The carnival of publicity attending the publication of Time on the Cross suggests that the authors, Robert Fogel and Stanley Engerman, desire an audience embracing not only econometric historians but all reasonable men. I am not an econometric historian or a specialist in the history of slavery, but I am a reasonable man and, as such, entitled to judge the plausibility of the authors’ argument. Fogel and Engerman contend that slave labor was more efficient than free labor. This contention appears to rest on a dubious inference that vitiates several of the book’s most striking conclusions.
The most troublesome phase of any quantitative study is the translation of numerical procedures into plain English. In their research and calculations, Fogel and Engerman may have considered all the objections raised below. But even if their conclusions turn out to be procedurally well founded, their presentation still fails, for they have not exposed to the reader’s view any process of reasoning adequate to justify their conclusions.
The crux of the problem is that Fogel and Engerman appear to have drawn unjustifiable inferences from data based on the “geometric index of total factor productivity”—inferences which that index is inherently unable to support. The index is essentially a ratio of output to input. They use it to compare the “efficiency” of Southern (slave) agriculture with Northern (free) agriculture. They conclude that in the single year tested, 1860, “Southern agriculture as a whole was about 35 percent more efficient than northern agriculture….”1
One would never know from the authors’ discussion of this index that economists are not entirely sure what it measures, or what causal factors it reflects, even in its most conventional applications. Fogel and Engerman’s interpretation of it as a measure of efficiency is defensible, but it would have been delightfully frank of them to tell their readers that Evsey Domar, the economist who formulated the “geometric” version of the index, was so wary of misinterpretation that he called it simply the “Residual,” rather than an index of efficiency. Commenting on a comparative study of the relative efficiency of the USSR and the US, Domar noted that “if the Index shows that the average factor productivity in one country is markedly inferior to another, greater efficiency of the latter is not an unreasonable hypothesis. But there may be other explanations as well.” Another economist referred to this entire class of aggregate productivity indices as a “measure of our ignorance.”2
But let us grant that the index can be construed as a measure of efficiency in some sense. What does it mean in the particular case—a static comparison of Northern and Southern agricultural production in the year 1860—to which Fogel and Engerman apply it?
The critical difficulty arises from the fact that the index, which is basically a ratio of output to input, states output not in physical units—bushels, bales, or pounds per worker—but in total market value of the product. It is, therefore, as much a measure of profitability as of the intrinsic technological efficacy of the production process. The authors explain in their technical supplement that the figures for Northern output are derived from estimates of “income originating in agriculture for the year 1860,” and that the figures used for Southern output are computed from a sample of physical output data “weighted by prices.”3 Both procedures permit the index to vary with the market value of the product. Indeed, there would appear to be no other way to aggregate total output—barrels of tobacco, plus pounds of chicken, plus bushels of peas, plus bales of cotton—except by reducing everything to dollar value.
Because the index expresses output in market value, the “efficiency” it measures has an Alice-in-Wonderland quality under certain unusual conditions. If the demand for a product, and therefore its price, is sufficiently high, its producers may appear to be very “efficient,” according to this index, even if their work habits are slovenly and their mode of production irrational. Since there was an extraordinary international demand for the dominant Southern agricultural product, cotton, it is possible that this exceptional demand situation explains (in part or even wholly) the so-called “efficiency” of Southern agriculture. The possibility may prove ephemeral, but it should be explored. At no point in their presentation, either in the expository volume or in the supplementary volume devoted to evidence and method, do the authors openly entertain this possibility. On the contrary, they attribute the “efficiency” gap primarily to “the combination of the superior management of planters and the superior quality of black labor.”4 This is not a permissible inference, given the limitations of the index.
If the index measured economic output in physical units—bushels or bales per worker—then it could serve as a basis for inferences about the comparative diligence of workers and skill of managers in the two regions. If workers on one farm produce eight bales per hand and workers on another farm using similar inputs of capital and land produce only four, there can be no doubt that the first farm is more efficient (although before assigning merit to the workers involved we will want to know more about the role of impersonal factors such as climate, soil quality, economies of scale, etc.). But since the index used by Fogel and Engerman measures output by the price the product brings in the market, rather than its physical volume, the index is necessarily influenced not only by the behavior of producers but also by the behavior of consumers.
Price, after all, is a function of both supply (efficiency of producers) and demand (eagerness of consumers). If a Southern farm produces cotton worth $800 per hand and a Northern farm using similar inputs produces wheat worth only $400 per hand, the index used by Fogel and Engerman will unhesitatingly rate the Southern farm more “efficient.” But does this superior “efficiency” reflect superior productive performance—greater energy, perseverance, rationality of organization, more pounds per man hour?—or does it reflect merely the different intensities of consumer demand for cotton and wheat? The index used by Fogel and Engerman cannot distinguish between these two quite different sources of “efficiency.”5
Some economists have recognized this limitation of the index, but the literature on productivity often ignores it because what finally counts, from the economists’ most common perspective, is “the bottom line”—profit. To a businessman wondering where to invest his money, or an economist looking for the sources of growth, the efficiency of an enterprise is adequately expressed by the ratio between income taken out and investment put in. The diligence of the work force and the rationality of management are, in the last analysis, irrelevant if intensive consumer demand for the product creates a large income relative to a given investment of resources.
But there are situations in which the limitations of the index become critical. Thus, to take a fanciful case, if one applied the economists’ index of efficiency to various enterprises in Holland during the Great Tulip Craze, even the most slovenly tulip producer might look highly “efficient,” compared to other agriculturists, simply because of the booming market for tulip bulbs. Here the index would not be a true measure of the energy of the producer or the rationality of the production process, but would merely register the extraordinary value placed on the product by consumers. Likewise, to take another case equally hypothetical but not dependent on transitory boom conditions, it is conceivable that Turkish opium growers, using primitive agricultural methods, might consistently be more “efficient” than highly mechanized, scientifically trained wheat farmers in the United States. Again, strong demand can create high “efficiency,” especially when producers are partially shielded from competition by legal or natural barriers to production.
The economists’ index of efficiency is a perfectly good guide to economic growth and profit maximization. I do not deny the validity or usefulness of the index itself, I challenge only the careless use to which Fogel and Engerman put it. The question is: What kinds of inferences can the index justify in this particular case? The answer is clear: If an unusual demand situation exists, the index cannot serve by itself as a basis for inferences about the quality of the labor force or the rationality of the process of production. Yet this is exactly the inference that Fogel and Engerman draw from the “efficiency” gap between slave and free agriculture. The advantage of slave over free labor, they say, was due to the “special quality of plantation labor.”6 Nowhere do the authors acknowledge that international consumer demand for cotton might have helped create the observed advantage. Instead they assert,
The advantage of plantations, at least that part which has been measured thus far, was due to the combination of the superior management of planters and the superior quality of black [sic] labor. In a certain sense, all, or nearly all, of the advantage is attributable to the high quality of slave labor, for the main thrust of management was directed at improving the quality of labor. How much of the success of the effort was due to the management, and how much to the responsiveness of workers is an imperative question, but its resolution lies beyond the range of current techniques and available data.7
Neither the slaves nor their masters deserve this kind of credit if their apparent “efficiency” is merely an incidental result of extraordinary demand for the crop they produced.
Pointing to the inferior “efficiency” of the North, Fogel and Engerman cast aside the image of the South as a comparatively traditional, noncommercial culture. Southern planters, they say, were “on the whole a highly self-conscious class of entrepreneurs…steeped in the scientific agricultural literature of the day….”8 With decisive aid from black overseers and drivers, slaveowners fashioned the “first large, scientifically managed business enterprises….”9 These work forces were made up of functional gangs and teams that worked at a level of intensity comparable to that of modern assembly lines.10 The authors recognize that such impersonal factors of collective discipline and specialization contributed to the superior “efficiency” of slave plantations, but they lay greatest stress on the effort and ingenuity displayed by the average slave. This moral dimension is what most impresses them. Slaves typically were “diligent and efficient workers”; they were even “imbued like their masters with a Protestant ethic.” The average slave field hand was “harder working and more efficient than his white counterpart.”11
The primary evidence for this portrait of the achievement-oriented slave is the index of “efficiency”—which, as shown above, may say more about the behavior of cotton consumers than about the diligence of producers.
Fogel and Engerman also use the “efficiency” gap as a launching pad for their most spectacular polemic, the attack on the “myth of black incompetence.”12 This regrettable phrase, which might be rendered more accurately as “the myth (if such it is) of slave apathy,” is a catchall rubric under which the authors lump together virtually all previous observers and analysts of slavery. Both Stanley Elkins’s “Sambo” thesis and Kenneth Stampp’s thesis of “day-to-day resistance” are condemned by Fogel and Engerman as mere variations on a myth fabricated, ironically, not by slaveowners but by abolitionists:
What bitter irony it is that the false stereotype of black labor, a stereotype which still plagues blacks today, was fashioned not primarily by the oppressors who strove to keep their chattel wrapped in the chains of bondage, but by the most ardent opponents of slavery, by those who worked most diligently to destroy the chains of bondage. 13
Fogel and Engerman declare that the myth has been perpetuated principally by “racism” and “racist myopia.”14 But there may be a simpler explanation for the myth’s persistence. It may conform to the observed facts—not of black incompetence, certainly, but of slave apathy. It is no racial slur to suppose that self-interest evokes a sense of task and dedication that the whip cannot impose. What Fogel and Engerman seem not to realize is that a considerable degree of slave apathy—indeed, even psychological trauma or a measure of deliberate sabotage—is perfectly compatible with their finding of high “efficiency.” This follows from the inherent limitations of the index. One can grant that slavery was a profitable institution and that the Southern economy was thriving as a whole. One can also grant that Southern agricultural production was “efficient” in the narrow economic sense measured by the index. All three of these conditions are compatible with indifference and even “day-to-day resistance” by slave laborers. If the demand for cotton was sufficiently strong, even the most cumbersome production process might have yielded a net profit for planters, economic growth for the region, and a favorable “efficiency” rating for slave labor.
How intense was the demand for cotton? All the objections I have raised thus far hinge on this question. If the demand for cotton was not significantly stronger than the demand for Northern agricultural products, then Fogel and Engerman are correct to infer a superior quality of Southern labor from the “efficiency” gap (though one would still need to distinguish between the role of individual work habits on the one hand and impersonal factors like collective discipline and economies of scale on the other).
But the reputation of King Cotton is well known, and Fogel and Engerman build upon it. Although they never relate the question of cotton’s profitability to the subject of efficiency, they take great pains to show that the demand for cotton was insatiable in real life as well as legend. Even though the long-term trend in cotton prices was downward throughout the antebellum period
…the 1850s constituted a period of sustained boom in profits for cotton planters. It was an era that outstripped even the fabled prosperity of the 1830s. Nearly every year of the decade was one of above-normal profit. What is more, profits remained high during the last four years of the decade, with prices averaging about 15 percent above their trend values. No wonder cotton production doubled between 1850 and 1860. It was clearly a rational economic response to increase cotton production by over 50 percent between 1857 and 1860. If planters erred it was not in expanding cotton production by too much. Quite the contrary—they were too conservative. Their expansion had not been adequate to bring prices down to their trend values and profits back to normal (equilibrium) levels.15
If supply lagged so far behind demand as this, surely we have a prima facie case for attributing at least part of the “efficiency” gap to the extraordinary demand for cotton. If the index is influenced by demand at all, Fogel and Engerman are obliged to ascertain the influence of the demand for cotton before they can claim to know what caused the “efficiency” gap.
The case for King Cotton as a source of the “efficiency” gap is not limited to the contention that 1860 was an atypical year in which the output of cotton producers was artificially inflated by a transitory lag between demand and supply. There was such a lag and the authors ought to have taken it into account. But more important, what makes the authors’ calculations suspect is the continuing high demand for cotton throughout most of the nineteenth century and the inability of producers outside the South to respond effectively to that growing demand. Even if there had been no lag at all between demand and supply in 1860, there still would be grounds for attributing part of the “efficiency” gap to cotton because the South enjoyed a semimonopoly in the production of the crop. On the eve of the Civil War 80 percent of the cotton consumed by Great Britain’s voracious textile mills came from the South. The South grew all but a third of the world’s total crop.16 No conceivable margin of advantage of slave over free labor could account for such a gargantuan share of the world market. Instead, the South’s dominance in large part must be attributed to the region’s natural advantages of soil and climate.
The South’s competitors in cotton production were notoriously weak by comparison. India’s short-staple cotton accounted for 18 percent of world production, but it was so inferior to “American Upland” that it was seldom used alone, even in the coarsest cloth.17 In spite of repeated British efforts, Egypt’s fine, long-staple cotton did not become a serious competitor until much later because the crop required extensive irrigation and posed special problems of ginning and spinning.18 The South’s natural advantages are demonstrated by its recovery after the Civil War. In spite of extensive devastation that prevented the region from matching its prewar production record until 1878, by the turn of the century the South was again supplying well over half of the world’s cotton—without the supposed benefits of slave labor.19 No wonder a recent econometric study concludes that although the South did not exploit its position by deliberately restricting production, it “did indeed possess substantial monopoly power in world cotton markets….”20
Although the South faced little international competition, there was of course competition among producers within the South. No planter or combination of planters controlled the price of cotton. But it is essential to recognize that this domestic competition did not reach the mode of production. All large planters relied on the same basic mode of production: slave labor. As a result, slavery as a system of cotton production faced little competition either at home or abroad. Fogel and Engerman note the total absence of large free farms in the South and attribute it to the superior efficiency of slave labor.21 But it might just as well be attributed to the shortage of free laborers and the ideological opposition that slaveowners no doubt would have mounted against an alternative mode of production which, if proved successful, would have undermined the value of their investment in slaves. If an entrepreneur wished to produce cotton on a large scale in the ante-bellum South, he probably had to use slave labor, whether or not it was efficient.
Given these semimonopoly conditions of mild competition and strong demand, the Southern labor force did not need to be efficient in order to achieve a high ratio of dollar output to input. Just as in the case of Turkish opium, even primitive production methods might rank high in measured “efficiency” if demand was strong and competition was limited by legal or natural barriers. Under such circumstances producers would have no compelling incentive to adopt new methods because the old ones—strongly sanctioned by cultural or ideological values—would continue to yield adequate profits.
In conclusion, two puzzles in Fogel and Engerman’s presentation can be resolved handily if one adopts the view that cotton demand has something to do with the superior “efficiency” of Southern agriculture. First, the authors fail to ask why, if slave labor was so efficient, did Northern businessmen not make stronger efforts to import it, not only into the disputed territories, but into the North itself? What moral scruple could have persuaded entrepreneurs to forego a 35 percent margin of advantage, in an era when even abolitionists were “racists”? If Southern planters were rational enough to overcome “racist myopia” and acknowledge the efficiency of slave labor, why weren’t Northern businessmen? Perhaps the answer is that although slave labor was potentially mobile, the crop that made it “efficient” was not.
Second, Fogel and Engerman find striking differences of efficiency within the slave South. “The slave plantations of the Old South exceeded the efficiency of free northern farms by 19 percent, while the slave plantations of the newer southern states exceeded the average efficiency of free northern farms by 53 percent.”22 Again the authors fail to ask themselves an obvious question: Why was slave labor so much more “efficient” in the new states—the heart of the cotton belt—than in other parts of the South? Are we to believe that slaves grew stronger and more diligent as they moved from Virginia to Mississippi? Or that their masters gained in entrepreneurial skill under the hot Alabama sun? Occam’s razor would suggest that it was not the special qualities of slave labor per se that made for “efficiency,” but the good conditions for cultivating cotton.
Are Fogel and Engerman correct to believe that the special quality of slave labor made Southern agriculture 35 percent more efficient than Northern agriculture? What has been said here does not settle the question, but only holds it open. I have not tried to prove that slave labor was of low quality, but only to show that the opposite contention is unproved. The question now is one of degree—how much of the observed “efficiency” gap is attributable to the diligence and rationality of producers, how much to the impersonal advantages of large-scale collective discipline, and how much to consumer demand for cotton?23 To settle the question will require the techniques of the cliometricians, and I will gladly defer to their final judgment. As a start, we need to do for cotton what Professor Fogel has already done for the railroad—imagine it out of existence and thereby ascertain the consequences of its presence.
Perhaps this has already been done. Perhaps all the considerations proposed here already have been taken into account and are incorporated into the final conclusions of Time on the Cross. If the authors can show that this is the case, then we can accept their conclusions and find fault only with their unconvincing manner of presentation. Given only the evidence presented in the book, however, a reasonable man must conclude that the “efficiency” gap is probably more the result of extraordinary consumer demand than extraordinary producer performance. But there may be more to this matter than meets the eye. It may be that the emperor is in reality fully clothed, in which case he now needs only to make his full outfit visible to reasonable men.
September 19, 1974
Robert William Fogel and Stanley L.Engerman, Time on the Cross, volume I: The Economics of American Negro Slavery (Little, Brown, 1974), p. 192, reviewed in NYR, May 2, by C. Vann Woodward. My understanding of the issues has been sharpened by friendly correspondence with Professor Engerman and by the assistance of Rice University colleagues too numerous to name. ↩
Evsey D. Domar, “On the Measurement of Technological Change,” The Economic Journal, 71 (December, 1961), pp. 709-729; “On the Measurement of Comparative Efficiency,” in Comparison of Economic Systems: Theoretical and Methodological Approaches, ed., A. Eckstein (University of California, 1971), p. 229; Moses Abramovitz, “Resource and Output Trends in the United States Since 1870,” American Economic Review, Papers and Proceedings 46 (May, 1956), pp 5-23. ↩
Fogel and Engerman, Time on the Cross, volume II: Evidence and Methods: A Supplement, pp. 131, 138, emphasis added. The market value of the nation’s agricultural output was taken from estimates by Marvin W. Towne and Wayne D. Rasmussen, “Farm Gross Product and Gross Investment in the Nineteenth Century,” in Trends in the American Economy in the Nineteenth Century: Studies in Income and Wealth, volume 24 (Princeton, 1960), pp. 255-312. Fogel and Engerman describe their initial procedure for calculating regional output as follows: “The allocation of crops [to the two regions] was based on census data regarding the physical product of each crop. Thus in the case of wheat, for example, the Towne-Rasmussen value of national wheat output in 1860 was $151.0 million. According to the 1860 census, the southern and northern shares of national wheat output were 22.4 and 73.2 percent, respectively. Therefore the value of southern wheat output in 1860 was measured as $33.8 million, while that of the North was $110.5 million” (p. 131). ↩
Time on the Cross, volume I, p. 210. ↩
Some economists have recognized that the index does not discriminate between two different aspects of efficiency, one reflecting producer performance per se, the other reflecting producer responsiveness to consumer demands. Abram Bergson distinguishes between “efficiency in the sense of realization of production possibilities” and efficiency in the sense of “the degree of optimality of the output structure.” He acknowledges that the index measures “performance in the two spheres together.” In short, part of the index’s test of efficiency is whether producers select the optimal mix of outputs, given a certain demand structure. But this supposes that all producers are equally capable of responding to demand. This is a valid assumption in many cases, but it is absurd to suggest that Massachusetts farmers flunked any test of optimality by failing to grow cotton. Edward F. Denison also has recognized the need to take into account the effect of demand pressure upon fluctuations in productivity. See Bergson, “Comparative Productivity and Efficiency in the Soviet Union and the United States,” in Eckstein, op. cit., p. 195, and Denison, What Growth Rates Differ: Postwar Experience in Nine Western Countries (Brookings Institution, 1967), pp. 273-276. In a forthcoming review essay in the Journal of Economic History, Paul A. David and Peter Temin make the same point. ↩
Time on the Cross, volume I, p. 209. ↩
Ibid., p. 210. ↩
Ibid., p. 201. ↩
Ibid., p. 208. ↩
Ibid., pp. 263, 231, 5. ↩
Ibid., p. 223. ↩
Ibid., p. 215. ↩
Ibid., pp. 223, 215. ↩
Ibid., p. 93. ↩
Stuart Bruchey (ed. and comp.), Cotton and the Growth of the American Economy, 1790-1860: Sources and Readings (Harcourt, Brace & World, 1967), tables 1A, 2A, and 2B. ↩
A.W. Silver, Manchester Men and Indian Cotton, 1847-1872 (Manchester University, 1966), pp. 32, 9, 227. ↩
E.R.J. Owen, Cotton and the Egyptian Economy, 1820-1914 (London: Oxford University Press, 1969), pp. 50, 31, 199-202; J.A. Todd, The World’s Cotton Crops (London:A. & C. Black, 1924), pp. 239-249. ↩
Todd, The World’s Cotton Crops, pp. 98, 395. ↩
Gavin Wright, “An Econometric Study of Cotton Production and Trade, 1830-1860,” The Review of Economics and Statistics, 53 (May, 1971), p. 111. I do not claim that planters conspired to control prices. I claim only that the South enjoyed such competitive leeway that it might have realized a profit even if its labor force was neither diligent nor well managed. ↩
Time on the Cross, volume I, p. 194. ↩
Ibid., p. 196. ↩
This is not to say that cotton demand is the only point on which Fogel and Engerman’s efficiency calculation can be challenged. The “efficiency” gap also rests on their dubious assumption that the quality of Northern farmland was 2.5 times higher per acre than Southern farmland. Nor do they consider the possibility that slavery made its chief contribution to “efficiency” by repelling free yeoman farmers, thus reducing land prices and the level of capital investment in Southern agriculture. See also the extensive critique of their labor and land indices by David and Temin in a forthcoming issue of the Journal of Economic History. If the calculations of David and Temin are correct, the “efficiency” gap may not only be reduced, but reversed. ↩