My critique of Nick Szabo’s “horse theory” of the Great Divergence between Western Europe and East Asia. This part is about China in the 18th century. See Part 2 for the general issue of industrialisation and transport costs in England.
Week before last in the comments section at Marginal Revolution, I exchanged words with Nick Szabo about his pet theory about the rise of the West. Then I noticed he posted his last rejoinder at his blog, “Transportation, divergence, and the industrial revolution“. The following is my summary of his views.
On the eve of the Industrial Revolution, the North Sea Basin, especially England, had many more draught animals per capita than East Asia. In particular, the English dray horse powerfully complemented all other forms of transport, acting as a “force multiplier” for movement of goods to market by canal, river, and sea. The combination enabled the transport of goods produced from many more acres of land than other livestock, let alone human portage. Szabo claims 20 : 5 : 1 was the megabeast / oxen / human ratio in “acreage access”.
 |
 |
Industrialisation required bringing together bulk goods such as coal and iron ore from disparate locations, which were prohibitively expensive to move by land. Thus, the “potential for a region to be the first to industrialize, was very sensitive to small changes in land transportation costs”. Szabo then invokes Metcalfe’s Law to assert that even small reductions in the cost of transport in England during the 18th century were multiplied by non-linear-dynamic-exponential processes into… I’m not sure exactly what. The last time I heard this, Jeff Goldblum was talking about butterfly wings causing hurricanes in the movie Jurassic Park.
By contrast, in China, land haulage depended completely or overwhelmingly on human porters pulling carts or wagons. (Szabo seems to rely primarily on vintage photos and films of rickshaws on the Bund.) Therefore, industrialisation was not possible until the Chinese were freed by technology from dependence on human muscle draught.
 |
 |
Szabo is not saying differences in transport infrastructure — like bad roads or a poorly maintained canal network — made the difference. He’s saying, horses, acting as complements to the networks, made the difference. Even if you were shipping coal 1000km down river or down the coast, you still had land haulage costs from colliery to port and from port to final destination. The land portion may have represented a small fraction of the distance, but those costs exacted an “exponential” impact. (Check out the tremendous emphasis he puts on the horse as a driver of the European-Asian divergence in posts such as this, this, this, and this.)
I think it’s a pretty silly “theory”, but it’s a good excuse to talk about other topics.
First, I can’t find any evidence that in the 18th century that the Chinese and the Japanese were totally or even overwhelmingly dependent on human portage for land 
transport. No doubt they were much more so than in Europe or the Middle East. But the only actual evidence of animal counts that I could find was a large farm survey conducted by John Lossing Buck, agricultural economist and husband of Pearl S., who reported that in 1930 the number of draught animals per acre in China was approximately half of England’s. But that doesn’t tell us anything about haulage animals during the Qing. (If anyone has any information about it please pass it on.) Nonetheless, carts and wagons drawn by oxen, mules, donkeys and (in the north) horses and camels must have played a part — especially if we are going by vintage photos :
  |
  |
My response to Szabo will make the following points :
- There’s no evidence, contrary to Szabo’s claim, that the early industrial revolution in England was “dynamically” sensitive to small changes in land transport costs. Industry location was the primary means of economising on them. (See part 2 on England.)
- Without arbitrarily large magical effects, transport is just another factor of production. If stagnation in transport productivity slowed “Smithian” growth (efficiency gains from more internal trade and more regional specialisation), then marginal cost savings via production technology could have compensated. You can still industrialise with higher transaction costs ; you would just be a little poorer than if you had lower transaction costs.
- Szabo’s argument is ultimately about energy — horsepower, kilowatts, joules, kilocalories, etc. But he only considers output and not the cost relative to output. The gap between English and Chinese transport costs, even on land, was not nearly as big as implied by Szabo’s supposed 20:1 land “acreage access” ratio.
- Besides, the English cost structure is not necessarily relevant to the Chinese cost structure. The Chinese would have used more of the input that was less expensive for them.
- Finally, Qing China’s transport system permitted a well-functioning and -integrated market. It could have supported a nascent industrialisation. At worst, stagnation in transport productivity might have set a ceiling on the scale of modern industrial production in the event that it got off the ground in the first place. But it never did. So transport is a premature, superfluous explanation, anyway.
Then, like now, Chinese labour was cheap
Szabo’s Law of the Second Best says no such thing exists. Thanks to the legerdemain of “nonlinear effects”, all imperfect substitutes for the English megaequid energised on hay and oats would have been fatally inferior. As far as he’s concerned, anything else — pasture-fed horses, oxen, Bactrian camels, mules, donkeys, reindeer, or llamas — would have nipped industrialisation in the bud, no matter how many might have been used to generate 1 horsepower.
But of course there must be a degree of substitutability. Even Langdon, who argued the transition from oxen to horses caused a revolution in mediaeval England, cited evidence that oxen moved more slowly but could pull heavier loads. From Smil here are estimates of the energy supplied by various draught animals :
Insofar as weight is an indication of maintenance costs, mules look pretty good in power/speed/cost terms. (The marginal cost of mules, i.e., the ratio of daily feed-wage to the marginal product of labour in power terms, is pretty low !)
A Chinese worker in the 18th century was much cheaper than an English horse. That must have closed some, if not all, of the horse’s power advantage. I calculate, the unit labour cost (wage relative to productivity) of horse haulage in England, compared with the in extremis case of human-only portage in China, would have been 2:1 in silver-money terms and 3:1 in PPP terms. (See the first post in the comments section for my calculation.) This implies that for any given acre’s worth of goods made accessible by human portage, horse haulage was 2-3 times as cheap/productive. No, this does not take into consideration that using horses probably saved on the number of wagons/carts. And presumably you could not produce 5 hp with 50 men because, as a speculative example, the canal walkways weren’t wide enough. (As with most input substitutes, the isoquant curve for horses/substitutes is convex to the origin, and there is a diminishing marginal rate of technical substitution — but not inordinately.) But the point is, the 20:1 ratio is in no way plausible, especially since the Chinese also must have driven oxen, mules and donkeys.
What about actual costs ? I did a handful of quick calculations for absolute costs of land transport. (Again, see first post in the comments section for the details.) They don’t really mean anything, because these costs should be examined in relation to the total value of the goods transported, but they still tell you how cheap Chinese labour must have been :
- Beijing Road, 1723-35, pole-carrier : 2.3 g silver per ton-mile
- Beijing Road, 1460s : 1.4 g Ag per tm
- Northern Zhili province, 18th c. : 1.52 g Ag per tm
- England 1700-50 : 6.8g Ag per tm
In silver terms, in the first half of the 18th century, the daily wage for skilled Chinese labourers was 2-4 g/day, depending on region ; and, for English labourers, ~11 g in London and ~6-7g outside. (Allen et al.) So at best there were modest differences in cost. Land transport was expensive everywhere before the age of rail and autos.
Szabo has argued that Malthusian pressures did not allow the Chinese to keep many draught animals because land was cultivated as intensively as possible for human consumption, which did not leave much room for nutrient-dense fodder. I guess he didn’t figure that extremely high land productivity in China, which resulted simply in more population, would have made made labour cheap. This is just a roundabout way of saying, capital was relatively inexpensive in England, labour relatively dear. The reverse was true in China.
As I said earlier, the country’s relative cost structure should matter more. Shiue & Keller estimates that the costs of sea versus inland waterway versus overland transport in the mid-to-late 18th century were 1 : 2.7 : 9.5 for China and 1 : 2.4 : 7.1 for England. If the cost of land transport relative to water transport was greater in China than in Britain, then the Chinese would have relied much more on water than the British. Obviously. Which is why internal Chinese trade was concentrated in a chunk of land serviced by rivers and canals, which was about 10 times the size of Britain.
There are also estimates of cost per value of goods transported for river & marine transport. One estimate From Shiue & Keller :
Given all of the above, the most reasonable conclusion is : if the costs of land freight relative to other freight were higher in China than in Britan in the 18th century, then the cause was probably not the source of draught power, but, more likely, bad roads. That’s plentifully attested in the historical record.
In England, improvements to roads and canals during the 18th century largely took place at private-sector initiative.
Qing China had a well-functioning market across large distances
In the 18th century, Qing China possessed a transport capacity to bring a vast amount of goods to market across long distances. At Marginal Revolution, I cited in support a passage from Kenneth Pomeranz’s The Great Divergence, a book whose overall thesis I reject. I repeat it for the benefit of readers here :
“The huge preponderance of land transport in preindustrial Europe probably results in part from the availability of so many farm animals, who had to be fed everyday but were only needed part-time for farming. Did Europe then have a crucial advantage in capital equipment for land transportation? Perhaps so, compared to east Asia, where pasture land was so scarce, but the remarkable development of water transport in China and Japan surely offset this and represented an at least equally valuable form of capital in transport; east Asia’s overall advantage in transport was noted at the time by Adam Smith.[47] And in parts of Asia where, as in Europe, there was lots of meadow and grassland, rural transport was probably just as highly developed… And both China and India had long purchased warhorses and some other livestock from central Asia, which had enormous amounts of pasture. After 1700, the Qing dynasty controlled much of this territory and bred its own warhorses. Had the Chinese needed to import other animals, this would have been ecologically feasible, too.[50]
“Nor do we see other signs of a shortage of transport capital in Asia. Such a shortage would presumably inhibit marketing, particularly of bulky goods such as grain. Yet in one of the most crowded societies of all—China—the share of the harvest that was marketed over long distances seems to have been considerably higher than that in Europe. Wu Chengming has conservatively estimated that 30,000,000 shi of grain entered long-distance trade in the eighteenth century, [51] or enough to feed about 14,000,000 people. [52] This would be more than five times a generous estimate of Europe’s long-distance grain trade at its pre-1800 peak [53] and over twenty times the size of the Baltic grain trade in a normal year during its heyday. [54]
“Furthermore, Wu’s figure includes only the largest of many grain-trading routes in China and uses cautious estimates even for those. He omits, for instance, Shandong province, which had a population of about 23,000,000 in 1800 [55]—slightly larger than that of France—and was neither particularly commercialized nor particularly backward. It imported enough grain in an average eighteenth-century year to feed 700,000—1,000,000 people—more than the Baltic trade fed—and exported roughly the same amount. [56] Thus, if we treat the grain entering and exiting this nation-sized piece of China as the equivalent of “international trade” in Europe, we find that this one province engaged in a grain trade comparable to all of Europe’s long-distance grain trading; and there must have been quite a bit of grain trading within the province as well, since even this volume of imports could not have met the demand from its urban areas (not to mention its cotton and tobacco growers).
“Nor was China unique. Many cities in various parts of Asia (and probably one or two in precolonial America) were larger than any European city before eighteenth-century London, and several were larger than London as well. It has been estimated that 22 percent of Japan’s eighteenth-century population lived in cities, versus 10—15 percent for western Europe;[57] and the Malay archipelago, though sparsely populated overall, may have been 15 percent urban. [58] Many of these cities—as well as some in south Asia and the Middle East—were heavily dependent on long-distance shipments of bulky foods.”
Pomeranz’s qualitative assessments are supported by some careful studies. Did China in the 18th century have a well-functioning market ? The best way to judge is to see how similar the prices of a bulk commodity were across regions. Shiue & Keller, “Markets in China and Europe on the Eve of the Industrial Revolution“, examines grain prices in Chinese and European cities across a comparable geographical space.
The above shows the degree of convergence in grain (rice and wheat) prices between cities as a function of distance. The more negative the values on the vertical axis, the less difference there was between cities in terms of their grain prices. (I’m not using the word “correlated” because that is the incorrect term.) The inter-city difference in prices, if they were kept roughly constant by price arbitrage, should reflect the transaction costs of inter-city trade. So you can interpret the gap between the curves as an upper bound on the difference in transport costs between European cities, Yangzi river markets and Chinese provincial capitals.
Chinese market integration compared quite favourably with Northwest Europe as a whole in the 18th century. It’s only at the shortest distance that European integration was stronger, which is consistent with the claim that for the Chinese land haulage was more expensive than other kinds of transport. So the incentive in China was to move goods farther away than in Europe, because of the differences in relative costs of land and water transport between the two.
This study lacked 19th century data for China, but by 1825-49 Northwest Europe moved ahead of where the most advanced region of China had been in 1742-1795. England had also been ahead of everyone by the late 18th century. I should add, a more recent study argues that China’s market integration deteriorated between 1740 and 1820. But it explicitly rules out transport costs as an explanation because the deterioration in markets correlated strongly with provincial borders but not with river networks or postal routes.
Also, Britain, in size and population, was comparable to a middling Chinese province. If a couple of adjacent prefectures were specialised and mutually dependent on trade, then that would be the equivalent of integrating London, Manchester and Birmingham.
The exception that proves the rule : the British East India Company to the rescue of Chinese supply bottlenecks
Some time during the early Qing (1644-1911), farmers in Fujian and Guandong began a more intensive cultivation of rice as well as cash crops like sugar and tea. Cotton was being crowded out :
Marks also has material on the expansion of commerce based on coastal shipping.
So even if China’s own transport capacity had been too limited or too slow to expand, foreign traders and shippers in Canton were present to fill the gap — in spite of the infamous Qing restrictions on foreign trade. Just as now, this could have supported a coastal industrialisation. So transportation cannot possibly have been a check upon it.
Qing China’s transport capacity could have supported a nascent industrialisation. It’s true, all else equal, between any two countries, the one with the lower transaction costs (including transportation) should have more internal trade and regional specialisation, and therefore more gains from trade. But modern economic history suggests, such gains face diminishing returns, and efficiency gains from rapid and sustained technological innovation are much much bigger. That’s true even if there are feedback effects between commercial and technical expansions.
In the 18th century England experienced both kinds of expansion — “Smithian” and “technological”, even if the effect of the latter was much bigger. Who says you must have the same balance between the two as England had ? Almost nobody else did.
Qing China lacked the appropriate technological innovation
In 1800, continental European countries were well behind England but many of them still had had at least isolated episodes of industrialisation. That wasn’t the case with Qing China in 1700-1800. There was cotton textile production in the Yangzi delta (with the raw cotton transported from the north). But it was largely a domestic cottage industry of peasant wives spinning at home as a side activity to supplement their families’ low farm incomes. In this labour-intensive sector, there was no equivalent of modest, early Lancashire inventions as the spinning jenny or the water frame.
According to Joseph Needham, the Chinese “treadle-operated three-spindle spinning-wheel” had been the global technological standard for spinning cotton yarn until James Hargreaves invented the spinning-jenny in the 1760s. But the three-spindle wheel had been invented in the 14th century during the Ming dynasty. Soon thereafter, it was “upgraded” into a 32-spindle water-powered spinning-frame for ramie and hemp. :
But the Chinese did not or could not adapt that for cotton, when it became the dominant and preferred textile. Many people now argue the incentive for mechanisation was absent in China because labour was too cheap. (I reject the ironic “cheap labour” theory of Chinese non-industrialisation, but that must be the subject of another blogpost.)
I also contend, just as English industrial concentration shifted from southeast in the pre-mechanisation era to the Midlands and the north in the early mechanisation era, so Chinese industry might have moved closer to the sources of expensive inputs like coal, horses, pasture, and land suitable for fodder crops, in the provinces to the north of Beijing. Manchuria was still a vast frontier ready for exploitation, and about the size of Pakistan or Turkey. The problem in this case was probably political. The Qing dynasty was Manchurian and would not permit commercial expansion in the barely exploited Manchuria and Mongolia.
Which is why I say Szabo puts the cart before his hobbyhorse. There were much more basic failings in China than the lack of horses.
Also see : Szabo is wrong about England, too.
Filed under: China, Economic History, Great Divergence, Industrial Revolution, Rise of the West Tagged: Chinese economic history, great divergence, horses, Industrialization, Nick Szabo, transport
