McCloskey: Cotton wasn’t crucial to the Industrial Revolution

I (mostly) copy-and-paste Deirdre McCloskey’s argument that cotton was not crucial to the Industrial Revolution in Britain. I also have a very brief rant about historians’ erasure of Robert Fogel from historiographic memory.

In an earlier post I argued that American slave cotton may have accelerated the Industrial Revolution, but certainly was not essential to it because there were alternative sources of cotton. One of the speculations I aired was: if Lancashire mills had to rely exclusively on inferior-quality Indian cotton from the very beginning, then textile machinery would have been optimised for Indian lint. And now there’s a paper which finds in fact that did happen in the British textile industry during the “cotton famine” of the US civil war: Hanlon, “Necessity Is the Mother of Invention: Input Supplies and Directed Technical Change” (ungated version).

But an entirely different argument I might have made in the earlier post is that cotton itself was not crucial to Britain’s (or America’s) industrial takeoff. But such a post wouldn’t have allowed me to pontificate on the international economic impact of the American civil war — let alone post a neat vintage photo of Indians towing their oxen burdened with massive bales of cotton…

I wasn’t going to return to this tired subject, but I keep seeing irritating articles like this or this, or book presentations by Sven Beckert like this which insist on extravagant claims about the “centrality” of slavery and slave cotton to the rise of western industrial capitalism.

My previous post was inspired by a wild assertion by Edward Baptist in his crude book that American slave cotton was “absolutely necessary” if the “Western world was to burst out of the 10,000-year Malthusian cycle of agriculture”. More measured language and temperate claims appear in Beckert‘s non-emotional book, but nonetheless its central implication is that industrial capitalism just wouldn’t have been possible without the violent exploitation of all kinds of coerced labour outside of Europe, which he calls “war capitalism”.

There’s a quick way to short-circuit all that loose talk.

In chapter 25 of Bourgeois Dignity, Deirdre McCloskey has a brief argument that, contra Findlay and O’Rourke, international trade was not an important engine of growth for Britain in the period 1780-1860. Her counterfactual involves halving Britain’s actual cotton textile output as a result of being shut out of foreign markets. It’s a static argument (i.e., not general equilibrium), but McCloskey is convincing that it’s good enough for the job and any dynamic model would not generate very large magnitudes for the importance of cotton.

The logic can be transported, with a few adjustments, to the counterfactual of cotton made more expensive by the absence of any coercion in cultivation. If the primary raw material had been more expensive, presumably its impact would have been a reduction in the quantity of output and a reallocation of the unused inputs to other, presumably less productive sectors of the British economy.

Since the argument is about as clearly presented as feasible, I simply quote most of the relevant bits, excising only McCloskey’s extended parenthetical asides. Anything in bold is my emphasis.

Suppose the cotton textile industry were cut in half by an exclusion from foreign markets… During 1780-1860 therefore the share of cotton in national income would have been 3.5 percent instead of its actual 7 percent. The 3.5 percent of resources would have had to find other employment. Suppose that the released resources now put to use in road mending and silk manufacturing and so forth would have experienced a productivity change of 0.5 percent per year (on the low end of the available possibilities) instead of the princely 1.9 percent they in fact experienced in cotton.

The cotton industry in the actual, 1.9 percent world contributed a large amount-namely, (0.07) • (1.9 percent) = 0.133 percent per year-to the growth of national income. This one giant contributed some 24 percent of the conservatively measured total of about 0.55 percent-per-year growth of income per person nationally 1780-1860…


Now we can calculate the counterfact. With the hypothetical halfway cutoff of trade…you can make so to speak a mechanical “static-dynamic” argument as follows. The Harley revision of my table implies that noncotton productivity change can be calculated from (1.41-.07)• (the implied residual productivity change outside cotton) = (0.55-0.13). That is, the implied residual of productivity change outside cotton is 0.42/1.34, or 0.313 percent per year (I retain more than significant digits to avoid rounding errors). The resources in the hypothetical case would therefore contribute (0.035) – (1.9 percent) + (0.035) • (0.313 percent) = 0.077 percentage points a year. The fall in national productivity change can be inferred from the difference between the actual 0.133 percent per year attributable nationally to cotton and the hypothetical 0.077 percent per year attributable to a half-sized cotton industry and the industries its resources would go to. The difference is about a 0.056 percentage point per year fall in the national rate of productivity change, that is, a fall from 0.550 percent a year to 0.494 percent a year. In the eighty years 1780-1860 such a lag would cumulate at monthly interest, however, to merely 5 percent or so of national productivity change. Remember that we are speaking here of doublings of national income 1780 to 1860.

You could cut the productivity change in cotton to allow for alleged economies of scale in cotton and come to roughly the same result. No one has shown that such economies of scale were in fact important (though they are important in the imaginings of economists), or that economies or diseconomies of scale in other industries would not cancel the net gain. We are giving the protrade, “dynamic” argument all the advantages. Suppose the scale-effect productivity change were half of the princely 1.9 percent in cotton, or 0.945 percent per year. So now the calculation is (0.035) -(0.945 percent) + (.035) •(0.313percent), or 0.0440 percentage points a year (as against 0.077 without the lost “economies of scale” inserted). National productivity change attributable to cotton falls from 0.133 percent per year all the way down to 0.0440 per year, a drop of 0.089 each year. So national productivity growth would decline on this account in the hypothetical world from 0.55 actual to 0.461 percent per year. The difference in final attainment in 1860 is again small, merely 8 percent of productivity change, and a smaller percentage of national income.

Note that the result is forced by widespread character of productivity change (even under the implausible Crafts-Harley calculation of zero productivity change outside the industries I chose in 1981 as the leading ones). Resources displaced from cotton do not simply disappear. They relocate. Their relocation does not result in a fall of national income equal to what they earned in cotton, as assumed by noneconomists. The resources of labor and capital go into other industries, albeit ones with lower productivity change than the amazing cotton textiles. Since cotton was not the only industry experiencing productivity change even in the classic period of the early Industrial Revolution-a point that the economic historians Peter Temin and John Clapham and I insist upon, and historians of technology such as Margaret Jacob and Joel Mokyr have affirmed in detail-the imagined shift is not deadly to progress.’ The dynamic effect sounds promising. In quantitative terms, though, a cotton textiles industry counterfactually smaller (if foreign trade were shrunk) does not kill off growth. It’s another popular explanation that doesn’t work very well.

A “dynamic” argument, further, has a serious problem as an all-purpose intellectual strategy. If someone claims that foreign trade made possible, say, economies of scale in cotton textiles or shipping services, she owes it to her readers… to explain why the gains on the swings were not lost on the roundabouts. Why do not the industries made smaller by the large extension of British foreign trade end up on the negative side of the account? The domestic roads in Shropshire not constructed and the brass foundries unbuilt in Greater London because of Britain’s increasing specialization in Lancashire cotton textiles may themselves have had economies of scale, untapped.


The table above embodies what is known as the “Crafts Harley view” — the controversial idea that the first phase of the British industrial revolution was driven by rapid productivity growth in a few “star” industries and the rest of the economy was largely stagnant. Crafts-Harley is essentially a “remainder” argument. GDP growth was estimated for the period 1780-1860, from which economy-wide productivity growth is derived. Then the productivity growth estimates for the very few individual industries which have actually been micro-studied are subtracted from the aggregate. The remainder turns out to be low. Hence, the “star” industries dominated the IR.

There have been many counterarguments to Crafts-Harley, but for me the most important is that GDP calculation for this period has a large margin of error, so there could be a substantial discrepancy between the macro estimates of productivity growth and the aggregation of micro estimates from individual industries. And every time someone studies a new, previously unexamined industry, it turns out its productivity was advancing rapidly ! A whimsical example is candle-making (source: Allen, ch.10, CEHMB v1) :

B2z6MkJIgAILgGcI mention all this because IF the Crafts-Harley view is wrong, then that implies the Industrial Revolution even in its earliest phase brought about a broad, diversified, multi-sectoral advance in productivity. And the cotton textile industry, by itself, was even less crucial than McCloskey’s exercise would suggest.

Edit: 10 July 2016. I hadn’t seen this at the time of writing, but I should mention Clark, O’Rourke & Taylor, “The growing dependence of Britain on trade during the Industrial Revolution” [ungated version]. The paper estimates the effect of foreign trade on British welfare circa 1760 and 1850: “We find that while trade had only a small impact on British welfare in the 1760s, it had a very large impact in the 1850s”.

Postscript: In his review of a new book on the “business of slavery”, Eric Herschthal begins with this passage :

Perhaps the most durable myth about slavery is that it was utterly incompatible with capitalism. Well before historians in the twentieth century began legitimating the idea, abolitionists suggested the disconnect themselves. In Uncle Tom’s Cabin, Harriet Beecher Stowe portrayed slave-traders as underfinanced, disreputable fools—the furthest thing from competent, successful businessmen. Even Karl Marx, no friend of capitalism, believed that wage labor would destroy slavery. But in recent years, scholars have begun to demolish this myth, arguing not only that slavery was compatible with capitalism, but that the emergence of modern capitalism made slavery’s growth possible.

Herschthal, who specialises in the history of slavery, does not see fit to mention anywhere in the article the classic Time on the Cross by Robert Fogel and Stanley Engerman — even though it established 40 years ago, not only the compatibility of slavery with capitalism, but also the high productivity of slave agriculture and the flourishing of the market system in the slave economy of the US South. TOC even argued slavery would have worked just fine inside industrial factories ! Yet the “new historians of capitalism”, rather like the Argentine junta, have essentially disappeared Fogel and Engerman.

And then they claim to have invented the wheel. The narrative these “new historians of capitalism” (who are from US history departments) keep repeating is, before they started uncovering the truth, no one had noticed the capitalist nature of slavery.

Nobel laureate and chief desaparecido Robert Fogel is not mentioned even once in Sven Beckert’s Empire of Cotton (***) even though its subject is the centrality of cotton and labour coercion to the rise of industrial capitalism. Amazing !

[(***) Edit 8 Nov. 2016: This was true at the time of writing. My Kindle edition still has no mention of Fogel, but on Google Books the latest copy of the book shows that Time on the Cross has been appended to footnote 37 of chapter 5.]

It’s meme replication via memory erasure.

Ever seen the French-language film Danton by the Polish director Andrzej Wajda ? It’s about Georges Danton, the first president of the Committee of Public Safety who was later guillotined by Robespierre during the Terror. There’s a scene in the film with Robespierre visiting the studio of the painter Jacques-Louis David. In the background is the yet-unfinished famous painting, “Le Serment de Jeu de Paume” (The Tennis Court Oath). Robespierre orders David to erase the figure of Fabre d’Eglantine, who has just been purged. It’s a Polish director’s allusion to Stalin’s doctoring of photos to remove purged rivals.


This entry was posted in cotton, great divergence, historians of capitalism, Industrial Revolution, Slavery and tagged , , , , , , , , . Bookmark the permalink.

27 Responses to McCloskey: Cotton wasn’t crucial to the Industrial Revolution

  1. afinetheorem says:

    Spot on – huge evidence that cotton is not of singular importance in the IR, even when one takes financing into account.

    On slavery and capitalism, you may enjoy this new paper by my friend Guillermo Marshall, with Suresh Naidu and Felipe Gonzalez:


  2. Caroline Heindel, my high school American History teacher, insisted that we get beyond the survey book and encounter Beard, Stamp vs Fogel and Engermann, etc. I learned more than I did in my college survey course (at Oberlin) and indeed would have been better advised simply to deepen my exposure to the controversies and theories that she referenced two years earlier.

    I have the Nykert book and found it unreadable – perhaps because I’ve read enough history to already know about “War Capitalism”, etc. I kept asking – who is this book for? It seemed to be for people who know so little history that there is no chance they would be reading this book.

    But I am skeptical around this no-cotton IR counter-factual. It assumes there would have been markets for the products of all this redeployed capital. Would there have been? Especially equivalent export markets?


  3. McCloskey’s counterfactual was halving the cotton output, not no-cotton. Regarding export markets : no, you would have had less exporting, period. In the factual world, Britain specialised more and more in cotton production, and every marginal output of cotton went overseas, while it imported more and more food. So the effect of curtailing cotton output would have been a change in Britain’s pattern of trade and mix of production.


  4. The imports of food were related to increase in population, not just simply to urbanization and surely not to productive reallocation. Broadberry et al., “ENGLISH AGRICULTURAL OUTPUT AND LABOUR PRODUCTIVITY, 1250-1850: SOME PRELIMINARY ESTIMATES” (2008) show both acreage sown and yields increasing dramatically during this period. Hard to imagine it could have increased more. So some other export or drain of cash would have been needed to pay for this food. There wasn’t enough wool (pace Hudson) to do it, so what would it have been? (Sorry for my “macro” protest against a general equilibrium conclusion, but I tend to think top-down.)

    Note: I’m not arguing that “slave cotton” was necessary for the IR to happen. And I haven’t read McCloskey. But isn’t it odd to argue that the rising bourgeois didn’t “need” slavery, when they embraced it so eagerly? The willingness to exploit slavery seems part and parcel of the broader phenomenon McCloskey is pointing to. And if she is arguing that a certain set of attitudes made modern economic growth possible, don’t we have to take the bad with the good?

    *Yes, we know from your work that Broadberry’s series contain some perhaps-complacent assumptions, but this is the modern and hopefully more reliable end. Striking, though, that Mitchell has absolutely no pre-1815 agricultural numbers. Perhaps even the modern “preliminary estimates” are guesses!


  5. The ability to pay for food imports is proportional to the ability to export. So if higher raw cotton import prices reduced textile exports then the UK would have been obliged to import less food — or more precisely, a smaller volume of imports in money terms. Yes, agricultural productivity grew more slowly than industrial productivity, so the higher population did require more food imports. The combination of all these things implies that the UK would have been obliged to import fewer luxuries and spend the reduced export earnings on cheaper commodity foods.


    • And/or Britons would have reduced non-food consumption in general (i.e., increase savings) in order to export more of the foregone consumption, in order to afford the food imports.


  6. The willingness to exploit slavery seems part and parcel of the broader phenomenon McCloskey is pointing to. And if she is arguing that a certain set of attitudes made modern economic growth possible, don’t we have to take the bad with the good?

    Why should I care about that ? I like McCloskey the cliometrician. McCloskey the rhetorician, not as much. Bourgeois Dignity is useful as a chapter by chapter polemic against every pet theory of the IR. Its value is negative; when it gets positive with the bourgeois dignity stuff I commune with my inner positivist.


  7. Why should you care? Well, you might care about it because of the “Clio”, or “history”, half of cliometrics, which is a discipline that not only advocates economic modeling and performance measurement, but which emphasizes the impact of institutions and institutional changes on economy and, therefore, society. So the question is not just whether, from an economics perspective, American slavery was a necessary condition of British industrialization. The question is also whether, from a historical perspective, American slavery was a consequence, perhaps a necessary consequence, of the set of attitudes and inclinations that a newly-energized bourgeoisie applied to the spheres of manufacturing, trade, and colonization. Why ask whether these mojo-ed economic actors could have done without slavery? If they perceived it as expedient, it was going to happen. With capitalism, absent legal prohibition, is there any counter-factual to perceived advantage?


    • You mistake my meaning. I reject McCloskey’s view that the “great enrichment” was due to some ethical revolution. So why should I care if her “ethical revolution” is (allegedly) incongruous with the fact of industrial-scale slavery ?


    • The question is also whether, from a historical perspective, American slavery was a consequence, perhaps a necessary consequence, of the set of attitudes and inclinations that a newly-energized bourgeoisie applied to the spheres of manufacturing, trade, and colonization.

      Why pin it on some new-fangled “attitudes and inclinations” when slavery had always existed, and the scale and character of its exploitation quite obviously depended on the production regime ? 19th century cotton agriculture in the US was essentially industrial and “proto-Taylorist”. That hadn’t been quite the case in the 18th century with tobacco, but had been with sugar in the Caribbean. Romans had latifundias, but Arabs generally didn’t.

      I’m not saying attitudes & values don’t come into play. The transition from slavery to serfdom in Europe is not quite understood.


  8. No claim here that slavery was a bourgeois invention. Only that a cliometric approach shouldn’t be too selective in noting the effects of its chosen causes. Meanwhile, I’m going to reread “Time On The Cross”. I might just have it in some box downstairs.


    • Only that a cliometric approach shouldn’t be too selective in noting the effects of its chosen causes.

      I have no idea what that means. The McCloskey passage above is intended as a preliminary accounting of the impact of the cotton textile sector on productivity growth in the 1st phase of the IR. It is not asking “what caused the Industrial Revolution” (including the high productivity growth in textiles).

      Clark had this, though the method was cruder and even more static than McCloskey’s :

      I’m going to reread “Time On The Cross”. I might just have it in some box downstairs.

      Without Consent or Contract is more up to date and has superseded TOC, which I mention primarily for historiographic purposes.


  9. Yes, but it was TOC that was accused of the fault around which you have “no idea”. I don’t mean the no-fault-at-all of being insufficiently moralistic. I mean the fault of emphasizing economic over social impact to the point of seeming to wear blinders. I think that was probably a bad rap on TOC, which is why I want to go back to it.

    North: “…by economic performance I include not only the standard quantitative measures: changes in the size and composition of population, aggregate income, income per capita, and the distribution of income; but also other less measurable but important characteristics of well-being such as the quality of the physical and social environment.”

    North also counted taboos among “institutions”. So perhaps my counter-factual, as opposed to the simple “unavailability” of export markets for cotton, is “what if, in addition to the institutions that McCloskey does study, not least the removal of the stigma around commerce, there had been a taboo around slavery?”


    • The more you elaborate, the less I understand.

      The narrow quantitative focus of the counterfactual is compelled, induced, motivated, prompted, and dictated by the nature of the initial claim — American slave cotton was indispensable to the IR. So the natural reaction is to investigate the quantitative importance of cotton textiles to the IR.

      Other things — the willingness to exploit slavery, the “attitudes that made slavery possible”, the social impact of slavery, the institutions that supported slavery, taboos — are all interesting and fit subjects in themselves. They are just not germane to the issue immediately at hand. (I think cultural evolution is an important element of the change in attitude toward slavery.)

      Besides, those topics are well covered by qualitative histories. There are always lots of qualia in clio. The lacunae are usually in the metrics.


  10. Actually you do understand. I was just saying that the “initial claim” was, for me, the wrong question. Since McCloskey’s subject was the impact of changing attitudes towards commerce upon economic growth, I was asking whether it wasn’t more interesting to “change the mix” in that attitude set. Given her hypothesis (and I only know it crudely from the publisher’s blurb) I find it more interesting to ask whether a bourgeois-driven economic renaissance would have been as “successful in general”* without slavery as with (and not just for the BIR). Your question is without cotton (and implicitly American slavery) or with (and just for the BIR). I grant that your question is more precise and more answerable. But it seems that slavery was something of a growth industry over the whole early-modern, European colonization period. What, if for the same period, it had been “against the rules”? Instead of accepting the cotton-for-slavery proxy, my counter-factual is slavery itself.

    What “Time on the Cross” suggested, was that slavery was a form of economic rationality. Whether or not it was necessary for growth in an absolute sense, it worked from the standpoint of certain rational economic actors and economic outcomes. Given the huge rise in economic rationality, in its capitalist form, in the early modern period, slavery, absent some taboo, was bound to increase as well. (That’s just a top-of-my-head hypothesis, as was my earlier growth-of-slavery statement. I have no numbers at hand.)

    Now, “no slavery” does not necessarily mean better, even if equivalent economic growth were achieved without it. It seems that Fogel and Engermann were all too clear-eyed about this, and hence the uproar. That nineteenth century slaves might have been better off, by some measures, than nineteenth century factory workers or a twentieth century black underclass was not a comfortable question. I have no problem with it at all. It does not trivialize slavery or make slavery acceptable to say that, in some places and times and in some ways, there are other things in the world that are as bad or worse.

    meaning, not using just the most common measures of economic performance, GDP and per capita GDP, but looking at growth, distribution, and social consequences. That was the reason for the North quote in the earlier post.


  11. In the previous post, the last section was supposed to be a footnote to the asterisk in the first paragraph.


  12. While we’re on the subject of Robert Fogel, find in the link below a Harvard-MIT duo’s paper that revisits his famous “no railroads” counter-factual. The reasoning in this one is way over my head, but the conclusion is not:

    “Removing all railroads in 1890 is estimated to decrease the total value of US agricultural land by 60%, with limited potential for mitigating these losses through feasible extensions to the canal network or improvements to country roads. Further, in the absence of the railroads, population
    levels and consumer welfare are predicted to decline substantially.”

    Donaldson, Dave, and Richard Hornbeck. 2015. “Railroads and American Economic Growth: A “Market Access” Approach”.


    • Yes, I know that paper. But in some ways it’s a vindication of Fogel. Instead of Fogel’s finding that the social savings of the railroads in the agricultural sector were 2.7% of GDP, D & H find 3.22%. Not a trivial difference, but also not an earth-shattering one….

      I am going to do a post one day on Gutman’s The Numbers Game.


  13. An earlier (2012) version of the article had this, to-be-revised-later claim:

    “Removing all railroads in 1890 would decrease the total value of US agricultural land by 73% and GNP by 6.3%, more than double social saving estimates (Fogel 1964). Fogel’s proposed Midwestern canals would mitigate only 8% of losses from removing railroads.”

    Even in the final paper there quite dramatic statements about impacts that were outside the social savings computation, such as the loss of value of agricultural land. It’s the spread of effects that interests me, perhaps partly because I’m too new to the literature to get my mind around conclusions based upon a single unfamiliar measure, such as social savings, or even TFP.

    As I reread your post yet-again, I also realize that my point argued above is really no different than than Eric Herschthal’s in the review quoted. I was just taking (a hypothetical,version of) McCloskey to task for leaving slavery out of the balance sheet, which maybe she doesn’t even do.

    Enough of this one, except to note that one pair of writers who do not fail to cite Fogel and Engermann are our favorite Guldi and Armitage, p. 97f. They graciously note that, having been “banished for its sins”, cliometrics may recover in the new big data ere that they herald. Isn’t that nice?


    • I recommend this as an introduction to the history of the social savings concept, its limitations, comparison with alternate methods, etc.


      • Thank you for the reference.

        Meanwhile, yesterday, one of those fortunate accidents that can happen with new retail omni-channel fulfillment systems: the “Evidence and Methods” supplement to “Time on the Cross” arrived first. So I got to begin by reading Appendix A, which is exactly the sort of dissection of the differences between “history” as commonly written and “social science history” and worthy successor to Braudel’s ‘combat’ that “The History Manifesto” isn’t. The defense of the difficult and even-then-becoming-unfashionable ideal of dispassionate and unbiased analysis is particularly good – and a wonderful antidote to much that has been written since.

        But I also figured out that I can’t possibly have read this in high school. I graduated in 1967 and the famous explosion at the Economic History Conference, revisited in Appendix A, occurred in September of that year. So the dimly-recalled controversy to which I was introduced the previous year must have been Stampp vs. Elkins (the other Stanley). It’s possible, of course. My teacher (and her Ph.D. candidate husband) may well, even probably, have known of Conrad and Meyer’s essay, and she might have mentioned it in class to introduce the profitability question, but I don’t remember those names.


  14. Pingback: Baptism by Blood Cotton | Pseudoerasmus

  15. Pingback: Piketty & Slave Wealth | Pseudoerasmus

  16. Pingback: Was slavery necessary to western industrialisation ? | Pseudoerasmus

  17. Pingback: Plant breeding, not working slaves harder, drove cotton productivity gains in the US South | Pseudoerasmus

  18. Pingback: The Baptist Question Redux: Emancipation & Cotton Productivity | Pseudoerasmus

  19. “Her counterfactual involves halving Britain’s actual cotton textile output as a result of being shut out of foreign markets. It’s a static argument “ahhh…the counterfactual, you’re talking about fiction, right?


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s