"Why Isn't the Whole World Developed? Lessons from the Cotton Mills", Clark 1987; excerpts:
"In 1910 one New England cotton textile operative performed as much work as 1.5 British, 2.3 German, and nearly 6 Greek, Japanese, Indian, or Chinese workers. Input substitution, and differences in technology, management, and workers' training or inherent abilities do not explain this. Instead local culture seems to have determined worker performance. Such differences, if widespread, would explain much of the international variation in wages. They also have important consequences for understanding labor migration, the choice of technique, and the sources of economic growth.
Perhaps the most prevalent modern view is that poor countries have remained poor because they cannot absorb the technologies of the advanced industrialized countries. To explain this, some cite the lack of education in poor countries; others, the lack of management skills or an entrepreneurial tradition, the lack of appropriate institutions, or the economies of scale inherent in the new industrial technologies which hinder their small-scale or isolated adoption.2 In this article I examine underdevelopment through a detailed study of one industry, cotton textiles, in the early twentieth century. For most of the countries I studied, factory production of textiles was a first step on the path to industrialization. I conclude that the failure of poor countries to dominate textile production in the way their labor cost advantage would suggest proceeds overwhelmingly from inefficient labor rather than from failure to import technology or management skills, or from failure of local capital or input markets, or from scale economies within the textile factories or at the industry level. In cotton textiles all these problems proved minor
When Britain launched the modern factory system in textile production in the late eighteenth century, people had every reason to expect that the Industrial Revolution would soon be a worldwide phenomenon. The new textile technologies were not particularly complex, most textile factory jobs did not require great skill, and every country had a ready local market for yarn and cloth. Despite attempts by the British to prevent the export of industrial technology-the emigration of artisans was prohibited until 1825, and most machinery exports were forbidden until 1842-the new machinery and methods spread quickly to North America and to Europe. By 1825, despite the prohibitions, an estimated two thousand British artisans had emigrated to the Continent.3 Thus while Britain established the first modern spinning factory in 1771, the French had established one by 1778, one was built in the Ruhr by 1784, the United States and Prussia had spinning mills by 1791, Russia had its first spinning mill by 1793, Switzerland by 1794, the Netherlands by 1795, and Belgium and Saxony by 1799.4 Even India, 12,000 miles by sea from Britain and very underdeveloped, joined the group in 1817, when a British trading firm shipped out to Calcutta a modern mill powered by steam.5 Nicolas Koechlin, a prominent Alsatian cotton manufacturer, reported in 1834 that when several leading Alsatian cotton manufacturers visited English factories in 1833 they "saw nothing particularly interesting" and "except in the high numbers, Alsace was not a whit behind hand."'6 Moreover, the Continental manufacturers apparently had great advantages in terms of the cost of labor in the 1830s. Hourly wages compared with Manchester were about 37% in Mulhausen in the Alsace, 28% in Zurich, 25% in some mills in the Vosges, 38% in Ghent in Belgium, 47% in Rouen, about 36% in Austria, and less than 25% in Prussia.7 Continental manufacturers did face higher costs for capital, machinery, repairs, and power (if they were using coal). But in 1834 wages were about two-thirds of the conversion costs (the cost of turning raw cotton into yarn or cloth) in spinning in Manchester, so that the costs of capital, machinery, repairs, and power would have to be at least 128% higher in Alsace to equalize costs, and 150% higher in Prussia.8 The cost differences reported by Koechlin are only of the order of 70% for Alsace.9 As these textile industries developed, there was every expectation that such cost differentials would decline substantially, when the size of the local machinery and the repair and capital markets increased, and so too would decline British dominance of export markets.
Yet despite the great labor cost advantages of the Continental countries, and the even greater labor cost advantages of the Asiatic countries, few of them were able to compete with the British outside their protected home markets before World War I. In 1911, 140 years after the establishment of the first spinning mill, 40% of the factory cotton spindles in the world were still in Britain; of those outside Britain 22% were in the United States or Canada, the only countries with higher wage costs than Britain. Germany, France, Russia, Belgium, Switzerland, Italy, Spain, Portugal, Austria, India, Japan, China, Mexico, and all the other low-wage competitors had only 39% of the world stock of spindles.'0 More importantly in 1911 Britain still accounted for 81% of the net exports of cotton yarn and cloth in international trade.1 Britain faced little competition in unprotected markets, since except for the Asiatic countries with their extraordinarily low wages, manufacturing costs in all the low-wage countries were higher than in Britain, making tariff protection necessary if the British were not to overwhelm even their domestic markets.
...given the labor cost advantages of Britain's competitors, what really requires explanation is not the ultimate demise of the British industry but why British textiles continued to thrive for so long. Second, the specific failures attributed to British managers prior to 1914 had very little impact on total costs and should have been completely swamped by labor cost effects. For example, one of the major mistakes attributed to British managers was a failure to switch from mule spinning to ring spinning.13 William Lazonick has argued that this failure stemmed from the more fundamental error of retaining the horizontal specialization of the British industry into spinning and weaving factories. With separate spinning and weaving mills, yarn has to be shipped from the spinning mills to the weaving sheds. It cost more to ship the ring weft (filling) yarn than to ship mule weft yarn, which he argued made it less costly to retain mules for weft spinning.'4 When this transport cost is calculated, however, the most it could have cost the British for sticking with mules for this reason could be only 1.4% of total value added in cotton textile manufacturing.'5 In comparison, British wages per hour were about 6.5 times those of the Indian industry, and since labor costs were about two-thirds of the value added in the industry, the British suffered a 57% cost disadvantage on account of wages. The correct question is not why bad choice of technique doomed the British industry, but why an industry apparently so burdened by high costs for its major input could remain so long successful, competing against a host of low-wage competitors.
To illustrate the magnitude of Britain's competitive disadvantage, consider the costs of inputs other than labor in cotton textile production. Cotton was the major input, accounting for about two-thirds of the total cost of cotton cloth, but its cost to Britain's competitors was generally the same or less. Britain had to import all its cotton, and in 1910 it actually cost slightly more to ship cotton from New Orleans to Liverpool than to ship it to Fall River in New England, or to Bremen, Le Havre, Antwerp, or Genoa on the Continent.
Table 1 shows the cost of labor per 55-hour week, the cost of coal per ton, and the cost of a complete spinning and weaving mill (including erection costs) in 15 countries or regions, the data gathered principally from the Special Agents Series of the U.S. Bureau of Foreign and Domestic Commerce. These reports, together with the U.S. Tariff Board report on Cotton Manufactures and other sources, provide a detailed summary of conditions in the international industry around 1910. As the table shows, labor costs varied by extraordinary amounts, Chinese labor costing 10.8% of that in Britain, and a mere 6.1% of that in the United States. It is true that other costs offset the labor cost advantages of the low-wage countries. The mill and the machinery could cost 100% more than in Britain, because most countries imported their machinery from Britain, and thus had to pay transport fees and import duties. Sometimes English mechanics had to be imported to erect the machinery. Shipping added about 25% to the cost of English machinery which was imported by U.S. mills, and about 20% overall to the cost of French mills built with English machinery.17 Interestingly, import duties rather than real cost elements were often the most significant factor in increasing costs. Russia in 1911 had a tariff on English machinery which was effectively 45% ad valorem, and the total cost of an erected mill in Russia was 63% greater than in England.
Table 1 does not take into account economizing on capital through the use of night work or longer work weeks. Table 2 shows the hours per year that mills in various countries operated around 1910, and the effect this had on the capital employed per spindle per year of output. In some cases, particularly Japan and Mexico, mills operated more than twice as long as in England. Table 2 also shows the manufacturing costs and the profit rates implied by these hours of operation, given the costs of inputs displayed in Table 1. With capital utilization considered, the profit rate (selling against British competition) in China should have been 32.9% not merely 22.1%, compared with the British rate of 8%. Also total manufacturing costs would be less than half those of Britain. With operating hours taken into effect, only Switzerland, among the low-wage countries, had a lower profit rate than Britain in international markets. Yet in practice only the Japanese, Indian, and Chinese mills provided any competition
The magnitude of these cost advantages can be appreciated in two ways. The first is to consider that $1 million invested in the Lancashire cotton industry to produce cloth for the Chinese market would accumulate at an 8% rate of return to $4.66 million in twenty years. If instead the same money were invested directly in the Shanghai industry, it would, based on this cost data alone, accumulate in twenty years to at least $54.24 million. If we also consider the differences in capital utilization, then the $1 million invested in China should have produced $295.46 million. British firms which invested $1 million for twenty years in the Indian industry should have accumulated $67.04 million. If the investors in India and China had ploughed these putative profits back into the industries, they would have grown at a minimum, without any outside investments, at rates of 23.4% and 32.9%. While Chinese and Indian mills were profitable prior to 1914, they were not this profitable, and production did not expand anywhere near as fast as these profit rates would suggest.22 The Indian industry grew at an annual rate of 5.1% in the twenty years before 1910, and the Chinese industry at 10% per year between 1895 and 1915. If there was such a difference in profit rates, why did Lancashire experience a major expansion between 1900 and 1910?
The prima facie case for differences in worker efficiency comes from examining how many machines each worker tended. The broadest measure is an index of the number of spindles and looms per worker. Since the relative importance of spinning and weaving varied across countries, the index is constructed by giving looms a weight of 1, ring spindles a weight of 0.011, and mule spindles a weight of 0.008. This corresponds to the relative numbers of workers needed to man weaving sheds, ring spinning mills and mule spinning mills in Britain in 1910. The measure shown in Table 3 is thus the number of "loom-equivalents" per worker in cotton textiles. Since some countries worked shifts, which would reduce the apparent machines per worker, the index is adjusted to a per-shift basis. The index varies from 0.46 in Greece to 2.97 in New England, or by a factor of 6.46:1. That is, in 1910 each worker in New England tended more than six times as much machinery per shift as did workers in the Greek industry
The only industries with lower costs than Britain, based on factor prices alone, were China and India. Even if we correct for capital utilization, only Japan, India, and China have lower costs and higher profits in competition with the British. Thus taking into account differences in labor efficiency, the only countries which ought to have been able to compete with the British on international markets were India, Japan, and China-the only countries which did in fact compete.
Differences in worker efficiency on the order of 6.5: 1 at the extreme will astonish many readers. They may attribute the difference to a number of causes, the first being that the aggregate measure of machines per worker does not take into account each country's selection of machinery types. Only North America, however, had by 1910 adopted the automatic loom on a large scale, which undoubtedly did somewhat inflate the number of loom-equivalents per worker in America. But a detailed examination of the staffing levels on ten particular machines, from those in the preparatory processes (such as carding) through the machines putting twist into the yarn and drawing it into finer strands, up to the final weaving of the yarn into cloth, reveals the same pattern as the aggregate data.
Yet the Special Agents Reports and other sources for 1910 show that low-wage countries were generally getting lower output per machine than the countries with high wages. In Switzerland machinery generally ran more slowly.24 Clark reports that in Austria: "The mills based their speeds on the English catalogues and production tables and run as near the speeds given as the quality of their help and material will permit."25 For France, Forrester noted only that "much of the machinery runs just as fast as in Lancashire."26 Ring spindles on the continent ran at an average speed which was "at least as low as in England," and lower than the speed in New England, the region with the highest wages 27
One of the tasks in spinning mills was doffing, which consisted of removing full spools of yarn from the spinning machines. This task would not allow much opportunity for substituting labor for capital, since the speed and output of the machines was independent of how many doffers were used. Yet where the numbers of spindles per doffer are recorded they show the same pattern as overall manning levels. Table 5 compares output per machine per hour in ring spinning and in weaving around 1910 in a number of different countries, controlling for the fineness of the yarn and the width and type of cloth. The outputs given are ounces per ring spindle spinning 20s count yarn per hour, and yards of plain cloth produced per loom weaving 36-inch-wide cloth with 40 to 50 threads per inch. Table 5 clearly shows that low-wage countries were not getting greater utilization of their capital except through running longer hours. In fact they seem to have gotten significantly lower output per hour around 1910. There is other, more fragmentary evidence for this. The efficiency of spindles or looms was defined as the%age of a machine's theoretical production, determined by the speed the machine was set to run at, which was actually achieved. British and American manufacturers seem to have assumed that ring spindles would achieve 90% efficiency and looms on plain cloth 87.5% efficiency.28 Spindle efficiency was reported at 90% in Japan, but only 77% in Russia, and loom efficiency 87.5% in Japan, but only 80% in Russia, and 77.5% in Italy.29
It was possible to some extent, in some processes, to save on cotton costs by using more labor. The price of cotton increased with the fiber length, and though yarn of a given fineness could be produced with cottons of various fiber lengths, the yarn was weaker if cheaper cotton was used. Weaker yarns broke more often in the manufacturing process, requiring more labor to repair the broken threads, but this could save on total costs if labor was cheap. Can we explain all or a significant part of the extraordinary manning levels in the low-wage countries by such a cost-minimizing strategy? The answer is no for five separate reasons. Only a few countries with high manning levels used inferior cotton. Those which did employed inferior cotton only on the coarser counts of yarn, though in these countries overmanning was just as great as machines making finer yarns. Even processes entirely unaffected by cotton quality, such as doffing, were overstaffed to as great an extent in low-wage countries. In processes where cotton quality did matter, workers did not spend much more time repairing yarn breakages. And finally where cheap cotton was used, it did not clearly create real cost savings since the value of the output appeared to be correspondingly reduced...Pearse visited one Indian mill in the 1920s, "a very good mill," where the cotton quality was high: "The mixings throughout are too good rather than too low."34 Importing long-staple Ugandan cotton, they produced counts as fine as the 60s, but even on the finest counts one male ring tender looked after only 260 spindles. In the United States, at such counts, a girl tender would oversee between 1,000 and 1,400 spindles, or four to five times as many. In another mill, where the average count was 29s (a count where American yarns were not much stronger than Indian yarns), a weaver still attended only two looms and a ring spinner tended only 170 spindles.35 Finally, in a mill manufacturing 60s from Egyptian and American cotton, which produced yarn as strong as that in the United States, a tender still watched only 300 spindles, less than a quarter of the number an American girl would attend at such counts.36 The strength of yarn does not seem to have much to do with differences in manning levels.
Another compelling argument that poor-quality cotton did not cause the low labor productivity of poor countries comes from the amount of time operatives spent fixing yarn breakages. In 1948 operatives in the card room of a British mill using machinery dating from before 1915 spent from 2.1% of their time repairing breakages in carding to 6.7% of their time on the fly frames. They spent the rest of their time emplacing and removing yarn packages, cleaning and monitoring the machinery, or resting.37 But in Japan, India, and China, where each operative had only one-fourth or fewer machines to attend as in Britain, over three-quarters of worker time is unaccounted for by British standards. Did they use that time to repair yarn breakages because of inferior cotton? No observers of mills in low-wage countries report any such difference in the nature of the operatives' tasks. Copeland notes in European mills only "as many breakages" as in the United States. Pearse remarked that Indian spinners "have hardly anything to do" and that "I watched two ring frames for three minutes; there was not a single end down [which means there were no broken threads]."
Though New England made many technological and organizational advances, Lancashire was the more important center because of its textile machinery industry, which supplied all or most of the textile machinery (and the manuals giving operating instructions and speeds) to most manufacturers outside North America. Thus textile mills around the world used very similar machinery. The first Indian mills, for example, were completely British in design and equipment, including even the stones that the machines sat upon.42 In 1927 an authority firmly stated that "the Bombay Cotton mills are in no way inferior to the Lancashire textile factories in their general equipment and manufacturing resources. "43 French mills erected just before World War I "would stand comparison with the best in England or the United States."44 In Brazil British mechanics sent by British machine makers erected most of the equipment.45 Where non-English machinery was used, as it was to some extent in weaving in Germany, Switzerland, France, and Italy, commentators found it generally equivalent to English machinery. Thus Besso notes, "The differences in style and design between machinery of Swiss and English make are no greater than those between the work of different English firms".46
Britain also supplied the managers and skilled workmen to operate the machinery and train the workforce in many countries, particularly India, China, Russia, Mexico, and Brazil. In 1895, of Bombay's 55 mill managers, 27 were British, as were 77 of the 190 weaving masters, spinning masters, carding masters, and engineers.47 At least one-third of the Chinese industry was under British management in 1915, as were some of the mills owned by Chinese entrepreneurs.48 Most Brazilian mills had British managers, room bosses, and engineers.49 Many non-British managers, particularly in Germany and Switzerland, received their training in Lancashire. Besso in fact claims that the average Swiss manager was better trained than the average English manager.50 Nor did the low-wage countries employ secondhand machinery which required more labor because it embodied a lower machine-labor ratio or because it was partly worn out. The literature never discusses an active secondhand machinery market, nor does it quote prices for secondhand machines in any countries. One reason for the absence of used machinery markets was the expense involved in setting up machinery in a mill. Clark's estimates for an Italian spinning mill show setting-up costs to be nearly 20% of the final cost of the erected machinery.51 These fitting-up expenses reduced the incentive for an established mill to sell off old but still serviceable equipment. In 1911, despite the industry's reasonably fast growth rate of 3.3% per year from 1860 to 1910, 22% of the spinning frames and 25% of the looms in the United States were more than twenty years old.52 Thus the low-wage countries generally purchased new machinery.
There is also direct evidence that the personal efficiency of workers in low-wage countries was low. American and British visitors to cotton mills elsewhere in the world almost universally noted the low quality of the labor. Clark remarked that English mills had the advantage over Swiss mills, "because of the greater efficiency of their operatives, which more than compensates for the higher wages paid."53 Besso concurred that "more work people are required to tend the machinery" in Switzerland and that "it is largely due to the inferiority of Swiss operatives."54 Clark noted of Egypt that "the inefficiency of the help available ... has probably had the largest effect in retarding the industry" and that "it takes two or three operatives to do the work of one English operative."55 Equivalent comments can be found for Germany, France, Italy, Spain, Peru, Japan, India, and China. [Apart from those in the Reports of Clark and Odell, such comments can be found in Forrester, Cotton Industry in France, pp. 42-43; Great Britain, House of Commons, Report of the Indian Factory Labour Commission, British Parliamentary Papers, Cd. 4292, 1908, p. 20; Great Britain, House of Commons, Minutes of Evidence taken before the Indian Industrial Commission, British Parliamentary Papers, Cmd. 234, 1919, p. 278; Moser, Cotton Textile Industry of the Far Eastern Countries, pp. 15, 68; Pearse, Cotton Industry of Japan and China, pp. 165, 172; Pearse, Cotton Industry of India, p. 53.]
And at least in India textile workers were small because they were inadequately nourished in childhood. The Indian Factory Labor Commission of 1908 found that a large sample of adult male factory operatives weighed on average only 105.4 pounds.62 This is certainly light by modern standards, and indeed by the standards of late-nineteenth-century British industry. In Britain in 1878 adult male artisans weighed on average 132.5 pounds, which probably represents fairly well the weights of Lancashire textile operatives.63
But the nature of operatives' tasks in textile work has to be considered. Few tasks demanded any literacy whatsoever. Most demanded only the ability to perform the same few simple operations repeatedly. It might take months to acquire the necessary dexterity and stamina, but the job for most workers was highly routine. Strength did not matter in most tasks, so small size should not have hampered operatives. Moser, visiting Indian mills in the 1920s, noted that "the workers in this mill, as in most others I visited, were lean of frame and appeared nimble. of hand. "64 Certainly if the current nutrition of operatives impeded their efficiency, the employers could cheaply supplement the operatives' diet and get great gains in output. In Japan employers maintained and fed employees in dormitories at the mill. And at least some German mills had company dining rooms with subsidized food.
Further evidence against the view that the workers in low-wage countries were inherently incapable comes from the New England industry. It employed the most efficient operatives, but in fact many workers came from countries whose own textile industries had very inefficient employees. The Immigration Commission Report in 1911 shows 27.8% of the workers in northern cotton mills to be Polish, Portuguese, Greek, or Italian, even though in Poland, Portugal, Greece, and Italy between three and six times as many workers were required per machine as in New England.65 The Poles, Portuguese, Greeks, and Italians who emigrated to New England probably had little formal education, and certainly little more than their compatriots in the textile mills of Europe. Perhaps only the most robust and energetic workers in these countries found their way to New England, but the necessary degree of selectivity would exceed any reasonable expectation. And why then, if immigrants are simply in general more vigorous, did Brazilian mills, which employed large numbers of Portuguese and some Italians and Germans, have operatives only one third as efficient as those in New England?66 Both New England and Lancashire mills employed many Irish migrants. Why did the Irish who happened to get on the boat to Lancashire exhibit lower efficiency as cotton operatives than those who instead took the boat to New England?
Whatever limits the efficiency of workers in low-wage countries seems to attach to the local environment, not to the workers themselves. The workers were, it appears, capable of tending to more machinery but either chose not to do so or were constrained from doing so. It should be noted that local textile managers were often not free to select the machinery manning levels they desired. The local managers in many low-wage countries insisted that the workers could tend more machinery but refused to do so. Pearse notes of Indian spinning mills in the 1920s that "there is only one side for each tender to look after; they have hardly anything to do, but they will not take on more spindles except at full rate of wages."67 He notes of another Bombay mill: "The operatives in this mill refuse to attend to more machinery. I watched two ring frames for three minutes; there was not a single end down, yet the work-people would not look after more than one side. They said that they are satisfied with the present wage, and that there are so many men who want work and cannot get it that it would be unfair if they were to attend to more machines."68
Moser, an American visitor to India in the 1920s, is even more adamant about the refusal of Indian workers to tend as many machines as they could ". . . it was apparent that they could easily have taken care of more, but they won't .... They cannot be persuaded by any exhortation, ambition, or the opportunity to increase their earnings."69 In 1928 attempts by management to increase the number of machines per worker led to the great Bombay mill strike.70
Similar stories crop up in Europe and Latin America. Clark noted that Mexico could not introduce automatic looms because "the Mexican operatives are very conservative, and as they have been accustomed to running two to four looms, usually not over three, it has as yet been found impossible to persuade them to run any larger number of automatic looms." 71 In the 1920s one big labor-policy issue for the new Soviet state was an attempt to force weavers to operate three looms as opposed to the traditional two. The attempt to introduce six or more looms per worker in Britain in the early twentieth century caused a long struggle between the managers and the weavers' union. In France between 1900 and 1904 many of the industrial disputes were "concerned with the increase in the number of looms to be minded and the piece rate to be paid."72...That would not explain why the machine manning disputes occurred at a level which was predictable from the real wage level of the country. In India the struggle was over two looms per weaver as opposed to one, in Russia three versus two, in France and Mexico, four versus three, and in Britain six versus four.
a study of European and American agriculture in the early nineteenth century finds great differences in the length of time it took workers to do simple manual tasks between the areas of advanced agriculture and high incomes and the backward areas. Grain was hand-threshed three times more quickly in North America than in Eastern Europe, with no discernible sacrifice of output in America. British and North American agricultural workers also were more efficient at simple manual tasks than medieval English peasants. [Gregory Clark, "Productivity Growth without Technical Change: European Agriculture before 1850" https://pdf.yt/d/rzhDtOaVOKEjnZYk / https://dl.dropboxusercontent.com/u/85192141/1987-clark-productivityagricultural.pdf (unpublished manuscript). See also Gregory Clark, "The Cost of Capital and Medieval Agricultural Technique" http://www.researchgate.net/publication/222450860_The_cost_of_capital_and_medieval_agricultural_technique/file/9c9605212962a1b948.pdf (unpublished manuscript).]
Another implication is that if the differences in worker efficiency had been eliminated, then wages of cotton textile workers in underdeveloped countries could have been raised close to those of the developed countries, since their nonlabor costs did not exceed those of the developed countries by much, as can be seen in Tables 1 and 2. Had the local industries been competing in a free market, the equilibrium wages (given the nonlabor costs) would have been $4.09 per week in the United States, $4.35 in China, $3.93 in India, $3.01 in Japan, $3.16 in Russia, $3.96 in Italy, $4.12 in France, and $3.80 in Germany. Britain, with its low-cost machinery industry and cheap coal, could have paid $5.00 per week. If, as I anticipate, other industries were like textiles, then the major source of the underdevelopment of poor countries was the inefficiency of their labor, rather than their inability to absorb modern industrial technology.
One remarkable feature of world industrialization since 1780 has been the extent to which workers have migrated to a few industrial centers, instead of capital and management migrating to the available supplies of cheap labor. The costs of movement for workers would seem to be much higher than those for other factors of production, given the difficulties of adjusting to new languages and cultures and the desire to remain close to parents and other relatives and friends. The existence of these costs is borne out by the substantial wage differentials which often exist even within the same country. Adam Smith, observing that while commodity prices had equalized across Britain in 1776 there were substantial differences in wages between parishes only a few miles apart, commented "it appears evidently from experience that a man is of all sorts of luggage the most difficult to be transported."74 In Germany in 1913 per capita income varied by over two to one between different provinces.7
In analyzing choice of technique between countries or over time it is often taken that there is a fundamental unit of measurement, an hour or day of unskilled labor, whose price relative to other factors of production determines the appropriate choice of technique. But this study shows that there is no such universal constant. The true relative cost of capital and labor cannot be inferred simply from the price of each, and in fact in textiles varied within rather narrow bounds. Thus while wages in textiles varied by a factor of 16 to 1, the true labor cost varied by only 3 to 1, and some of this variation is measurement error. Nor did any monotonic connection exist between wages and the true labor cost. Some low-wage countries had labor costs quite as high as those of the highest wage areas. Much discussion has considered whether and how the high cost of American labor influenced U.S. technological development.77 If cotton textiles are any guide, it may well have been that nineteenth-century American labor costs were no higher than those elsewhere, undercutting the premise of this whole line of enquiry. Outputs per worker increased greatly in all the national textile industries over the nineteenth and early twentieth centuries. In England in 1850 the average weaver tended only 2.2 power looms compared with 3.44 in 1906, despite the fact that looms in 1906 were about 50% faster than those of 1850.78 The number of spindles per operative increased also. In 1833 the mule spinning frame had 440 spindles on average, but by 1910 this had increased to 1080, with no increase in workers per frame, despite the fact that the speed of spindles had more than doubled.79 Differences in manning levels among countries suggest that it is unsafe to infer that the increase in output per worker resulted solely from technical progress, whether this was embodied in new machinery or disembodied learning-by-doing. The same forces that created national differences in the efficiency of labor could be operating over time, and labor intensification could thus be an important source of productivity increases"
"In 1910 one New England cotton textile operative performed as much work as 1.5 British, 2.3 German, and nearly 6 Greek, Japanese, Indian, or Chinese workers. Input substitution, and differences in technology, management, and workers' training or inherent abilities do not explain this. Instead local culture seems to have determined worker performance. Such differences, if widespread, would explain much of the international variation in wages. They also have important consequences for understanding labor migration, the choice of technique, and the sources of economic growth.
Perhaps the most prevalent modern view is that poor countries have remained poor because they cannot absorb the technologies of the advanced industrialized countries. To explain this, some cite the lack of education in poor countries; others, the lack of management skills or an entrepreneurial tradition, the lack of appropriate institutions, or the economies of scale inherent in the new industrial technologies which hinder their small-scale or isolated adoption.2 In this article I examine underdevelopment through a detailed study of one industry, cotton textiles, in the early twentieth century. For most of the countries I studied, factory production of textiles was a first step on the path to industrialization. I conclude that the failure of poor countries to dominate textile production in the way their labor cost advantage would suggest proceeds overwhelmingly from inefficient labor rather than from failure to import technology or management skills, or from failure of local capital or input markets, or from scale economies within the textile factories or at the industry level. In cotton textiles all these problems proved minor
When Britain launched the modern factory system in textile production in the late eighteenth century, people had every reason to expect that the Industrial Revolution would soon be a worldwide phenomenon. The new textile technologies were not particularly complex, most textile factory jobs did not require great skill, and every country had a ready local market for yarn and cloth. Despite attempts by the British to prevent the export of industrial technology-the emigration of artisans was prohibited until 1825, and most machinery exports were forbidden until 1842-the new machinery and methods spread quickly to North America and to Europe. By 1825, despite the prohibitions, an estimated two thousand British artisans had emigrated to the Continent.3 Thus while Britain established the first modern spinning factory in 1771, the French had established one by 1778, one was built in the Ruhr by 1784, the United States and Prussia had spinning mills by 1791, Russia had its first spinning mill by 1793, Switzerland by 1794, the Netherlands by 1795, and Belgium and Saxony by 1799.4 Even India, 12,000 miles by sea from Britain and very underdeveloped, joined the group in 1817, when a British trading firm shipped out to Calcutta a modern mill powered by steam.5 Nicolas Koechlin, a prominent Alsatian cotton manufacturer, reported in 1834 that when several leading Alsatian cotton manufacturers visited English factories in 1833 they "saw nothing particularly interesting" and "except in the high numbers, Alsace was not a whit behind hand."'6 Moreover, the Continental manufacturers apparently had great advantages in terms of the cost of labor in the 1830s. Hourly wages compared with Manchester were about 37% in Mulhausen in the Alsace, 28% in Zurich, 25% in some mills in the Vosges, 38% in Ghent in Belgium, 47% in Rouen, about 36% in Austria, and less than 25% in Prussia.7 Continental manufacturers did face higher costs for capital, machinery, repairs, and power (if they were using coal). But in 1834 wages were about two-thirds of the conversion costs (the cost of turning raw cotton into yarn or cloth) in spinning in Manchester, so that the costs of capital, machinery, repairs, and power would have to be at least 128% higher in Alsace to equalize costs, and 150% higher in Prussia.8 The cost differences reported by Koechlin are only of the order of 70% for Alsace.9 As these textile industries developed, there was every expectation that such cost differentials would decline substantially, when the size of the local machinery and the repair and capital markets increased, and so too would decline British dominance of export markets.
Yet despite the great labor cost advantages of the Continental countries, and the even greater labor cost advantages of the Asiatic countries, few of them were able to compete with the British outside their protected home markets before World War I. In 1911, 140 years after the establishment of the first spinning mill, 40% of the factory cotton spindles in the world were still in Britain; of those outside Britain 22% were in the United States or Canada, the only countries with higher wage costs than Britain. Germany, France, Russia, Belgium, Switzerland, Italy, Spain, Portugal, Austria, India, Japan, China, Mexico, and all the other low-wage competitors had only 39% of the world stock of spindles.'0 More importantly in 1911 Britain still accounted for 81% of the net exports of cotton yarn and cloth in international trade.1 Britain faced little competition in unprotected markets, since except for the Asiatic countries with their extraordinarily low wages, manufacturing costs in all the low-wage countries were higher than in Britain, making tariff protection necessary if the British were not to overwhelm even their domestic markets.
...given the labor cost advantages of Britain's competitors, what really requires explanation is not the ultimate demise of the British industry but why British textiles continued to thrive for so long. Second, the specific failures attributed to British managers prior to 1914 had very little impact on total costs and should have been completely swamped by labor cost effects. For example, one of the major mistakes attributed to British managers was a failure to switch from mule spinning to ring spinning.13 William Lazonick has argued that this failure stemmed from the more fundamental error of retaining the horizontal specialization of the British industry into spinning and weaving factories. With separate spinning and weaving mills, yarn has to be shipped from the spinning mills to the weaving sheds. It cost more to ship the ring weft (filling) yarn than to ship mule weft yarn, which he argued made it less costly to retain mules for weft spinning.'4 When this transport cost is calculated, however, the most it could have cost the British for sticking with mules for this reason could be only 1.4% of total value added in cotton textile manufacturing.'5 In comparison, British wages per hour were about 6.5 times those of the Indian industry, and since labor costs were about two-thirds of the value added in the industry, the British suffered a 57% cost disadvantage on account of wages. The correct question is not why bad choice of technique doomed the British industry, but why an industry apparently so burdened by high costs for its major input could remain so long successful, competing against a host of low-wage competitors.
To illustrate the magnitude of Britain's competitive disadvantage, consider the costs of inputs other than labor in cotton textile production. Cotton was the major input, accounting for about two-thirds of the total cost of cotton cloth, but its cost to Britain's competitors was generally the same or less. Britain had to import all its cotton, and in 1910 it actually cost slightly more to ship cotton from New Orleans to Liverpool than to ship it to Fall River in New England, or to Bremen, Le Havre, Antwerp, or Genoa on the Continent.
Table 1 shows the cost of labor per 55-hour week, the cost of coal per ton, and the cost of a complete spinning and weaving mill (including erection costs) in 15 countries or regions, the data gathered principally from the Special Agents Series of the U.S. Bureau of Foreign and Domestic Commerce. These reports, together with the U.S. Tariff Board report on Cotton Manufactures and other sources, provide a detailed summary of conditions in the international industry around 1910. As the table shows, labor costs varied by extraordinary amounts, Chinese labor costing 10.8% of that in Britain, and a mere 6.1% of that in the United States. It is true that other costs offset the labor cost advantages of the low-wage countries. The mill and the machinery could cost 100% more than in Britain, because most countries imported their machinery from Britain, and thus had to pay transport fees and import duties. Sometimes English mechanics had to be imported to erect the machinery. Shipping added about 25% to the cost of English machinery which was imported by U.S. mills, and about 20% overall to the cost of French mills built with English machinery.17 Interestingly, import duties rather than real cost elements were often the most significant factor in increasing costs. Russia in 1911 had a tariff on English machinery which was effectively 45% ad valorem, and the total cost of an erected mill in Russia was 63% greater than in England.
Table 1 does not take into account economizing on capital through the use of night work or longer work weeks. Table 2 shows the hours per year that mills in various countries operated around 1910, and the effect this had on the capital employed per spindle per year of output. In some cases, particularly Japan and Mexico, mills operated more than twice as long as in England. Table 2 also shows the manufacturing costs and the profit rates implied by these hours of operation, given the costs of inputs displayed in Table 1. With capital utilization considered, the profit rate (selling against British competition) in China should have been 32.9% not merely 22.1%, compared with the British rate of 8%. Also total manufacturing costs would be less than half those of Britain. With operating hours taken into effect, only Switzerland, among the low-wage countries, had a lower profit rate than Britain in international markets. Yet in practice only the Japanese, Indian, and Chinese mills provided any competition
The magnitude of these cost advantages can be appreciated in two ways. The first is to consider that $1 million invested in the Lancashire cotton industry to produce cloth for the Chinese market would accumulate at an 8% rate of return to $4.66 million in twenty years. If instead the same money were invested directly in the Shanghai industry, it would, based on this cost data alone, accumulate in twenty years to at least $54.24 million. If we also consider the differences in capital utilization, then the $1 million invested in China should have produced $295.46 million. British firms which invested $1 million for twenty years in the Indian industry should have accumulated $67.04 million. If the investors in India and China had ploughed these putative profits back into the industries, they would have grown at a minimum, without any outside investments, at rates of 23.4% and 32.9%. While Chinese and Indian mills were profitable prior to 1914, they were not this profitable, and production did not expand anywhere near as fast as these profit rates would suggest.22 The Indian industry grew at an annual rate of 5.1% in the twenty years before 1910, and the Chinese industry at 10% per year between 1895 and 1915. If there was such a difference in profit rates, why did Lancashire experience a major expansion between 1900 and 1910?
The prima facie case for differences in worker efficiency comes from examining how many machines each worker tended. The broadest measure is an index of the number of spindles and looms per worker. Since the relative importance of spinning and weaving varied across countries, the index is constructed by giving looms a weight of 1, ring spindles a weight of 0.011, and mule spindles a weight of 0.008. This corresponds to the relative numbers of workers needed to man weaving sheds, ring spinning mills and mule spinning mills in Britain in 1910. The measure shown in Table 3 is thus the number of "loom-equivalents" per worker in cotton textiles. Since some countries worked shifts, which would reduce the apparent machines per worker, the index is adjusted to a per-shift basis. The index varies from 0.46 in Greece to 2.97 in New England, or by a factor of 6.46:1. That is, in 1910 each worker in New England tended more than six times as much machinery per shift as did workers in the Greek industry
The only industries with lower costs than Britain, based on factor prices alone, were China and India. Even if we correct for capital utilization, only Japan, India, and China have lower costs and higher profits in competition with the British. Thus taking into account differences in labor efficiency, the only countries which ought to have been able to compete with the British on international markets were India, Japan, and China-the only countries which did in fact compete.
Differences in worker efficiency on the order of 6.5: 1 at the extreme will astonish many readers. They may attribute the difference to a number of causes, the first being that the aggregate measure of machines per worker does not take into account each country's selection of machinery types. Only North America, however, had by 1910 adopted the automatic loom on a large scale, which undoubtedly did somewhat inflate the number of loom-equivalents per worker in America. But a detailed examination of the staffing levels on ten particular machines, from those in the preparatory processes (such as carding) through the machines putting twist into the yarn and drawing it into finer strands, up to the final weaving of the yarn into cloth, reveals the same pattern as the aggregate data.
Yet the Special Agents Reports and other sources for 1910 show that low-wage countries were generally getting lower output per machine than the countries with high wages. In Switzerland machinery generally ran more slowly.24 Clark reports that in Austria: "The mills based their speeds on the English catalogues and production tables and run as near the speeds given as the quality of their help and material will permit."25 For France, Forrester noted only that "much of the machinery runs just as fast as in Lancashire."26 Ring spindles on the continent ran at an average speed which was "at least as low as in England," and lower than the speed in New England, the region with the highest wages 27
One of the tasks in spinning mills was doffing, which consisted of removing full spools of yarn from the spinning machines. This task would not allow much opportunity for substituting labor for capital, since the speed and output of the machines was independent of how many doffers were used. Yet where the numbers of spindles per doffer are recorded they show the same pattern as overall manning levels. Table 5 compares output per machine per hour in ring spinning and in weaving around 1910 in a number of different countries, controlling for the fineness of the yarn and the width and type of cloth. The outputs given are ounces per ring spindle spinning 20s count yarn per hour, and yards of plain cloth produced per loom weaving 36-inch-wide cloth with 40 to 50 threads per inch. Table 5 clearly shows that low-wage countries were not getting greater utilization of their capital except through running longer hours. In fact they seem to have gotten significantly lower output per hour around 1910. There is other, more fragmentary evidence for this. The efficiency of spindles or looms was defined as the%age of a machine's theoretical production, determined by the speed the machine was set to run at, which was actually achieved. British and American manufacturers seem to have assumed that ring spindles would achieve 90% efficiency and looms on plain cloth 87.5% efficiency.28 Spindle efficiency was reported at 90% in Japan, but only 77% in Russia, and loom efficiency 87.5% in Japan, but only 80% in Russia, and 77.5% in Italy.29
It was possible to some extent, in some processes, to save on cotton costs by using more labor. The price of cotton increased with the fiber length, and though yarn of a given fineness could be produced with cottons of various fiber lengths, the yarn was weaker if cheaper cotton was used. Weaker yarns broke more often in the manufacturing process, requiring more labor to repair the broken threads, but this could save on total costs if labor was cheap. Can we explain all or a significant part of the extraordinary manning levels in the low-wage countries by such a cost-minimizing strategy? The answer is no for five separate reasons. Only a few countries with high manning levels used inferior cotton. Those which did employed inferior cotton only on the coarser counts of yarn, though in these countries overmanning was just as great as machines making finer yarns. Even processes entirely unaffected by cotton quality, such as doffing, were overstaffed to as great an extent in low-wage countries. In processes where cotton quality did matter, workers did not spend much more time repairing yarn breakages. And finally where cheap cotton was used, it did not clearly create real cost savings since the value of the output appeared to be correspondingly reduced...Pearse visited one Indian mill in the 1920s, "a very good mill," where the cotton quality was high: "The mixings throughout are too good rather than too low."34 Importing long-staple Ugandan cotton, they produced counts as fine as the 60s, but even on the finest counts one male ring tender looked after only 260 spindles. In the United States, at such counts, a girl tender would oversee between 1,000 and 1,400 spindles, or four to five times as many. In another mill, where the average count was 29s (a count where American yarns were not much stronger than Indian yarns), a weaver still attended only two looms and a ring spinner tended only 170 spindles.35 Finally, in a mill manufacturing 60s from Egyptian and American cotton, which produced yarn as strong as that in the United States, a tender still watched only 300 spindles, less than a quarter of the number an American girl would attend at such counts.36 The strength of yarn does not seem to have much to do with differences in manning levels.
Another compelling argument that poor-quality cotton did not cause the low labor productivity of poor countries comes from the amount of time operatives spent fixing yarn breakages. In 1948 operatives in the card room of a British mill using machinery dating from before 1915 spent from 2.1% of their time repairing breakages in carding to 6.7% of their time on the fly frames. They spent the rest of their time emplacing and removing yarn packages, cleaning and monitoring the machinery, or resting.37 But in Japan, India, and China, where each operative had only one-fourth or fewer machines to attend as in Britain, over three-quarters of worker time is unaccounted for by British standards. Did they use that time to repair yarn breakages because of inferior cotton? No observers of mills in low-wage countries report any such difference in the nature of the operatives' tasks. Copeland notes in European mills only "as many breakages" as in the United States. Pearse remarked that Indian spinners "have hardly anything to do" and that "I watched two ring frames for three minutes; there was not a single end down [which means there were no broken threads]."
Though New England made many technological and organizational advances, Lancashire was the more important center because of its textile machinery industry, which supplied all or most of the textile machinery (and the manuals giving operating instructions and speeds) to most manufacturers outside North America. Thus textile mills around the world used very similar machinery. The first Indian mills, for example, were completely British in design and equipment, including even the stones that the machines sat upon.42 In 1927 an authority firmly stated that "the Bombay Cotton mills are in no way inferior to the Lancashire textile factories in their general equipment and manufacturing resources. "43 French mills erected just before World War I "would stand comparison with the best in England or the United States."44 In Brazil British mechanics sent by British machine makers erected most of the equipment.45 Where non-English machinery was used, as it was to some extent in weaving in Germany, Switzerland, France, and Italy, commentators found it generally equivalent to English machinery. Thus Besso notes, "The differences in style and design between machinery of Swiss and English make are no greater than those between the work of different English firms".46
Britain also supplied the managers and skilled workmen to operate the machinery and train the workforce in many countries, particularly India, China, Russia, Mexico, and Brazil. In 1895, of Bombay's 55 mill managers, 27 were British, as were 77 of the 190 weaving masters, spinning masters, carding masters, and engineers.47 At least one-third of the Chinese industry was under British management in 1915, as were some of the mills owned by Chinese entrepreneurs.48 Most Brazilian mills had British managers, room bosses, and engineers.49 Many non-British managers, particularly in Germany and Switzerland, received their training in Lancashire. Besso in fact claims that the average Swiss manager was better trained than the average English manager.50 Nor did the low-wage countries employ secondhand machinery which required more labor because it embodied a lower machine-labor ratio or because it was partly worn out. The literature never discusses an active secondhand machinery market, nor does it quote prices for secondhand machines in any countries. One reason for the absence of used machinery markets was the expense involved in setting up machinery in a mill. Clark's estimates for an Italian spinning mill show setting-up costs to be nearly 20% of the final cost of the erected machinery.51 These fitting-up expenses reduced the incentive for an established mill to sell off old but still serviceable equipment. In 1911, despite the industry's reasonably fast growth rate of 3.3% per year from 1860 to 1910, 22% of the spinning frames and 25% of the looms in the United States were more than twenty years old.52 Thus the low-wage countries generally purchased new machinery.
There is also direct evidence that the personal efficiency of workers in low-wage countries was low. American and British visitors to cotton mills elsewhere in the world almost universally noted the low quality of the labor. Clark remarked that English mills had the advantage over Swiss mills, "because of the greater efficiency of their operatives, which more than compensates for the higher wages paid."53 Besso concurred that "more work people are required to tend the machinery" in Switzerland and that "it is largely due to the inferiority of Swiss operatives."54 Clark noted of Egypt that "the inefficiency of the help available ... has probably had the largest effect in retarding the industry" and that "it takes two or three operatives to do the work of one English operative."55 Equivalent comments can be found for Germany, France, Italy, Spain, Peru, Japan, India, and China. [Apart from those in the Reports of Clark and Odell, such comments can be found in Forrester, Cotton Industry in France, pp. 42-43; Great Britain, House of Commons, Report of the Indian Factory Labour Commission, British Parliamentary Papers, Cd. 4292, 1908, p. 20; Great Britain, House of Commons, Minutes of Evidence taken before the Indian Industrial Commission, British Parliamentary Papers, Cmd. 234, 1919, p. 278; Moser, Cotton Textile Industry of the Far Eastern Countries, pp. 15, 68; Pearse, Cotton Industry of Japan and China, pp. 165, 172; Pearse, Cotton Industry of India, p. 53.]
And at least in India textile workers were small because they were inadequately nourished in childhood. The Indian Factory Labor Commission of 1908 found that a large sample of adult male factory operatives weighed on average only 105.4 pounds.62 This is certainly light by modern standards, and indeed by the standards of late-nineteenth-century British industry. In Britain in 1878 adult male artisans weighed on average 132.5 pounds, which probably represents fairly well the weights of Lancashire textile operatives.63
But the nature of operatives' tasks in textile work has to be considered. Few tasks demanded any literacy whatsoever. Most demanded only the ability to perform the same few simple operations repeatedly. It might take months to acquire the necessary dexterity and stamina, but the job for most workers was highly routine. Strength did not matter in most tasks, so small size should not have hampered operatives. Moser, visiting Indian mills in the 1920s, noted that "the workers in this mill, as in most others I visited, were lean of frame and appeared nimble. of hand. "64 Certainly if the current nutrition of operatives impeded their efficiency, the employers could cheaply supplement the operatives' diet and get great gains in output. In Japan employers maintained and fed employees in dormitories at the mill. And at least some German mills had company dining rooms with subsidized food.
Further evidence against the view that the workers in low-wage countries were inherently incapable comes from the New England industry. It employed the most efficient operatives, but in fact many workers came from countries whose own textile industries had very inefficient employees. The Immigration Commission Report in 1911 shows 27.8% of the workers in northern cotton mills to be Polish, Portuguese, Greek, or Italian, even though in Poland, Portugal, Greece, and Italy between three and six times as many workers were required per machine as in New England.65 The Poles, Portuguese, Greeks, and Italians who emigrated to New England probably had little formal education, and certainly little more than their compatriots in the textile mills of Europe. Perhaps only the most robust and energetic workers in these countries found their way to New England, but the necessary degree of selectivity would exceed any reasonable expectation. And why then, if immigrants are simply in general more vigorous, did Brazilian mills, which employed large numbers of Portuguese and some Italians and Germans, have operatives only one third as efficient as those in New England?66 Both New England and Lancashire mills employed many Irish migrants. Why did the Irish who happened to get on the boat to Lancashire exhibit lower efficiency as cotton operatives than those who instead took the boat to New England?
Whatever limits the efficiency of workers in low-wage countries seems to attach to the local environment, not to the workers themselves. The workers were, it appears, capable of tending to more machinery but either chose not to do so or were constrained from doing so. It should be noted that local textile managers were often not free to select the machinery manning levels they desired. The local managers in many low-wage countries insisted that the workers could tend more machinery but refused to do so. Pearse notes of Indian spinning mills in the 1920s that "there is only one side for each tender to look after; they have hardly anything to do, but they will not take on more spindles except at full rate of wages."67 He notes of another Bombay mill: "The operatives in this mill refuse to attend to more machinery. I watched two ring frames for three minutes; there was not a single end down, yet the work-people would not look after more than one side. They said that they are satisfied with the present wage, and that there are so many men who want work and cannot get it that it would be unfair if they were to attend to more machines."68
Moser, an American visitor to India in the 1920s, is even more adamant about the refusal of Indian workers to tend as many machines as they could ". . . it was apparent that they could easily have taken care of more, but they won't .... They cannot be persuaded by any exhortation, ambition, or the opportunity to increase their earnings."69 In 1928 attempts by management to increase the number of machines per worker led to the great Bombay mill strike.70
Similar stories crop up in Europe and Latin America. Clark noted that Mexico could not introduce automatic looms because "the Mexican operatives are very conservative, and as they have been accustomed to running two to four looms, usually not over three, it has as yet been found impossible to persuade them to run any larger number of automatic looms." 71 In the 1920s one big labor-policy issue for the new Soviet state was an attempt to force weavers to operate three looms as opposed to the traditional two. The attempt to introduce six or more looms per worker in Britain in the early twentieth century caused a long struggle between the managers and the weavers' union. In France between 1900 and 1904 many of the industrial disputes were "concerned with the increase in the number of looms to be minded and the piece rate to be paid."72...That would not explain why the machine manning disputes occurred at a level which was predictable from the real wage level of the country. In India the struggle was over two looms per weaver as opposed to one, in Russia three versus two, in France and Mexico, four versus three, and in Britain six versus four.
a study of European and American agriculture in the early nineteenth century finds great differences in the length of time it took workers to do simple manual tasks between the areas of advanced agriculture and high incomes and the backward areas. Grain was hand-threshed three times more quickly in North America than in Eastern Europe, with no discernible sacrifice of output in America. British and North American agricultural workers also were more efficient at simple manual tasks than medieval English peasants. [Gregory Clark, "Productivity Growth without Technical Change: European Agriculture before 1850" https://pdf.yt/d/rzhDtOaVOKEjnZYk / https://dl.dropboxusercontent.com/u/85192141/1987-clark-productivityagricultural.pdf (unpublished manuscript). See also Gregory Clark, "The Cost of Capital and Medieval Agricultural Technique" http://www.researchgate.net/publication/222450860_The_cost_of_capital_and_medieval_agricultural_technique/file/9c9605212962a1b948.pdf (unpublished manuscript).]
Another implication is that if the differences in worker efficiency had been eliminated, then wages of cotton textile workers in underdeveloped countries could have been raised close to those of the developed countries, since their nonlabor costs did not exceed those of the developed countries by much, as can be seen in Tables 1 and 2. Had the local industries been competing in a free market, the equilibrium wages (given the nonlabor costs) would have been $4.09 per week in the United States, $4.35 in China, $3.93 in India, $3.01 in Japan, $3.16 in Russia, $3.96 in Italy, $4.12 in France, and $3.80 in Germany. Britain, with its low-cost machinery industry and cheap coal, could have paid $5.00 per week. If, as I anticipate, other industries were like textiles, then the major source of the underdevelopment of poor countries was the inefficiency of their labor, rather than their inability to absorb modern industrial technology.
One remarkable feature of world industrialization since 1780 has been the extent to which workers have migrated to a few industrial centers, instead of capital and management migrating to the available supplies of cheap labor. The costs of movement for workers would seem to be much higher than those for other factors of production, given the difficulties of adjusting to new languages and cultures and the desire to remain close to parents and other relatives and friends. The existence of these costs is borne out by the substantial wage differentials which often exist even within the same country. Adam Smith, observing that while commodity prices had equalized across Britain in 1776 there were substantial differences in wages between parishes only a few miles apart, commented "it appears evidently from experience that a man is of all sorts of luggage the most difficult to be transported."74 In Germany in 1913 per capita income varied by over two to one between different provinces.7
In analyzing choice of technique between countries or over time it is often taken that there is a fundamental unit of measurement, an hour or day of unskilled labor, whose price relative to other factors of production determines the appropriate choice of technique. But this study shows that there is no such universal constant. The true relative cost of capital and labor cannot be inferred simply from the price of each, and in fact in textiles varied within rather narrow bounds. Thus while wages in textiles varied by a factor of 16 to 1, the true labor cost varied by only 3 to 1, and some of this variation is measurement error. Nor did any monotonic connection exist between wages and the true labor cost. Some low-wage countries had labor costs quite as high as those of the highest wage areas. Much discussion has considered whether and how the high cost of American labor influenced U.S. technological development.77 If cotton textiles are any guide, it may well have been that nineteenth-century American labor costs were no higher than those elsewhere, undercutting the premise of this whole line of enquiry. Outputs per worker increased greatly in all the national textile industries over the nineteenth and early twentieth centuries. In England in 1850 the average weaver tended only 2.2 power looms compared with 3.44 in 1906, despite the fact that looms in 1906 were about 50% faster than those of 1850.78 The number of spindles per operative increased also. In 1833 the mule spinning frame had 440 spindles on average, but by 1910 this had increased to 1080, with no increase in workers per frame, despite the fact that the speed of spindles had more than doubled.79 Differences in manning levels among countries suggest that it is unsafe to infer that the increase in output per worker resulted solely from technical progress, whether this was embodied in new machinery or disembodied learning-by-doing. The same forces that created national differences in the efficiency of labor could be operating over time, and labor intensification could thus be an important source of productivity increases"