is therefore a cause for concern that official numbers show that the United States remains in the productivity slow lane it has occupied since the early 1970s: the annual increase in output per worker hour remains in the vicinity of 1 percent per year, far below the nearly 3 percent annual rate of the 50s and 60s.
Many business leaders, however, find these official statistics hard to credit. For one thing, they find it implausible that the digital revolution, which has had so much impact on the way business is conducted, should not have shown more visible payoff. Furthermore, many businesses perceive themselves as having been forced by intense competition to engage in radical measures to increase productivity; again, they cannot believe that these measures have not paid off for the economy as a whole.
To those who believe that dreary official statistics on productivity are wrong, it seems obvious that dreary conventional views about the limits to growth are also wrong. Suppose, after all, that we believe that actual productivity growth in the 90s has been more than twice as high as its measured 1 percent - say 2.5 percent. Then we must also believe that the economy's potential growth, the sum of its 1 percent labor force growth and productivity growth, is actually 3.5 rather than 2 percent. So shouldn't the Fed let the economy rip?
Let's take this in stages. First, one must admit that the critics of official productivity statistics have a case, although there are counterarguments. Techno-skeptics like to point out that although digital technology is flashy and glamorous, it arguably does less for the actual productivity of workers than many less photogenic innovations of the past. (My own favorite example of an utterly unglamorous technology that had a profound effect on the economy was freight containerization, which was introduced in the 1960s and eliminated the need for literally hundreds of thousands of longshoremen and other freight handlers). And after initial enthusiasm, businesses have found that some new technologies - the desktop computer among them - have large hidden costs. It has also been pointed out that much business restructuring does not so much eliminate jobs as outsource them, from large corporations that pay high wages to smaller suppliers who often have lower pay scales. From the point of view of the restructured company it may seem as if the same work is being done with far fewer people; from the point of view of the economy as a whole output per worker may not have increased much if at all. For what it is worth, economists who try to estimate productivity growth are thoroughly divided over the issue; a few believe that productivity has been greatly understated, many believe that it has been at least mildly understated, but a substantial number also believe that the official numbers are more or less right.
However, this discussion is really a side issue, not central to the main point - because if we are asking what growth target is appropriate it doesn't matter whether the official numbers are right. The important point to remember is that productivity, by definition, is measured as output per worker. When we talk about productivity in the U.S. economy as a whole, we are talking about real GDP per worker employed in the United States - nothing more, nothing less. (In anticipation of the next section, perhaps it is also worth remembering that neither the output generated nor the workers employed by U.S.-based companies outside the United States play any role in the calculation of either GDP or productivity). Now suppose that it were true that productivity - real GDP per worker - has actually grown at 2.5 percent since 1990. Does this mean that the Fed should have set a growth target of 3.5 percent over that period, and that by allowing GDP to grow at only 2 percent it stifled the economy's potential? Not at all. After all, since productivity by definition is equal to real GDP per worker - and since nobody is claiming that the numbers on employment are wrong - a claim that true productivity growth has actually been 1.5 percent higher than the statistics say is necessarily also a claim that true GDP growth has been higher by exactly the same amount. You should therefore not fault the Fed for failing to give the economy the 3.5 percent growth it deserved - you should instead congratulate it for getting the growth rate exactly right!
Or put it a bit differently: if the Fed had tried to achieve a measured - not actual - growth rate of 3.5 percent, it would have in fact been seeking a true growth rate of 5 percent, well above the economy's potential. And to have tried that over the period 1990-96 would have meant driving the unemployment rate down far below current levels, to roughly 2 percent. Few people think that is a feasible rate. The question of whether our economic statistics are understating productivity growth is an important one for many issues. One issue for which it is irrelevant, however, is the question of whether the target growth rate as measured using those same statistics should be higher.
GLOBALIZATION AND INFLATION
The claim that the U.S. economy, despite the drab statistics, is actually experiencing high productivity growth is one of two main planks of the new paradigm. The other is the claim that expanding demand will not lead to inflation, even at very low unemployment rates, because of the new importance of global competition. Unlike in the past, the story goes, U.S. companies now have to face actual or potential competition from rivals in Europe and Asia; thus even in the face of strong demand they will not dare raise prices, for fear that these rivals will seize the market.
As in the case of the claim of understated productivity growth, it is possible to question this assertion on the facts. There are without question many American firms facing international competition to an unprecedented degree. However, such global competition mainly occurs in the goods-producing sector - very few services are traded on international markets - and even within manufacturing there are many industries that remain largely isolated from foreign competitors (seen any Chinese refrigerators lately?). Since we are mainly a service economy, this means that no more than 25 and probably less than 15 percent of employment and value-added are actually subjected to the kind of global market discipline that the new paradigm emphasizes.
But discussion of the true extent of globalization, like discussion of the true rate of productivity growth, is beside the point. Even if global competition played a bigger role in the U.S. economy than it really does, such global competition would not raise the economy's speed limit - because no matter how big the world economy may be, once any economic slack has been taken up the maximum possible growth rate of any piece of that economy is still equal to the sum of productivity and labor force growth in that piece.
This seems to be a tricky point to grasp, perhaps because many people wrongly suppose that a global economy is somehow more than the sum of the national economies of which it is comprised. One way to correct this false impression is to recall a parable introduced by MIT's Paul Samuelson more than 30 years ago; I refer to this as the story of Samuelson's Angel.
Samuelson's idea was, roughly speaking, to imagine running history backward. Normally we think of the global economy as coming about through greater economic integration - which is indeed the way we actually got to where we are. But Samuelson suggested alternatively imagining what would happen if you started with a unified economy and split it apart. Specifically, he suggested the following parable: Imagine that initially all of the world's resources were able to work freely with each other - neither distance nor national boundaries existed. But then an angel descended and scattered the resources to many lands; thereafter the nations could trade with each other, but at least some resources - such as labor - became immobile. (This parable is clearly inspired by Genesis 11, the story of the Tower of Babel; presumably the factors of production had dared to challenge heaven). The important point about this parable is that the resulting global economy - produced by pulling apart an economy that did no international trade, because there was nobody else to trade with - would be indistinguishable from a global economy produced, as happened in fact, by partly pulling together previously separate national economies.
Now it is easy to fool yourself into thinking that putting economies together somehow changes the rules, that it removes old constraints on economic policy. One would hardly expect, however, to gain new policy freedom by pulling an integrated economy apart. And yet the actual, imperfectly integrated world economy can be viewed in either way - suggesting that there is something wrong with the claims that the old constraints are gone.
What the new paradigmatics claim, once again, is that because of globalization monetary expansion can now be pursued without risk of inflation. Can this be right? Let's think it through with the help of the angel.
First, imagine a pre-angel world in which all resources can work together without impediment, and in which there is a single language and a single currency - say one consisting of red banknotes. This world economy is just a big baby-sitting coop: while expanding the quantity of money can expand output up to a point, beyond that point it will be counterproductive, and will usually be dissipated in inflation. (True, the world economy is unthinkably immense - gross world product is probably about 25 trillion dollars. But the U.S. economy, with its $7 trillion GDP, is unthinkably immense even by itself - yet it can usefully be thought of as a baby-sitting coop. It is hard to see that going up a notch in scale would make a qualitative difference). Now suppose the angel descends, and divides this world into two nations, each with its own resources and currency - say that one country now uses blue notes, while the other uses green. How does this change the picture?
Well, suppose that the world as a whole is close to full employment, and that both countries simultaneously expand their money supplies - say that they both double the number of notes in circulation. Clearly, this will be no different from what would happen if the pre-angel economy increased its money supply in the same proportion; once the slack in the world economy has been taken up, further expansion will lead to inflation.
Matters would be somewhat different if only one country tried to expand its money supply - say if the green money country simply expanded the number of green notes, while the blue-note nation did not. It might seem that in that case competition from the non-inflating country would limit price rises in the expansionary economy. So perhaps the world as a whole is not exempted from speed limits, but at least the inflationary consequences of a unilateral expansion are muted. But there is a problem with this line of argument, too. What is supposed to be happening to the exchange rate? If the exchange rate is floating, an increase in the number of green notes will lead to a depreciation of green currency against blue. This will lead directly to an increase in the prices of goods that the expanding nation imports (when measured in green currency); it will also lead to a rise in the prices of blue-currency products that compete with green goods, presumably giving those firms room to raise prices. In fact, until the rise of the new paradigm the conventional wisdom was that monetary expansion leads to more inflation, or at least feeds into inflation more quickly, when it is undertaken unilaterally by a single country rather than taking place simultaneously in many nations. And the United States has a floating exchange rate; the dollar rises or falls quite promptly on indications that the Fed is likely to tighten or loosen its policies.
Nor is this just a conceptual argument. It is worth remembering that while large-scale international trade may still seem somewhat novel to the United States, most other countries have long been highly dependent on foreign trade; even now, the share of exports and imports in America's GDP remain well below levels that have been commonplace elsewhere for many decades. Yet international evidence shows without question that excessive monetary expansion leads to inflation just as surely in highly open economies as in ones that do little trade.
All of this is not meant to downplay the importance of globalization for many economic issues. For example, without question growing world trade has been an important factor in sparking economic development in many poor nations; less happily, it has played at least some role in the growing inequality of incomes in advanced countries. One thing globalization has not done, however, is to change in any important way the rules about how fast the US economy can grow. Globalization or no globalization, if the Fed tries to expand the economy faster than the sum of labor force and productivity growth, the result will be inflation.
A PARADIGM'S POPULARITY
The popularity of the new paradigm poses something of a puzzle. The two key arguments of that paradigm are that high productivity growth justifies higher growth targets, and that global competition prevents inflation. As we have seen, however, both arguments collapse - indeed, look quite silly - when given even a cursory critical examination. Moreover, the criticisms offered in this article are neither deep nor unusual: my own experience is that when one tries to explain the new paradigm to an academic macroeconomist who is unaware of the doctrine's growing influence, he or she produces essentially the same critique you have just read within a minute or two, and finds it hard to believe that anyone would take the doctrine seriously.
Yet many people, especially in the business community, do take the doctrine very seriously indeed. Why? One answer is that the critiques described here do not come naturally to businesspeople. The business of doing business is essentially microeconomic: it involves understanding what happens at the level of an individual market, not the way those markets interact. A business leader in general has no need to understand macroeconomic issues, in which the interaction of markets is of the essence. Why should a CEO know or care about how an increase in the money supply affects GDP, or how that effect changes when an economy has a floating as opposed to a fixed exchange rate? A second reason for the popularity of the new paradigm is, of course, that it tells businesspeople what they would like to hear. Who would not be attracted to a doctrine that promises that the economy can expand without limit for the indefinite future?
There is, however, one more reason for the special appeal of the new paradigm: it is extremely flattering to the businesspeople who constitute its audience. Imagine an advocate of the new paradigm speaking to a group of, say, several hundred top executives. The new paradigmatist tells the group that their new tough management style and their application of cutting-edge technology have brought a productivity revolution; meanwhile, they all know that they can't raise prices because they must now face intense global competition. Now in that group there will surely be at least a few executives whose honest reaction should be: "Well, that may be the way it is in other lines of business. In my industry, however, the truth is that we haven't made great strides in productivity lately. But that hasn't really hurt our bottom line: the fact is that there isn't much international competition in the stuff we sell, and my domestic rivals and I have a tacit understanding that it's in all of our interest not to get into price wars". Realistically, how likely is it that anyone will actually stand up and say this?
Perhaps, then, we should not be surprised that the new paradigm, which makes businesspeople feel good both about their economic prospects and about themselves, has spread so rapidly. But it is time to get serious: an economic doctrine, no matter how appealing, must be rejected if it cannot stand up to well-informed criticism. We would like to believe that America can grow much faster if only the Fed would let it; but all the evidence suggests that it cannot.