The economic historians have not so far discovered any single material factor essential to British industrialization. A long time ago Gerschenkron argued that the notion of essential prerequisites for economic growth, single or multiple, is one that needs skeptical handling.¹ He gave examples from the industrialization of Germany, Italy, and Russia that showed substitutes for what looked like prerequisites from the British history. The big banks in Germany in the 1870s and state enterprises in Russia in the 1890s, he claimed, substituted for vigor in entrepreneurship and honesty in trade that were by 1750 taken for granted in Britain.

Gerschenkron’s economic metaphor that one thing can “substitute” for another applies to Britain itself as much as to the other countries (there is some doubt, actually, concerning the other countries). Economists believe with good reason that there is more than one way to skin a cat. If foreign trade or entrepreneurship or saving had been lacking, the economist’s argument goes, other impulses to growth could have taken their place (with a loss, but usually a modest one). A vigorous domestic trade or a single-minded government or a forced saving from the taxation of agriculture could take the place of the British ideal of the merchant left alone by government to reinvest his profits in a cotton factory.

Transportation, for example, is often cast in the hero’s role. The static tale is most easily criticized. Canals carrying coal and wheat to the docks at a lower price than cartage, better public roads bringing coaching times down to a mere day from London to York, and then the railway steaming into every market town were all of course Good Things. But their effect on national income can be shown to be small.

The way it can be shown is a technique much used by economists, which will be worked hard here. Think of a sector such as transportation as having a certain share of national income and a certain percentage increase in productivity. If you multiply the two you have calculated the national gain from the increase in productvity. The technique depends on the economist’s metaphor of the economy as a “production function,” a sort of sausage machine of inputs yielding outputs—the Q = F (K, L) mentioned earlier. The robustness of the calculation is a consequence of what is known informally among economists as Harberger’s Law (after A. C. Harberger, a Nobel-worthy economist at Chicago and then UCLA, famous for such calculations). That is, if one calculates a gain amounting to some fraction from a sector that amounts to again a fraction of the national economy one is in effect multiplying a fraction by a fraction. Suppose G percent of gain comes from a sector with a share of s percent of national income. It follows from highly advanced mathematics (don’t try this at home) that the resulting fraction, G times s, is smaller than either of its terms, since both are fractions less than 1.0. For most sectors and most events—here is the crucial point that will make the technique work for the story here—the outcome is a small fraction when set beside the 1,500 percentage points of growth to be explained 1780 to the present, or even beside the 100 percentage points of growth to be explained 1780 to 1860.

Transportation is never more than 10 percent of national income—in Britain it was something like 6 percent 1780 1860. Britain was well supplied with good harbors for its massive coastwise transportation, and in England the rivers flowed gently like sweet Afton when large enough for traffic at all. Mother Nature had given Britain a low cost of transportation by water, even when the waterways were unimproved by river dredging and stone-built harbors. The further lowering of the cost by introducing canals and railways would yield an improvement of, say, 50 percent (a figure easily justified by looking at freight rates and price differentials). But the 50 percent fall in transport cost applies only to the portion of traffic not carried on unimproved water—say likewise 50 percent. By Harberger’s Law, 50 percent of 50 percent of 10 percent will save a mere 2.5 percent of national income. One would welcome a tiny share of 2.5 percent of national income as one’s personal income; and even spread among the population it is not to be scorned. But it is not by itself the stuff of “revolution,” and it is nothing like 1500 percent.

Yet did not transportation above all have “dynamic” effects? It seems not, though historians and economists have quarreled over the matter and it would be premature to claim that the case is entirely settled.² The most powerful case against the importance of dynamic effects was mounted by Robert Fogel on a long evening in Toronto against the speculations of the economic historian Paul David to the contrary.³ David had harshly criticized on “dynamic” grounds Fogel’s calculation of social saving in Fogel’s book of 1964, Railroads and American Economic Growth. In a 54-page rebuttal (which Fogel read that night after dinner in its entirety), he calculated the possible dynamic effects and found them small.⁴

In framing the calculation a few points need to be kept in mind. For one thing the attribution of dynamism sometimes turns out to be erroneous double counting of the static effect. Historians will sometimes observe with an air of showing the large effects of transport that the canals or the railways caused transport costs to fall and increased the value of coal mines or made possible larger factories—”dynamic” effects (the word is protean). But the coal lands and factories were made more valuable simply because the cost of transporting their outputs was lower. The higher rents or the larger markets are alternative means of measuring what is the same thing, the fall in the cost of transporting coal or pottery or beer.⁵ To add them together is to count the same effect twice.

For another, some of the dynamic effects would themselves depend on the size of the static, 2.5 percent effect. For example, one “dynamic” effect is that new income is saved, to be reinvested, pushing incomes up still further, by the much honored logic of “accumulate, accumulate.” The trouble is that the additional income in the first round is very small. A 2.5 percent first round leads to a much smaller second, and a still smaller third round.

And if it would lead to a bigger second and still bigger third round, there’s something strange about the model—perhaps “economies of scale” have been thrown into the model at just the right time to make it explosive, as in modern growth theory. In that case anything, simply anything, could have started off the dynamo, and at any time from Tyre and Rome to the present. Explosive models that give no reason for becoming explosive exactly in 1700 or 1800 have not explained the sharpest upturn of real incomes per head in history. They have merely renamed the upturn “economies of scale.” The new growth theory in economics revives an idea of Alfred Marshall in 1919 and of Allyn Young in 1928 that bigger is better, if you have smart neighbors, especially in its economies-of-scale and especially in its economies-of-neighborhood form initiated by among others Paul Krugman and David Romer and Charles Sabel (the latter two, I am pleased to note, were my students as undergraduates; I wish Krugman had been, too: I would have taught him some intellectual humility). For example, people gathered in cities sometimes do a little better. But sometimes a little worse. The theories anyway are often a trifle exiguous. Though humility is not Krugman’s most prominent virtue, he does charmingly admit that his version of the “new economic geography” shares some of these handicaps. He quotes against himself a “sarcastic physicist” as remarking, “So what you’re saying is that firms agglomerate [in cities; or in economic growth] because of agglomeration effects?”⁶ And measurements of such effects show them to be small, on the order of perhaps 10 percent. That’s enough to explain why Chicago beat out Moline or St. Louis, and so explains the geography of production and consumption. A good thing to know. Interesting. Let me show you the mathematical model. But the 10 percent does not go very far to explaining an enrichment of 100 percent or 1500 percent.

And there’s a deeper problem with transport dynamics. Such truly dynamic effects as externalities leading to agglomeration effects may arise from expensive as much as from cheap transportation. Forcing more industry into London in the early nineteenth century—imagine for example humming cotton mills down at Kew in lilac time—might have achieved economies of scale in 1776 or 1815 which were in the event dissipated by the country locations chosen under the regime of low transport costs (and, to be serious about the history, without the constraints of regulations in the literal City of London or its westward extensions). In fact, precisely because of its advantages in transport costs to its numerous consumers at home and abroad, greater London before the eighteenth century was the manufacturing center of England, having fully ten percent of the English population in the mid-seventeenth century. Once you introduce the possibility of economies of scale, in other words, the balance of swings and roundabouts has to be calculated, not merely asserted—after all, that is the anti-invisible-hand point of industrial policy and infant-industry protection and path dependence and other allegedly practical implications of what economists call “non-convexities.” Manufacturers did relocate to Manchester and Birmingham at the call of a little cheaper labor and a little cheaper transport. So?

Sector by sector the older heroes have fallen before the research of the economists and historians. Marx put great emphasis for instance on the enclosure of open fields, that is, the dissolution of the medieval agricultural community and its translation into compact, individualistic farms. Marx claimed that enclosure enriched the investing classes and drove workers into the hands of industrialists. Most educated people believe the tale as gospel truth, and are quite sure that a lot of industrial investment came from the profits from enclosures, and that the workforce for industrialization was “pushed off the land.” Sellar and Yeatman capture the bits we can remember: “there was an Agricultural Revolution which was caused by the invention of turnips and the discovery that Trespassers could be Prosecuted. This was a Good Thing, too, because previously the Land has all been rather common, and it was called the Enclosure movement and was the origin of Keeping off the Grass, . . . [culminating] in the vast Royal Enclosure at Ascot.”⁷

But by now several generations of agricultural historians have argued (contrary to the Fabian theme first articulated in 1911, following Marx) that eighteenth-century enclosures were in many ways equitable and did not drive people out of the villages.⁸ True, Parliament became in the eighteenth century an executive committee of the landed classes, and made the overturning of the old forms of agriculture easier than it had been under earlier and royal supervision. Oliver Goldsmith lamenting the allegedly deserted village wrote in 1770 that “Those fenceless fields the sons of wealth divide,/ And even the bare worn common is denied.” But contrary to the pastoralism of the poem—which reflects aristocratic traditions in poetry back to Horace and Theocritus more than evidence from the English countryside—the commons was usually purchased rather than stolen from the goose. One can point with sympathy to the damaging of numerous holders of traditional rights without also believing what is false—that industrialization in any important way depended on the taking of rights from cottagers to gather firewood on the commons. Industrialization, after all, occurred first in regions long enclosed and far to the north and west, such as Lancashire or Warwickshire, not in the East Midlands or East Anglia or the South, the places where the Parliamentary acts of the eighteenth century did transform many villages. And in such freshly enclosed areas the local populations increased after enclosure.

The result of enclosure was a little more efficient agriculture. Perhaps the efficiency is why enclosure increased employment, because it raised the quantity demanded for now more productive workers. But was enclosure therefore, to take the optimistic view, the hero of the new industrial age? By no means. Nothing much would have changed had English agriculture, like agricultures on the Continent, resisted enclosure until a century after industrialization.⁹ The productivity changes were small, perhaps a 10 percent advantage of an enclosed village over an open-field village, and the profits small in national terms, though a high percentage of the previous rents (about doubling, which explains why they happened: that’s the most reliable method of calculating the productivity change).¹⁰ Agriculture was a large fraction of national income (shrunk perhaps to a third by 1800), but the share of land to be enclosed was only half of the land of England (the rest were those “regions long enclosed”).¹¹ Harberger’s Law asserts itself again: ( 1/3) (½) (10 percent) = 1.6 percent of national income was to be gained from the enclosure of open fields. Improved road surfaces around and through the enclosing villages might well have been more important than the rearranging of scattered plots on which most historical attention has been lavished (straightening and resurfacing of roads went along with enclosure, but in the historical literature is seldom stressed).

Nor was Adam Smith correct that the wealth of the nation depended on the division of labor. To be sure, the economy specialized. Ann Kussmaul’s pioneering work on rural specialization shows it happening from the sixteenth century onward.¹² Maxine Berg and Pat Hudson have emphasized that modern factories need not have been large, yet the factories nonetheless were closely divided in their labor.¹³ Most enterprises were tiny, and accomplished the division of labor through the market, as Smith averred. It has long been known that metal working in Birmingham and the Black Country was broken down into hundreds of tiny firms, anticipating by two centuries the “Japanese” techniques of just in time inventory and detailed sub contracting. A division of labor certainly did happen, widely.

That is to say, the proper dividing of labor, like the proper marshalling of transport and enclosure, made the economy more efficient. Gains were to be had, which suggests why they were seized (compare agglomeration effects explaining specialization of, say, Chicago in meat packing). French engineers at the time were amazed by the division of labor in Britain. But the division of labor was much noted in China at the time as well, yet did not result in an industrial revolution. And a new technique of specialization, like an advantage from agglomeration in Chicago, can be profitable to adopt yet lead to only a small effect on productivity nationally. The modest, if by no means unimportant, productivity changes from the puddling and rolling of iron amounted 1780 to 1860 to about 0.9 percent per year in the industry, which itself was not gigantic.¹⁴ The national gains were modest in the absence of dynamic effects, because the static gains from more complete specialization are limited by Harberger’s Law.

Consider the following extreme thought experiment. Specialization in the absence of technological change can be viewed as the undoing of bad locations for production. Some of the heavy clay soil of the Midlands was put down to grazing, for example, which suited it better than wheat. Or the labor of the Highlands was ripped off the land, to find better employment—higher wages, if less Gaelic spoken—in Glasgow or Nova Scotia or North Carolina. The size of the reallocation effect can be calculated, à la Harberger. Suppose a quarter of the labor of the country was misallocated. And suppose the misallocation was bad enough to leave, say, a 50 percent wage gap between the old sector and the new. This would be a large misallocation, indicating a large-scale irrationality of laborers in not moving to better jobs—or, more likely, a large-scale blockage laid down by bosses or a government controlled by bosses. The wage gap created by South African apartheid were even greater than 50 percent, but it seems unlikely that British wage gaps were so large as can be created by a sophisticated and powerful modern state intent on discrimination. Now imagine the labor moves to its proper industry, closing the gap. As the gap in wages closes the gain shrinks, finally to zero. So the gain from closing it is so to speak a triangle (called in economics in fact a Harberger Triangle), whose area is half the rectangle of the wage gap multiplied by the amount of labor involved. So again: ( ½ ) ( ¼ ) (50 percent) = 6.25 percent of labor’s share of national income, which might be half, leaving a 3 percent gain to the whole. The gain, as usual, is worth having. But it is not itself the stuff of revolutions. The division of labor: No.

The economic historian Jeffrey Williamson would in some ways disagree. In 1990 he argued that in the early nineteenth century in Britain “imperfect capital markets starved industry for funds, driving a wedge between rates of return in industry and agriculture. Since the industrial capital stock was, therefore, too small, industrial jobs were fewer than they would have been had capital markets been perfect.”¹⁵ That is, he claims there was an economically relevant gap between high returns to capital and labor in cotton mills and coal mines against low returns in agriculture. He uses a four-sector general equilibrium model of the sort he has pioneered in economic history and economic development to argue that factually speaking the capital-market gap and the labor-market gap amounted by 1850 (say) to a 7 percent lower GDP than would have obtained in a perfected world.¹⁶ Perhaps. One can quarrel with details of his model. And seven percent is not the stuff of revolutions. Further, Williamson himself—who is always generously comprehensive in his historiography—notes that many people (such as Crouzet and, with much less authority, since unlike Crouzet she has not done primary research into the matter, McCloskey) do not believe the imperfections existed in the first place. Long ago the economist George Stigler wrote a devastating essay against the conversation-ending rhetoric of “imperfections in the capital market.”¹⁷ An historian ignores Stigler to his peril.¹⁸

And Williamson makes the crucial point against his own argument: “the view that wage and rate of return gaps represent disequilibrium and factor market disequilibrium may also be challenged.”¹⁹ Yes, it may. The question is whether an observed gap is “economically relevant.” A higher rate of return to the owner of a wool mill as against a sheep farm may come from a greater degree of risk in making cloth than in raising wool. A higher wage in the industrializing North than in the agricultural South of England may come from costs of moving, including cultural tastes, and the disamenities of smoky Halifax—a point that Williamson himself demonstrates is factually relevant to the period. He writes, “some portion of the higher earnings of urban residents may be simply compensation for the disamenities of urban life and work.”²⁰ If so, the gaps represent reasonable adjustments to available opportunities, not sluggish stupidity. A southern agricultural laborer ordered peremptorily to go north to Halifax would incur costs of travel, retraining, home-sickness, nastiness (in his southern mind) of northern life, tearing of social bonds that would overweigh the future returns from a higher money income. If at liberty, and nobody’s fool, he would disobey the order. The capital and labor markets would then be, the economists would say, “in equilibrium,” despite the observed wage gap. Free lunches from reallocation would not be sitting around un-eaten, because they would not in fact be free of relevant cost.

Gaps between industrial and agricultural wages have persisted in every country in the world for decades, even centuries. For example, they persisted for the whole of the nineteenth century, as Williamson notes, in the fabled land of mobility and liberty and being nobody’s fool but your own, the United States. Such persistent gaps on the order of 50 or 100 percent, most economists would suspect, cannot be viewed simply as stupidly ignored free lunches. A mill owner could think of twenty different ways to pick up the lunch if it were merely a matter of stupidity. And the laborers have every incentive to pick it up themselves. Yet some economists have felt comfortable calculating the gain from reallocating labor across the wage gap, decade by decade, as though it were a free lunch sitting on the kitchen counter for a hundred years, to be slowly, persistently dined on. In honor of a great economist scientist who made the error fashionable, the calculation might be called the Kuznets Fallacy. The Fallacy is to believe without historical inquiry that every price divergence represents an opportunity for arbitrage, buying low to sell high without costs of transaction. It doesn’t seem so.

---

1. [back] Gerschenkron 1957 (1962).
2. [back] For the pro transport side in Britain, against my argument, see Szostak 1991 and 2003.
3. [back] David 1969. Fogel's reply was his presidential address to the Economic History Association meeting that year in Toronto.
4. [back] Fogel 1979.
5. [back] The point was made as long ago as 1970 by Roger Ransom.
6. [back] Krugman 1997, p. 52. The jibe apparently registered, since Krugman mentions it again in Krugman 2000, p. 55. Compare Luciani 2004, p. 4: "To say the clustering [of an industry in a city] is the result of localized external economies is too vague. It is a bit like saying agglomeration takes place because of agglomeration effects."
7. [back] Sellar and Yeatman 1931, p. 94.
8. [back] McCloskey 1972a and works cited there.
9. [back] Federico 2005, p. 151.
10. [back] McCloskey 1972a using the robust method of rent increases; compare McCloskey 1983; confirmed by Allen 1992, though using Arthur Young's dubious surveys, and processing them with dubious statistical methods (misusing statistical "significance," for example; see McCloskey 1995a).
11. [back] McCloskey 1975a; Wordie 1983. Re-confirmed by later studies by Allen 1992.
12. [back] Kussmaul 1981.
13. [back] For example in Hudson 1989.
14. [back] McCloskey 1981, 1994; reprised in Harley 1993, p. 200.
15. [back] Williamson 1990, p. 203.
16. [back] Williamson 1990, p. 207.
17. [back] Stigler 1967.
18. [back] Williamson does not in fact ignore Stigler. In Williamson 1975, p. 317n16 he argues, just as he does here, that Stigler ignores dynamic effects.
19. [back] Williamson 1990, p. 212.
20. [back] Williamson 1990, p. 232.

You may reply to this post without registering. Comments may be edited for content and clarity.

Name (required)

Mail (will not be published) (required)

Website