A Farewell to Alms: A Brief Economic History of the World (Princeton Economic History of the Western World)
by Gregory Clark
Reviewed by Robert M. Solow
New York Review of Books
Look at the graph reproduced on this page. It shows the material standard of living of the average English person over a span of eight centuries beginning in 1200. In 1800 the average English person was literally no better off, in food, clothing, and shelter, than in 1200. Although there had been ups and downs, there was no long-term improvement. Nobody would have been talking about "growth."
Then, in the early 1800s, not exactly suddenly but quite rapidly on this long time scale, it all changed. Less than two hundred years later, the previously stagnant standard of living had multiplied by six or more. That was the Industrial Revolution. Nothing closer to a Big Bang had occurred in human history. Now nobody talks about anything but growth in standard of living. This is not a new observation: the broad facts had long been noticed and discussed by Paul Bairoch, Carlo Cipolla, and other economic historians.
According to Gregory Clark, who teaches economic history at the University of California at Davis, this picture could be extended to cover the whole world and the thousands of years since the beginnings of settled agriculture. He even catches the reader's attention with a graph looking just like this one, but extending from 1000 BCE to the present. That graph, as far as one can tell, is largely made up. Only a handful of scattered and crude estimates of income per person exists for ancient times. On the other hand, Clark is very good at piecing together figures from here and there, including those from isolated groups of hunter-gatherers alive today. He makes a plausible case for the basic pattern: for thousands of years before the Industrial Revolution, there was essentially no sustained improvement in mankind's general material standard of living, nor was there much variation from place to place around the world. The Industrial Revolution made all the difference.
It is worth emphasizing that this is all about material standards, not cultural evolution or the "level of civilization." Shakespeare, Purcell, and Newton had done their work in England long before 1780 or 1800. But ordinary people still lived in hovels and wore rags.
Accepting this big picture, Clark is after deep answers. He proposes to explain why the stagnation lasted so long, why it ended with the close of the eighteenth century and not much earlier or later, why the Industrial Revolution occurred first in England and not in China or India or somewhere else, and why some parts of the world, such as Europe and the United States, continued to improve standards of living long after the Industrial Revolution, while other parts of the world, including Africa, stagnated or deteriorated. And he attempts to do all this not encyclopedically but by applying a neat, clean theory.
There are many reasons to wonder what it could possibly mean to figure out the "cause" of a unique historical event. Any large-scale historical event surely has many "causes," some more important than others. How are we to know that a proposed explanation has not left out something important or insisted on the significance of something that is really extraneous? To put the question more generally, how can there be a test of a theory about a unique historical event? There is by definition no second experiment to provide a check, and some sort of check is needed.
Historians sometimes try to resolve this problem by imagining that the event at hand is really a member of a larger class of events. Students of "revolution" will think of the French "case" or the Russian "case" or the Iranian; students of war will speak of the Franco-Prussian "case" or the "case" of World War I. We do not have any trouble with "cases" of strep throat, although there are always some differences among them. Wars are not so straightforward. One has to decide whether the Franco-Prussian War and the Thirty Years' War are really so similar in their etiology that a theory about the causes of one has to apply successfully to the other, or else be rejected.
When it comes to the Industrial Revolution, part of the premise is that it was a truly unique event, so that there is nothing to compare it with. Even if Germany had its own industrial revolution some decades later, it would not be comparable to the English one if only because the social and technological changes that figured importantly in England were already known and available for imitation or adaptation or rejection, not to mention the fact that England was already exporting industrial products.
It seems that there are two sorts of less satisfactory "tests" available for judging Clark's -- or anyone's -- answers to his key questions. First, any theory will have various building blocks, bits of intellectual machinery, some of which can be tested against a broader class of facts. Second, we simply have to judge whether the story he tells hangs together and passes the test of plausibility. In the background there is always the "null hypothesis" trial horse beloved of statisticians and econometricians and no one else. Suppose, they say, there was always a very small probability that an industrial revolution would occur in each century since the year one, in any one of ten eligible places. It was bound to happen somewhere in some century; heads just happened to come up for the first time in England in 1800 after 180 (eighteen centuries times ten places) consecutive appearances of tails. Any decent explanation has to beat that one. 2.
Clark explains the long era of pre-industrial stagnation as a "Malthusian trap." In other words, he interprets it as the working-out of the process analyzed by T.R. Malthus in his Essay on the Principle of Population (1798) and refined by the great classical economist David Ricardo, who published his Principles of Political Economy and Taxation in 1821. Other economic historians have come to the same conclusion, though you might not realize that from reading Clark's text. He pursues the Malthus-Ricardo model with a great intensity and unusually broad empirical scope.
Here is how Malthus-Ricardo works. In any society at any time there is a reasonably well defined notion of "subsistence," a level of income, essentially wages, just adequate to support a standard of living that will lead the average family to reproduce itself. Subsistence has a hard physiological basis in calories, necessary nutrients, protection from weather, and the like, but it can be modified by cultural factors, social norms, and customs. If wages happen to exceed subsistence for a while, because of good harvests or a reduction in the supply of labor through war or disease, normal mortality will decrease, fertility may rise, and the population will increase. But not for long: the pressure of a larger working population on a fixed supply of land and resources will force labor productivity and wages to fall. (That is the famous law of diminishing returns: the idea is that as more and more workers are squeezed onto the same area of land, at some point each additional worker will be able to add less output than his predecessor did, simply because he has a smaller share of the land to work with.)
This process cannot stop until wages are back to the subsistence level. The population will be bigger, but its members no better off than they were. If harvests then go back to normal, productivity and wages will fall below subsistence and the process just described will go into reverse: higher mortality will cause population to fall until productivity and wages return to the subsistence level and then stabilize.
This is a simple and powerful story, and it has just the implications needed to explain the grim preindustrial history. The key implication is that the material standard of living of any population is determined only by the level of subsistence. Incremental technological progress, which certainly took place in England -- and elsewhere -- between 1200 and 1780, does not seriously improve living standards; it just allows a larger population to be supported. Malthus thought about "prudential checks" to population -- e.g., practicing abstinence -- but wistfully. (The consequences of a large and rapid burst of technological progress -- not part of Malthus's story -- may be different; and that will be important when we come to the Industrial Revolution itself.) Even more paradoxically, progress in health care and sanitation may mean that the subsistence wage falls, because now mortality can be contained with even less in the way of nutrition, clothing, and shelter. In principle, then, medical progress could lead to lower wages and poorer conditions of life in those other respects.
This Malthus-Ricardo trap can certainly account for the long failure of living standards to rise. Clark exploits it with a lot of interesting and useful circumstantial detail. Here is an example. The sustainable level of the subsistence wage and standard of living can be higher in one country than another if the first can maintain lower fertility (at any given wage) than the second; population will stabilize at a higher wage, more or less by definition. Clark notes that preindustrial England did in fact have a lower fertility rate than other parts of the world in roughly the same income class. The mechanism was not primarily individual contraception but rather social norms: people married later in life and there were larger numbers of spinsters. One could speculate why this was the case: perhaps it was a response to the custom of primogeniture. The origin is interesting; but the fact itself could obviously be important in shaping England's response to the beginnings of the Industrial Revolution. 3.
The coming of the Industrial Revolution has attracted more attention from economic historians than the prolonged preindustrial stagnation. Clark's story has some things in common with other accounts, and some twists of its own. It is not always easy to distinguish between them because he is not exactly assiduous about crediting earlier work. I will try to identify the specifically Clarkian twists as they arise.
The Malthusian process works itself out slowly. The chain of causation from a rise in wages to decreased mortality and increased fertility, then from induced changes in population to changes in the supply of labor, and from the latter back to a fall in wages could take years or decades to complete itself. Now imagine what might happen if there is a series of major technological innovations. Productivity and wages may have time to rise by a lot; and they may keep rising even after population starts to increase, if the favorable effect of rising productivity from later innovations outweighs the depressing effect of higher labor supply.
But then new possibilities emerge. For instance the experience of rising incomes could encourage lower fertility, with an emphasis on fewer but better-educated children. This would figure in the theory as a drastic change in the social definition of "subsistence." The point is that the dismal Malthusian prospect could be postponed, perhaps forever. Something like this did happen in the West. The so-called "demographic transition" to lower fertility is usually dated to later in the nineteenth century; but the birth rate in England actually started to fall in the 1840s. As Clark points out in an interesting discussion, there was undoubtedly more to the demographic transition than the simple mechanism of rising incomes.
This part of the story depends on -- and obviously leaves unexplained -- the great series of technological innovations and economic adaptations that unfolded in England after 1780 and transformed the cotton textile industry, transportation, and mining; the most famous of these was the steam engine. Perhaps the most significant transformation, described but insufficiently emphasized by Clark, was the shift from men and animals as sources of energy to coal, and all that this meant for iron and steel, machinery and railroads. Why did all this happen when it did, and in England? This is where Clark departs, with fanfare, from generally accepted ideas.
The conventional wisdom, endorsed by many economic historians, most notably by Douglass North, points to a connected set of legal, economic, and social institutions that are thought to be necessary for or at least specially conducive to sustained economic growth. The most important are the rule of law itself, secure property rights, relatively untrammeled markets, and a degree of social mobility. They function by reducing the uncertainty surrounding saving, investment, and entrepreneurial activity, and by sharpening the incentives for able people to devote themselves to economic activity instead of violence and prayer. The Industrial Revolution happened when it did because these background conditions were met as they had not been met before; and England is where they were met soonest and most fully.
Clark claims that this rather plausible explanation cannot be the whole story or even the central story line. His main argument is that the standard institutional preconditions were already in place centuries before 1800 in England, and perhaps also in China, Japan, and elsewhere; but no industrial revolution happened. The key ingredient must have been something else. To bolster his case, for example, he produces a rather bizarre table, comparing England in 1300 and in 2000 with respect to each of twelve criteria of the sort proposed to poor countries today by the International Monetary Fund and the World Bank as important for development: low tax rates, modest social transfers such as welfare payments, stable money, low public debt, security of property, security of the person, social mobility, free markets for goods, labor, capital, and land, and finally incentives and rewards for people who advance knowledge, particularly by allowing the opportunity to profit from technological inventions. In Clark's estimation, England in 1300 gets a grade of Yes on ten of the twelve with question marks for personal security and for providing rewards for creating knowledge. England in 2000 gets a No on the first four criteria and on a free land market, and Yes on the seven remaining criteria. So 1300 wins by a little better than ten to seven.
This is not exactly convincing. For one thing, the World Bank's current exhortations to today's developing countries don't conform to what North and other advocates of the centrality of institutions have in mind. More important, this kind of casual judgment is hardly enough to contradict extensive scholarship about the character of the feudal economy. Clark's picture of medieval England as a free-market economy would be unfamiliar to many medieval historians. They view the medieval economy as a place where land was not owned privately but held by right of occupation; the allocation of factors of production such as labor was governed by a web of customary rights and obligations.
This system was widespread in England, though there were local exceptions. The market economy is believed to have developed gradually from this base, with some elements of the manorial system persisting until the eighteenth century. This standard description may rest too much on legal documents and not enough on practical realities; but to overturn it -- for instance to classify the enclosure movement as an irrelevancy -- would require more than the existence of a few land transactions.
Even on logical grounds, Clark's dismissal of the institution-centered view is too summary. The most his examples can show is that a collection of the "right" institutional practices is not a sufficient condition to cause an industrial revolution and subsequent economic growth. It may still be a necessary condition: there may be instances where secure property rights and open markets have failed to generate sustained, rapid economic growth; but there are no instances of sustained, rapid economic growth without those basic institutional prerequisites. This is not a negligible fact.
It is not clear from his book if Clark accepts the extreme view advanced by some economists that dysfunctional institutions cannot be a fundamental barrier to economic progress because inefficient institutions will eventually crumble in the process of social evolution. That this sometimes happens hardly needs arguing in the aftermath of the collapse of Soviet communism and central planning. But that does not make it a historical law. And even if feudal restrictions were bound to give way, sooner or later, here or there, that would not be a reason simply to ignore their effects while they lasted. 4.
The Industrial Revolution in England was not an overnight miracle; to date it from the 1780s is only a short-hand convention. Clark's calculations indicate a slow improvement in productive efficiency beginning as early as the seventeenth century, a significant acceleration from the 1760s to the 1860s, and only then an even sharper acceleration to modern rates of growth. He thinks that much of the apparent suddenness of the change comes from a roughly simultaneous but causally unrelated burst of population growth in England. The graph on page 38 shows income per person, so it avoids that illusion; the three phases picked out by Clark are visible there.
He quotes the historian Kenneth Pomeranz as saying that the densely populated core of China in 1800 was indistinguishable from northwestern Europe in 1800 in its "commercialization, commodification of goods, land and labor, market-driven growth, and adjustment by households of both fertility and labor allocation to economic trends." Much the same could have been said about Japan and parts of India. So why did lightning strike a small country of six million people in northwestern Europe?
Clark's view is that China and Japan were indeed on a track that might have led to an independent industrial revolution, but their evolution was moving more slowly than England's. One reason is that population grew slightly faster in both China and Japan than in England between 1300 and 1750; presumably the Malthusian trap held more tightly.
The second reason brings us to the main novelty, the true Clarkian twist. According to the now traditional view, the presence of institutions with the required qualities serves to create and maintain incentives that favor innovation, enterprise, and trade. Clark claims that these prerequisites were adequately present in medieval England, China, and Japan, but without the expected result. Something must have been missing, and he identifies it as the capacity or willingness of people to respond effectively to economic incentives. You might think of these as bourgeois virtues. So why were late-eighteenth-century Englishmen able and willing while medieval Englishmen and eighteenth-century Chinese and Japanese were not?
Clark resolves the puzzle in a novel way. In England -- and everyplace else -- in those early days before the demographic transition, the better off had more children than the less well off. More of the well-off women married; they lived longer and survived childbirth better; they were healthier and cleaner, and more of their children survived to grow up. Clark calls this pattern of differential fertility of the well off the "survival of the richest." It is a neat phrase that may seem to have the right Darwinian overtone, though not as cute as the title of the book itself. Demographic statistics for China and Japan are not nearly as good as for England, but they seem to exhibit the same pattern. However, the pattern of differential fertility in favor of the well-off appears to have been much more pronounced in England than elsewhere.
Why should this matter for the big question at hand? Clark infers that before the Industrial Revolution, there must have been a substantial amount of downward mobility from the higher-income groups. They could more than reproduce themselves, but they could not reproduce the same positions of status for all their offspring. Primogeniture would see to that; and in its absence, division of inheritances would have the same effect. Younger sons would have moved into somewhat lower strata of the English income distribution, not into poverty, of course, but below the very upper crust. Along with that inference goes the hypothesis that capacities and dispositions characteristic of upper-income groups became diffused into English society along with their bearers. Among these was the ability and willingness to respond to economic incentives. Clark writes: "Thus we may speculate that England's advantage lay in the rapid cultural, and potentially also genetic, diffusion of the values of the economically successful throughout society in the years 1200–1800."
Notice, by the way, that "and potentially also genetic." It, or something like it, recurs throughout successive references by Clark to this key hypothesis. I have no idea whether pecuniary aptitudes and attitudes have a genetic basis or are simply passed on in family and social settings as acceptable norms of behavior. It does not matter a bit for Clark's argument, but that is a reason to avoid insinuating a possible biological basis for this story without any evidence at all.
Leaving that aside, is Clark's answer to the "Why England" question plausible? The first thing to say is that the basis in evidence is slim. The excess fertility of the preindustrial English wealthy class seems well enough established. Whether this force was weaker in China, Japan, and India is harder to pin down, if only because their demography is less well documented in general. The notion of downward mobility of younger sons has arithmetic on its side, but not much else. It is a question that could in principle be the subject of research, but we are not offered the results of research. The general idea is not self-evident. The pattern here is more frequent among the aristocracy; but the conventional picture is that the respectable destinations for younger sons were the army and the Church, neither of them a hotbed of pragmatic, pecuniary rationality.
The nonaristocratic rich are more interesting, because they are more likely to have taught the requisite mercantile virtues to their offspring. But the downward mobility is not so inevitable among them. A fortune can be divided as a dukedom cannot; and half of a large fortune may still be a small fortune. Clark's story about diffusion may be true nonetheless, but it cannot be taken for granted. One would really like to see some direct evidence that the trait of responsiveness to pecuniary incentives had spread more widely in English society than it had earlier or elsewhere by the end of the eighteenth century. Even if it had, only the most minute micro-research could hope to demonstrate that this was the result of the mechanism proposed by Clark. As things stand, he has built a very heavy structure on a very narrow base. 5.
The last section of the book, called (after Pomeranz) "The Great Divergence," starts off fairly well, but then peters out. During the long Malthusian plateau, average living standards differed somewhat from one part of the world to another, but the range of variation was not terribly wide. In each place, living standards were governed by the level of the subsistence wage, and so in large part by common human physiology. After the Industrial Revolution, however, some parts of the world took off into an era of sustained economic growth that is still going on and may occasionally have accelerated, while other parts of the world have stagnated or even declined, with living standards still close to preindustrial levels.
In 1800, Western Europe, North America, and Oceania -- including the Pacific islands of Polynesia, Micronesia, and Melanesia -- had 12 percent of the world's population and 27 percent of the world's income. In 2000 they still had 12 percent of the population -- relatively more of it in North America -- but 45 percent of the income. To take only the extreme contrast, Africa went from 7 percent of the population and 9 percent of the income to 13 percent of the population and 4 percent of the income. That is surely a great divergence. It is in some ways as hard to explain as the Industrial Revolution itself. (The story in Asia is more complicated, both temporally and geographically, but the details are not relevant here.)
Clark shows rather convincingly that the main source of this shocking gap is a large difference in productive efficiency, or what economists call "total factor productivity." That is jargon for the quantity of output that the economy is able to generate per unit of all input, including labor, capital, and natural resources. He goes on to make a reasonable, if sketchy, case that the primary culprit is not lack of access to technology or capital, both of which are available to poor countries that can use them effectively today, and have been available for a long time in a few places, especially imperial dependencies.
Nor, he claims, is it mainly a lack of manpower with the necessary skills. He also absolves management failure, on the ground that textile factories in colonial India with British managers did no better than those with Indian managers. (This is thin evidence, but perhaps other examples would show the same thing.) In the end, Clark puts the finger on the workers -- not their skills or native ability but their attitudes and aptitudes, their willingness to show up on time, work hard with little supervision, exercise local ingenuity, and so on.
In this context, too, he dismisses the prevailing view that dysfunctional or corrupt economic, social, and political institutions explain the divergence in efficiency. He reasons: if a factory in a poor country produces less than an essentially identical factory in a rich country, how can that be attributed to institutional failure? Here, too, he may be a little hasty. Cronyism at the top, failure to enforce laws, promotion by favoritism, inequitable taxation, capricious hiring and firing -- all those practices could easily breed disaffection or even sabotage, and thus inefficient production. Maybe.
Clark's pessimism about closing the gap between the successful and less successful economies may derive from the belief that nothing much can change unless and until the mercantile and industrial virtues seep down into a large part of the population, as he thinks they did in preindustrial England. That could be a long wait. If that is his basic belief, it would seem to be roundly contradicted by the extraordinary sustained growth of China and, a bit more recently, India. Embarrassingly for Clark, both of those success stories seem to have been set off by institutional changes, in particular moves away from centralized control and toward an open-market economy.
In an extensive industry-by-industry comparison of productive performance in Brazil, India, Korea, the US, and some European countries, the McKinsey Global Institute arrived at a conclusion somewhat different from Clark's. It found that large disadvantages in efficiency are traceable more often to failures of internal organization in the leading firms than to deficiencies of technology, capital, or workers' skills. This is not dramatically at odds with Clark's view, but puts much more emphasis on lack of sharp incentives for management than on the attitudes of workers. It is for that reason more optimistic about the prospects for change.
Toward the end of his book Clark spends a few paragraphs in stereotypical complaint about how modern economic theory has lost touch with any reality; its endless refinements are useless for dealing with the basic problems of economic growth that engage him and the world. This amounts to a severe bite at the hand that feeds him, since much of this sometimes fascinating and thought-provoking -- and sometimes irritating -- book is based quite precisely on applying the insights and methods of modern economic theory.
Robert M. Solow, Institute Professor Emeritus of Economics at MIT, won the 1987 Nobel Prize in Economics. His most recent book is Work and Welfare.