The standard measure of a national economy’s overall performance is its real gross domestic product (GDP). This is just the sum of all the goods and services produced during the year (except those used up in further production), valued at going market prices and then corrected for price inflation. The graph on this page shows yearly real GDP per person in the population of the United States from 1948 to 2000.

Two characteristics stand out clearly from the graph. The first is a rising trend, the statistical image of economic growth. If you were to pencil in a smooth curve it would rise from about $11,000 in 1948 to about $34,000 in 2000, or at 2.19 percent a year on average. But the rate of growth is not constant; there are intervals of faster and slower growth (for instance from 1990 to 2000 and from 1973 to 1983). The second characteristic is that real GDP per person fluctuates within this larger trend in what are often called business cycles, though the fluctuations are not very regular. The existence of business cycles suggests that it would be more accurate to plot the trend through the cyclical peaks than through the midpoints. That would reflect the commonly accepted notion that real GDP is more often below the economy’s normal capacity to produce than above it. This second sort of trend, from peak to peak, is usually called the economy’s “potential” output: not an absolute limit, but a level of production that cannot be exceeded for long without strain and likely inflation.

The distinction between trend and fluctuations is not merely descriptive. They represent different mechanisms. The conventional wisdom—which I believe to be correct—takes the general upward trend to be driven mainly by factors on the supply side: by improvements in the education, training, and skills of workers; by technological innovation; and by increases in the stock of machinery and equipment per worker as well as the replacement of obsolete equipment with new versions adapted to the latest and most productive technology.1 Of course education, capital investment, and innovation are not elemental forces. They result from choices and respond to pecuniary and other incentives.

The short-term fluctuations, on the other hand, are driven primarily by demand-side forces: by changes in the willingness and ability of families and firms (and foreigners) to buy goods and services. When real GDP falls below potential, it is not because some factor has diminished the economy’s capacity to produce, but because producers and sellers cannot find enough willing buyers at the prices they are charging. (Do not leap to the conclusion that a round of price reductions would cure this problem. Lower prices would mean lower incomes generally, and perhaps the expectation of further price reductions. Both are likely to lead to a further weakening of demand. The result might be little or no improvement in economic conditions.)

This way of looking at the trend in economic growth and the fluctuations within it has implications for public policy. If the goal is to steepen the growth trend, then policy should be directed at improving skills (often described as investment in human capital), encouragement of research and development, and promotion of capital investment. If the goal is to end a recession, a short-term slowdown of economic growth, and move the economy closer to its potential output, then policy should aim to increase private and public spending on goods and services for private consumption, public purposes, and, once again, business investment, housing, and other durable goods.

That business investment can stimulate both short- and long-term growth is a reminder that demand-side and supply-side forces are not wholly independent of each other. Some kinds of demand add to potential output; others do not. For example, spending to build a new factory increases potential output but spending to build a recreation room does not; nor does dining in a restaurant. Some supply-side changes will stimulate demand, as when new technology induces firms to spend on up-to-date computers and induces families to increase their total spending, especially on new goods. Economics is full of such two-way chains of causation.


The title of Jeff Madrick’s new book is Why Economies Grow, which suggests that it is mainly about the causes of the trend in growth: why it may be high or low, rising steeply or almost flat. But the subtitle is “The Forces That Shape Prosperity and How To Get Them Working Again,” and that sounds more like a tract on avoiding or limiting recessions. As I have already pointed out, there are connections between the two. The most important is that a chronically depressed economy may eventually put a damper on the search for new production technology and on the investment in new equipment that often goes along with it.

In fact, Madrick sometimes fails to be precise about the distinction between trend and fluctuations. This is hardly surprising: when you read in a newspaper that the economy “grew” last quarter at an annual rate of 4 percent, no one bothers to explain that some part of that increment represents a permanent rise in trend, while the rest marks a temporary upward movement in the business cycle. When Madrick, or anyone else, passes from one to the other without noticing, the result is occasional confusion, especially about the appropriate policies. But sometimes he hits a target that needs to be hit. This mixture of two themes causes trouble; but it allows the author to score some points.


The good moments occur, as you might expect, when Madrick validly insists that there are interactions between business cycle and growth trend, between the demand side and the supply side: for instance that businesses respond to demand by searching for new products, and that people try to equip themselves for jobs that are actually available. Some modern economists—the school is centered at the Universities of Chicago and Minnesota, but they are widely represented and respected—tend to (over)emphasize the self-balancing properties of the market economy, and like to play down the occurrence of persistent gaps between potential output and actual output. This allows them to claim that, on the whole, the economy follows the best path it can in the light of (a) the preferences of the inhabitants (with respect to work and leisure, and with respect to their present and future satisfactions) and (b) known technological possibilities.

For this school of thought, temporary aberrations in economic growth are the result either of unpredictable variations in tastes and technology or of misguided interventions by governments into the workings of the system in attempts to smooth fluctuations that cannot and should not be smoothed. Madrick is on strong ground when he urges that the development of the supply-side forces of human capital, physical capital, and technological progress are at least partially governed by necessarily political choices about public investment in infrastructure, education, and health care, and by the pressure of demand in expanding markets. I will come back to this.

The bad moments occur sometimes in attempts to find general principles about economic growth, and sometimes in inadequate or unproved analysis of more concrete aspects of the interplay between supply and demand. Several general questions are suggested by the title of the book: if we knew in any detail “why economies grow,” we would be close to holding the key to the economic universe.

Madrick seems to think, as a matter of principle, that there must be a single answer to the question. He writes as if there is necessarily a chain of causes and effects that can be written down from left to right; the task is to find the left-most item in this chain, what he sometimes calls the First Mover. This is altogether reminiscent of theological disquisitions. Causal chains in economics have a habit of doubling back on themselves. If there is a causal mechanism leading from A to B, there is very likely to be another one leading from B to A. (Malthus, for example, thought that high wages induced rising population, and also that a large population depressed wages.) Any attempt to work backward from right to left may very well end in such a web of mutual causation.

It is only fair to say that Madrick recognizes this in principle. There are sentences like: “In fact, many factors cause and enhance economic growth, and in turn affect each other.” Or again: “Most of these [a long list] are necessary conditions for growth…. All are both cause and consequence.”

But then one reads much sharper statements: “This book argues that the growth of markets…was the predominant factor in Western economic development.” Further: “…Market size and the dissemination of information are furthest left and are closest to first movers or true leaders. In addition, the first of these is the size and expansion of markets for goods and services.”

I am in the awkward position of thinking that the weak message is too weak and the strong message too strong. One important school in modern economics—the one I have already mentioned—seriously underplays the role of the demand side in economic growth. I take Madrick’s opposing message to be the strong one, but I think he does not get the balance right. The focus on market size is excessive.

With this starting point, Madrick interprets the standard theory of economic growth as a case of technological determinism. It has been found, as an empirical matter, that technological progress is the largest contributor to the upward trend of real GDP per person in modern industrial economies. Since many economists stop there, Madrick argues that they must regard technology as the First Mover. I think this overstates the case.


Particularly useful here is the economist’s concept of an “exogenous” variable, a factor whose effects are provisionally taken as given in order to limit the scope of an analysis. This device can work, provided that any causal chains leading from the things that are being analyzed—the “endogenous” variables—back to the exogenous variables are relatively weak. If I am studying the demand for restaurant meals, I will notice that it surely depends on—among other things—the degree of prosperity of the national economy. People eat out when they are flush. I know, of course, that the restaurant industry itself is one small contributor to that degree of prosperity; but the contribution is small enough for the error involved in neglecting it to be minor. For this reason, I can agree to take the degree of prosperity as exogenous, and therefore crucial for understanding the ups and downs of the restaurant industry as long as I am thinking about the market for restaurant meals. But that does not make it a First Mover. On the other hand, an analysis of the labor market, or even the automobile market, would have to be structured quite differently because what happens to jobs and wages, or what happens to car sales, has economy-wide effects that are too big to be neglected. In such cases we can’t take general prosperity as given.

Besides, there is a large literature on the theory of economic growth that does try to understand the causal chains leading to the pace of technological progress.2 I have my doubts about the success of that intellectual enterprise, just as I have about Madrick’s attempt to dig behind the advance of technology, but that is neither here nor there. Madrick’s charge that economic theorists are guilty of technological determinism is misplaced.

Madrick has his own proposal for going one step to the “left” of technology. His candidate for the First Mover of economic growth is the “expansion of markets.” There is an element of truth here, but not nearly enough. He has in mind that a strong, perhaps spontaneous, surge in the demand for X, especially one that puts pressure on the economy’s capacity to produce X, is likely to motivate the X-industry to seek new and more productive ways to produce X. Necessity is the mother of invention. Nobody would deny this.

But there are at least two serious problems with the argument. The first is that invention needs a father too: there will be occasions when the search for new technology comes up empty. Then the price of X will rise enough to choke off the rising demand. An expanding market by itself will not do the trick. Successful growth requires a web of causes, one of which may be the “ripeness” of the needed science and technology. An obvious example is the search for a low-cost source of energy to replace gasoline in the automobile. It may eventually succeed, but has not so far.

The second problem is more a matter of logic. On the scale of the national economy, not just a single industry, the only thing we can mean by “an expanding market” is rising real GDP. We measure a nation’s economic growth by the increase in its real GDP. So Madrick’s First-Mover hypothesis about expanding markets is altogether too close to the tautology that the prime cause of rising real GDP is rising real GDP. The escape from this box is to give up the extreme and unnecessary notion of a First Cause. Economies grow because of a complex interplay of exogenous variables and economic incentives that leads to technological innovation and investment in human and physical capital. The expansion of many markets, and the contraction of some, is only part of the cause, and only part of the effect.

There is a third problem that Madrick could not have known about, and that came as a surprise to me too. In a forthcoming paper,3 the economic historian Alexander Field, of Santa Clara University, argues convincingly that the fastest technological progress in the twentieth-century United States took place in the “decade” between 1929 and 1941. (Among the technologies that were improved dramatically during this period, for example, were chemical engineering—Plexiglas, Teflon, nylon—and heavy construction—dams, highways.) These were mostly depression years, and the period ended before the full mobilization for war. Markets were not expanding; quite the opposite. There is even a case to be made that the technological improvements stored up during that interval fueled the period of rapid growth immediately following the war.


Madrick’s attempts to shoehorn particular historical episodes into simple patterns in which market demand produces higher growth—“demand-pull” as economists put it—are hardly convincing. I will not dwell on his discussion of the industrial revolution of the late eighteenth and early nineteenth centuries. This is a staple of economic historians and there is an enormous literature on both sides of every question; and in any case the answers have little to tell us about practical recipes for sustaining growth in the twenty-first-century United States.

It is true, for example, that Watt’s steam engine allowed the railroads to happen, but it was not the prospect of the railroad age and the demand—the “growth of markets” as Madrick puts it—for better transport that created the steam engine. The need to pump water out of mines may have been more pressing but less productive. Nor did a preexisting urgent demand for cotton underdrawers inspire the textile inventions of the eighteenth century. It is more enlightening to realize that it was the technology-induced fall in the price of cotton cloth that made the consumption of all those cotton underdrawers economically possible.

It does not follow that the demand-pull aspect of innovation is unimportant; but it is not nearly the whole story. I share Madrick’s belief that modern economic theory tends to underplay the importance of strong and profitable markets, though not necessarily expanding markets, in eliciting the innovation and investment that eventually move the potential trend. (And don’t forget the reverse causation, from ongoing innovation to strong markets.) One-size-fits-all explanations, such as one based on expanding markets, are unlikely to be good enough.

Here is a more recent and relevant example of Madrick’s tendency to overinterpret events his way. Between 1950 and the early 1970s, productivity (i.e., output per hour worked in the business sector) grew very rapidly in the US, faster even than in the last half of the 1990s. Then, for the next twenty-five years, productivity growth slowed to a crawl. Why did the “productivity slowdown” happen? This is an especially poignant question for the author of a book called Why Economies Grow, which concludes with two chapters on “Making America Grow.” If expanding markets are the deep cause of improving technology and growing productivity, a productivity slowdown must have its source in forces that work against the expansion of markets.

Madrick’s hypothesis is that sometime around 1970 consumers became rich enough and educated enough to demand goods that were more narrowly tailored to their individual preferences. The 1970s and 1980s were thus, he argues, the age of the “niche market.” Whether in consumer electronics, packaged food, or car models, a more diverse range of products was made for increasingly choosy consumers. Production runs had to be smaller and product life-cycles shorter. Markets became fragmented; they could not continue to expand as they had been doing. Imports reduced the sales of domestic producers. Economies of large scale were less available than in the past. The economy’s capacity to innovate was spread too thin, and so productivity-growth slowed down.

There is obviously some anecdotal truth to this picture, but Madrick provides very little serious evidence to back up his interpretation of such an important test case.4 There are big holes in his story. The most striking is the fact that the post-1970 productivity slowdown was worldwide. It affected Japan and Germany at least as much as the US and it was felt in most industrial countries. But the story of increasingly finicky consumers does not make much sense for them. Besides, Germany and Japan were actually expanding their markets through exports.

There are also conceptual problems. If consumers were so eager for niche products, they were presumably willing to pay for them. Since productivity is measured by inflation-corrected value added per hour, successes in meeting consumer needs should in principle be fully reflected in the productivity statistics. If they were not, then part of the slowdown was merely a measurement error. Yet another puzzle arises from the fact that it was in part technological innovation in the form of computer-controlled production lines that made the shift to tailored products possible; this should have at least tempered any loss of productivity.

My own view is that much of the productivity slowdown is still unexplained. It is not even certain how much of it needs to be explained: Why should we assume that productivity will rise at a more or less steady rate unless something special happens? By the way, Madrick has no convincing explanation of the rather sudden and unexpected acceleration of productivity in the US after 1995, which was not shared by other industrial countries. Very likely some of it was the late-arriving benefits of information technology; but Madrick is rightly skeptical about exaggerated claims about the effects of that technology in creating a new economy. Productivity growth has been surprisingly well maintained even during the past two years of recession and hesitant recovery, a phase of the business cycle in which it is usually fairly weak. Evidently the productivity story is still unfolding.


The last third of Madrick’s book concentrates more directly on the contemporary US economy and its prospects. The logical connection to the earlier text is sketchy, but the immediacy of the issues he discusses makes up for that. Madrick lists a dozen “major economic challenges in the future.” The odd-numbered ones are high levels of private debt, stagnating stock prices, intense global competition, the rising costs of education and health care, an aging population, and more educationally demanding work. This is too large and diverse an agenda to be discussed compactly, and—confusingly—it is as much related to the short run as to the long.

In this context, however, Madrick’s emphasis on expanding markets has a useful function, if it is properly interpreted. He notes that there are two separate and oversimplified responses to any such list of future needs. One is a more or less exclusive focus on increasing national savings by such measures as expanded IRAs and favored tax treatment of dividends, which will in turn—it is too simply assumed—lead to more investment. The other is a more or less exclusive focus on improving technological research and education. He may be slightly exaggerating the exclusivity, but never mind. It is still useful to insist that these supply-side-oriented policies need to be accompanied by attention to the demand side, whether in the form of increased consumer spending or government investment, or the kind of pressure of demand on existing productive capacity that will encourage business investment.

Market economies can run into bad patches of weak demand: look at Japan and Germany today. Saving will not be translated automatically into high investment, and technological capacity will not be translated automatically into active innovation unless the demand for old and new goods and services is brisk and enduring. Madrick’s emphasis on the encouragement of expanding markets has validity here, and its persuasiveness does not depend on claims concerning a First Mover.

Madrick sorts his ideas about policy into four basic propositions. The first, of course, is the need for growing internal markets. (I trust that the casual insertion of “internal” is not a sop to protectionists.) The second is that inequality has gone too far in the contemporary US, and has become a drag on productivity and growth. The third is that a society that neglects public investment loses key benefits and may weaken the prospects for growth. The fourth is that nostalgia is dangerous. Growth requires change, and change can be uncomfortable. The necessary changes can have to do with the role of government and business or with expectations about the nature of a working life, or about other things.

These propositions seem entirely reasonable to me. The details, however, are a mixed bag. With so many items to cover, the treatment is inevitably cursory; for me to review them would simply be more of the same. I will focus instead on the issue of inequality:

Inequality of income undermines growth because it weakens the demand for goods and services, as well as most Americans’ ability to save…. [It] also undermines people’s ability to invest in themselves through education and healthcare.

To this brief passage one could add that inequality passed down from one generation to the next is likely to rob society of entrepreneurial, scientific, and other talents that it badly needs. This sounds good, and I certainly share Madrick’s sentiment. But what is the evidence that inequality undermines growth?

As it happens, there has been some research on this very issue and, for a while, it went Madrick’s way. The procedure was usually something like this. Collect statistics for a substantial number of countries, first of all on their degrees of income inequality and on their growth rates during a common time interval, such as between 1960 and 1990. Add country-by-country information on other possibly relevant characteristics: level of income at the start of the period, literacy rates, investment rates, measures of political stability, measures of monopolistic market distortion, measures of the size of government, and so on. Then use standard statistical techniques to answer the question: Did countries with more inequality grow faster or slower than countries with less inequality, after one allows for the influence of all those other measured characteristics?

The results of those studies were not uniform, but, after one allows for the difference in other factors, the answer that generally emerged was that less unequal countries grew faster, allowing for differences in other factors. Still, there were problems. The results were too sensitive to minor changes in the analysis, like slightly different time periods, slightly different choices of “other characteristics,” and the like. Besides, there is nothing in this kind of correlation to tell you what is causing what: Do fast-growing countries reduce income inequality or does lower income inequality favor growth?

And now the correlation itself has been powerfully challenged. In an article published two and a half years ago, Kristin Forbes revisited the question, using better data on inequality.5 In addition, her evidence covers forty-five countries, each observed during a sequence of five-year intervals from 1966 to 1995. The importance of this extension is that the analysis can extract implications from changes over time within each country, as well as from differences among countries at some fixed time. Forbes’s conclusion is that, after taking into account all the other influences she was able to measure and include, she found that greater inequality was associated with faster growth.

Should Madrick retreat? Not necessarily: this is a complicated issue, possibly not yet settled. I would not close the books until the causal mechanisms accounting for growth are pinned down, and that is never easy. Besides, much of what he advocates in this book is valuable, and ignored or denied in these conservative times. That strong demand promotes progress, that inequality wastes human resources, that public investment has important uses—these are all cases worth making, even if the connection to economic growth is more intricate and less certain than he realizes.

My own domestic agenda would not differ so much from Madrick’s, but I would sort it out differently. The case for strong and expanding markets is primarily in the here and now. Chronic slack in the economy wastes labor and capital resources that are dissipated if they are not used in the present. For the longer run, if growth is what we want, the goal of public policy should be to create the institutions and incentives that will divert production from consumption to investment in the broadest sense: public and private investment, human and physical investment, basic and applied research. As for inequality, I suspect its causal connection to the trend in growth is weak. But at a time when impersonal economic forces seem to be pushing by themselves in the direction of widening inequality, for public policy to be doing the same thing is not a technical mistake but a moral disaster.

This Issue

July 3, 2003