An earlier draft of this paper was presented to the Eleventh Annual Cliometrics Conference, the University of Wisconsin, Madison, Wisconsin, April 29, 1971. We are grateful to a number of those in attendance for their comments and especially to Paul David, Albert Fishlow and Joseph Swanson. As is so often the case, Stanley Engerman offered many valuable suggestions, as did Robert Gallman and an anonymous reviewer. The Social Science Research Institute, the University of Hawaii, provided clerical services, and Randall Weir provided able research assistance. Any shortcomings which remain, however, are our own responsibility.

¹ Taylor, George R., The Transportation Revolution, 1815–1860 (New York: Holt, Rinehart, and Winston, 1962).Google Scholar

² External developments, in ocean transportation, were also significant. According to Douglass C. North, “… the most striking fall of ocean freight rates occurred between 1815 and 1850 …”; North, Douglass C., Growth and Welfare in the American Past (Englewood Cliffs, N.J.: Prentice-Hall, Inc., 1966), p. 110Google Scholar; and productivity advance in ocean shipping, 1814–1860, was 3.5 percent annually; North, Douglass C., “Sources of Productivity Change in Ocean Shipping, 1600–1850,” The Journal of Political Economy, LXXVI (Sept./Oct. 1968), pp. 953–70.CrossRefGoogle Scholar

³ For examples of recent research on these subjects, see Scheiber, Harry S., The Ohio Canal Era (Columbus: Ohio University Press, 1969)Google Scholar; Ransom, Roger L., “Interregional Canals and Economic Specialization in the Ante-Bellum United States,” Explorations in Entrepreneurial History, 2nd series, V (Fall 1968), 12–35Google Scholar; Ransom, Roger L., “Social Returns From Public Transport Investment: A Case Study of the Ohio Canal,” The Journal of Political Economy, LXXVIII (Sept./Oct., 1970), pp. 1041–60CrossRefGoogle Scholar; Fogel, Robert W., Railroads and American Economic Growth: Essays in Econometric History (Baltimore: The Johns Hopkins Press, 1964)Google Scholar; Fishlow, Albert, Railroads and the Transformation of the Ante-Bellum Economy (Cambridge, Mass.: Harvard University Press, 1965)Google Scholar; and Boyd, J. Hayden and Walton, Gary M., “The Social Savings from Nineteenth Century Railroad Passenger Services,” Explorations in Economic History, IX (Spring 1972) 3, pp. 233–54.Google Scholar

⁴ The study-by Hunter, Louis C., Steamboats on the Western Rivers (Cambridge: Harvard University Press, 1949)Google Scholar, is an excellent history of steamboating, and it has proved invaluable to this study, but it contains little economic analysis. An important beginning towards some economic analysis of river transportation has been made by Erik F. Haites, “Ohio and Mississippi River Transportation, 1810–1860” (unpublished doctoral dissertation, Purdue University, 1969); by Haites, Erik F. and Mak, James, “Ohio and Mississippi River Transportation, 1810–1860,” Explorations in Economic History, VIII (Winter 1970), 153–80Google Scholar; and by Haites, Erik F. and Mak, James, “Steamboating on the Mississippi Before the Civil War: A Study of a Competitive Industry,” Business History Review, XLV (Spring 1971), 52–78.Google Scholar

⁵ North, Growth and Welfare in the American Past, p 111.

⁶ A study of productivity change in flatboating is in progress by Mak and Walton. Initial, but preliminary, results suggest substantial rates of change, 1830–1850, and averaging about 1 percent 1815–1860.

⁷ It should be noted that ideally, where the data are perfectly consistent, both measures should give the same findings. However, the data do contain imperfections; thus we have included both measures in the analysis. For instance, the indexes of inputs and outputs used in measure (1) are available only by decade. In addition, government expenditures on river improvements between Louisville and New Orleans (the route analyzed below; see text), though of modest size, did aid navigation, reduce accidents, and increase the output of steamboats. The resources used for these improvements should be included in our input index, but an appropriate way to include them has eluded us. Therefore, the input index is understated absolutely, although how this affects its rate of change is somewhat ambiguous. “The total amount devoted to the Mississippi, Missouri, Ohio, and Arkansas from 1824 to 1860 was about $3,130,000,” Lippincott, Isaac, “A History of River Improvement,” The Journal of Political Economy, XXII (1914), 648.Google Scholar In Lippincott's footnote no. 4 (p. 648), he indicates that about half of this sum was allocated to the Mississippi, but a portion of this amount would have been spent on the improvements of areas of the river other than the Louisville-New Orleans trunk section. Evidence from Haites (Table 19) on expenditures for river improvements on the Ohio and Mississippi, 1820–60, shows that the overwhelming portion of expenditures came during 1825–38, with little or no expenditures occurring in 1815–24, and only a relatively small amount being spent in the period 1839–60. This rise and fall of expenditures suggests that our measure (1) is possibly somewhat upward biased through the first half of the period and slightly downward biased over the last half. However, no discernable bias is suggested for the entire period. It should be noted that the impact of the capital expenditures, the useful life of snagboats and other equipment and improvements, would have been over a longer period and more evenly spread than that suggested by the dating of expenditures. For further information on river improvements see Hunter, Steamboats …, ch. iv.

⁸ A single route is used to isolate the effects of improvements on one route, rather than to consider changes between routes. Hopefully, this illustrates secular changes and excludes or minimizes cross-sectional influences. Data availability urged the use of the Louisville-New Orleans route, and it was the main traffic artery in the trans-Appalachian West. Because our findings are derived from a single route, certain qualifications are in order. Though the absolute level of freight rates per ton mile was higher (and the level of productivity lower) on back waterways and western rivers other than the Louisville-New Orleans trunk route, there is little reason to suspect that the rates of change of productivity (and rates of decline of freight rates) were noticeably different on these other routes. The average rates of advance, 1815–60, estimated in our study are probably indicative of the general rates of advance that occurred on most western rivers in the antebellum period. Of course, certain particular findings (given below) for our Louisville-New Orleans route, such as the average ship size by decade, the average number of round trips per year per vessel by decade, and other aspects unique to this route would not be applicable to other routes. Nevertheless, the average rates of change of the cost determinants discussed below were probably similar on most other major routes. We have no reason to suspect that the relative importance of the major improvements was radically different on other western routes.

⁹ Despite initial attempts by Fulton to monopolize steamboat services on the Mississippi, the industry was highly competitive throughout the entire antebellum period. Evidence of the degree of competition is given by Haites and Mak, “Ohio and Mississippi River Transportation, 1810–1860,” pp. 172–74 and pp. 177–79. See also Hunter, Steamboats …, pp. 32–33 and pp. 308–15.

¹⁰ In addition to the biases discussed above (see fn. 7), the selection of base year weights inevitably adds other biases to the measures. In this case, however, these biases appear quite insignificant. Our choice of 1840–49 base year weights is primarily because this was the earliest period for which the data were most complete, and hopefully, it is near enough to the middle of the time period not to impose a large bias. To illustrate the sensitivity of the selection of weights, moving weights slightly changes the above rates of 5.5 percent and 4.6 percent to 5.8 percent, and 4.3 percent, respectively; 1815–1819 base year weights give rates of 5.8 percent and 4.2 percent, respectively.

¹¹ The rate of advance of total factor productivity in railroads, 1839–1910, was approximately 3.5 percent (5.4 percent, 1839–59), Fishlow, Albert, “Productivity and Technological Change in the Railroad Sector, 1840–1910,” Output, Employment and Productivity in the United States after 1800, Studies in Income and Wealth, Vol. XXX (New York: National Bureau of Economic Research, 1966), pp. 583–646.Google Scholar The rate of change in ocean shipping, 1814–1860, was 3.5 percent according to North, but his comparison of deflated rates from periods of war to peace probably overstates the true productivity increase because ocean rates were unusually sensitive to international conflicts; North, “Sources of Productivity Change in Ocean Shipping, 1600–1850,” p. 954. Also, the rate of fall of costs of ocean shipping in the last half of the nineteenth century when steam replaced sail was less than over the period 1814–60. We are unaware of any productivity studies for canals during the nineteenth century, but it is doubtful that canal productivity increased significantly and it may have declined. The significance of productivity increases (or freight rate reductions) is dependent, of course, on the volume shipped. The relative importance is determined by a comparison of the total cost savings, or more specifically, the social savings in percentage terms. An estimate of the social savings of steamboats is beyond the scope of this study which prevents such a comparison.

¹² These rates are given in the source to Table 1.

¹³ The upstream rate was lower than the downstream rate in 1850–60 because of the decline in utilization upstream, 1840–60.

¹⁴ The adjustments to real rates are made from the Warren-Pearson general price index, U.S. Bureau of the Census, Historical Statistics of the United States from Colonial Times to 1957 (Washington, D.C.: Government Printing Office, 1960), p. 115.Google Scholar

¹⁵ We qualify this conclusion to the extent that the productivity gains from steam power, which are reflected in the fall in rates, 1815–20, may have been understated somewhat because of slow entry or limited competition in this early period. Nevertheless, there were 17 vessels in operation on western rivers in 1817, and 69 were in operation by 1820 (Hunter, Steamboats …, p. 33; also see fn. 9 above). It seems reasonable to assume that the initial impact of the steam engine had occurred by 1820. In any case, although an alteration of dates would change the relative importance of the steam engine vis a vis later improvements, it can hardly be disputed that later improvements were of major significance.

¹⁶ It is tempting, perhaps, to conclude from Table 1 that the sources of productivity advance were more labor-saving than capital-saving. Clearly, we do observe that an average steamboat in 1860 used more capital relative to labor than did its 1815 counterpart. However, changes in the relative prices of capital and labor affected the observed factor proportions as well as did technological change, or other sources of productivity advance. The average wages of steamboatmen rose 50–60 percent over the period while average capital costs per measured ton fell by almost half. This change in relative prices probably led to some movement along the production function, and a portion of the rise in labor productivity is probably due to the substitution of capital for labor. The shift of the production function led to a fall in capital inputs per payload ton despite the relative price decline of capital. Therefore, capital inputs per payload ton clearly were reduced. Labor inputs per payload ton probably also were saved, but we cannot be certain without knowing the exact position of the production function isoquants before and after the productivity advance.

¹⁷ Hunter, Steamboats …, p. 66.

¹⁸ Ibid., p. 75.

¹⁹ Ibid., p. 75.

²⁰ Ibid., pp. 79–80.

²¹ Insurance coverage was typically for a specific time period rather than on a per voyage basis. Coverage on steamboats was generally taken out by the month or the year, with three to six months as the most frequently mentioned periods (Hunter, Steamboats …, p. 367). However, the insurance rates would reveal the rate of steamboat losses, which would have been a direct function of the running time or the number of trips taken each year.

²² Except for the captain and first mate (who were hired on an annual basis) crews were paid monthly and were typically hired for the year's navigation season. See Haites and Mak, “Ohio and Mississippi River Transportation, 1810–1860,” p. 158.

²³ Depreciation and repairs may have varied with the number of trips per year, but the evidence, though somewhat inconclusive, suggests not. See Haites and Mak, “Ohio and Mississippi River Transportation, 1810–1860,” p. 160.

²⁴ The term “in operation” refers to that period in which the vessel was not out of water, and this remained constant at approximately 9 months per year. It should be emphasized that the estimates of passage times and number of round trips per year are derived from separate and independent sources.

²⁵ Kirkland, Edward C., A History of American Economic Life (4th ed.; New York: Appleton-Century-Crofts, 1969), p. 142Google Scholar. For another similar example, see Kent T. Healy, “American Transportation Before the War Between the States,” in Williamson, Harold I. (ed.) The Growth of the American Economy (New York: Prentice-Hall, Inc. 1949), pp. 172–88.Google Scholar See p. 180, in particular.

²⁶ See Hunter, Steamboats …, ch. iii, for descriptions of the mechanical improvements in steamboating.

²⁷ Hunter, Steamboats …, pp. 222–23. The six months navigation season before 1830 meant that 3 months of the period that the vessel was in operation (see fn. 24) was necessarily port time. The length of the navigation season, of course, varied according to river conditions and ship size, and the above estimates are rough averages.

²⁸ Although the increase in the navigation season appears to have been more important than the reduction in cargo collection times or passage times, it was only capital-saving, not labor-saving. Labor costs rose almost in proportion to the increase in the navigation season (only the payments to the first mate and captain did not vary with the navigation season; see fn. 22). The increase in the number of round trips per year caused by the decline in passage times and cargo collection times (which were approximately of equal importance), however, were both labor-saving and capital-saving, and the relative importance of the longer navigation season suggested in Table 3 is overstated in terms of the combined savings of labor and capital. When both input savings are considered, it appears that the fall in passage times was about equal in importance to the reduction of cargo collection times, and the increase in the navigation season was slightly less important.

²⁹ It should be noted that although changes in the physical characteristics came gradually, most of them occurred fairly early, and “by 1840 the western steamboat had reached the form in essentials it was to keep to the end of its day” (Hunter, Steamboats …, p. 64).

³⁰ It is important to emphasize that the period of technical diffusion, or the time between the development and adoption of new improvements was apparently quite short. The average stock of tonnage was completely renewed every 5 to 6 years, the expected life of an average steamboat. This relatively short average life changed little over the period; therefore, the advance of average productivity (the productivity of an average vessel) was about as rapid as the discovery and development of new improved physical characteristics. New improvements were rapidly introduced into the stock of tonnage.

³¹ Hunter, Steamboats …, p. 121.