Article contents
A Quantitative Approach to the Study of Railroads in American Economic Growth: A Report of Some Preliminary Findings*
Published online by Cambridge University Press: 03 February 2011
Extract
Leland Jenks's article describing the pervasive impact of the railroad on the American economy first as an idea, then as a construction enterprise, and finally as a purveyor of cheap transportation, has become a classic of economic history. The particular contribution of the Jenks article was not the novelty of its viewpoint, but the neat way in which it summarized the conclusions both of those who lived during the “railroad revolution” and those who later analyzed it through the lens of elapsed time. Out of this summary the railroad emerges as the most important innovation of the last two thirds of the nineteenth century. It appears as the sine qua non of American economic growth, the prime force behind the westward movement of agriculture, the rise of the corporation, the rapid growth of modern manufacturing industry, the regional location of industry, the pattern of urbanization, and die structure of interregional trade.
- Type
- Articles
- Information
- Copyright
- Copyright © The Economic History Association 1962
References
1 Jenks, Leland H., “Railroads as an Economic Force in American Development,” The Journal of Economic History, IV, No. 1 (May 1944), 1–20CrossRefGoogle Scholar; reprinted in Lane, F. C. and Riemersma, J. C., Enterprise and Secular Change (Homewood, Ill.: Richard D. Irwin, 1953), pp. 161–80Google Scholar; and in Lambie, J. T. and Clemence, R. V., Economic Change in America (Harrisburg, Pa.: Stackpole Co., 1954), pp. 52–68Google Scholar.
2 Savage, Christopher I., An Economic History of Transport (London: Hutchinson & Co., 1959). p. 184Google Scholar.
3 Rostow, W. W., The Stages of Economic Growth (Cambridge: The University Press, 1960), p. 55Google Scholar.
4 See, for example, Belcher, Wyatt W., The Economic Rivalry Between St. Louis and Chicago, 1850–1880 (New York: Columbia Univ. Press, 1947)Google Scholar, and “Cincinnati Southern Railway: Struggle Between Two Rival Cities for Metropolitan Dominance,” in Gras, N. S. B. and Larson, Henrietta M., Casebook, in Business History (New York: F. S. Crofts, 1939)Google Scholar.
5 A table of the canals abandoned during the nineteenth century is given in U. S. Congress, Senate, Preliminary Report of the Inland Waterways Commission, Doc. No. 325, 60th Congress, 1st Sess. (1908), pp. 205–9.
6 There was double counting in the data on which these ratios are based. U. S. Congress, Senate, Andrews Report, Executive Doc. No. 112, 32nd Cong., 1st Sess. (1853), pp. 903–6;; U. S. Census Bureau, Eleventh Census of the United States: 1890, Report on the Transportation’ Business in the United States, Part I, pp.452, 548, 640; Part II, pp. 9, 10, 163, 308, 384, 436, 479.
7 Isard, Walter, “Transport Development and Building Cycles,” Quarterly Journal of Economics, LVII, No. 1 (Nov. 1942), 90–112CrossRefGoogle Scholar; Jenks, “Railroads as an Economic Force,” pp. 4–5.
8 U. S. Census Bureau, Historical Statistics of the United States, Colonial Times to 1957 (Washington: Govt. Printing Office, 1960), p. 427.
9 U. S. Agriculture Dept., Yearbook, of Agriculturem, 1933, p. 748Google Scholar; Taussig, Frank W., “The Tariff, 1830–1860,” Quarterly Journal of Economics, II (1888), 379Google Scholar.
10 The relationship between the railroads and the growth of manufacturing industries is the subject of one of the other essays in Railroads and American Economic Growth: Essays in Econometric History.
11 New England is the only notable departure from this pattern of complete, or virtually complete, agricultural self-sufficiency. Even here the deficiency in grain appears to have been relatively small. Thomas Jefferson, writing in 1808, estimated that “90,000 persons in Massachusetts subsisted on imported flour.” This implies that, although deficient, the state's output of wheat was large enough to meet the needs of 80 per cent of the population. And the Massachusetts deficit was offset, at least in part, by the surpluses of Vermont and New Hampshire. Bidwell, Percy W. and Falconer, John I., History of Agriculture in the Northern United States, 1620–1860 (Washington: Carnegie Institution of Washington, 1925), p. 236Google Scholar; U. S. Census Bureau, Historical Statistics, p. 13.
12 U. S. Census Bureau, Eleventh Census of the United States, Compendium, Part I, p. 2.
13 Estimated local requirements and supplies for the North Atlantic region (in thousands of tons) are:
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20151019113355099-0892:S0022050700062719_eqnU1.gif?pub-status=live)
The procedure followed in die construction of these estimates is discussed below, pp. 179–85.
14 Wheat exports from ports in the North Atlantic region were approximately 1,260,000 tons. St. Louis Merchants’ Exchange, Annual Report, 1890, p. 168.
15 U. S. Congress, House, Report of the Industrial Commission on the Distribution of Farm Products, Doc. No. 494, 56th Cong., 2nd Sess. (1901), p. 37.
16 U. S. Statistics Bureau, Wheat Crops of the United States, 1886–1906, Bulletin No. 57 (1907), p. 18.
17 See below, pp. 178–79.
18 Ibid., pp. 179–84.
19 A progress report on the essay dealing with the impact of the railroad on the intraregional distribution of agricultural commodities (“The Social Saving Attributable to American Railroads in the Intraregional Distribution of Agricultural Products in 1890”) was presented at the New York meeting of the Regional Science Association in December 1961.
20 The definition of social saving used in this paper is the difference between the actual level of national income in 1890 and the level of national income that would have prevailed if the economy had made the most efficient possible transport adjustment to the absence of the interregional railroad. As noted in the text, this figure is larger than the more ideal social saving figure, which would take into account the production adjustments that would obtain with a different system of transportation.
In treating the differential in transportation costs as a differential in levels of national income, I am assuming that there would have been no obstacles to an adjustment to a nonrail situation. In other words, I am abstracting from market problems by assuming that national income would have dropped only because it took more productive resources to provide a given amount of transportation, and that all other productive resources would have remained fully employed. The relationship between the railroad and the demand for output is the subject of one of the other essays in my study (cf. note 10).
21 See, for example, Ringwalt, J. L., Development of Transportation Systems in the United States (Philadelphia: the author, 1888)Google Scholar; Lucas, Walter A., ed., 100 Years of Steam Locomotives (New York: Simmons-Boardman, 1958)Google Scholar; Clarke, Thomas C., et al., The American Railway: Its Construction, Development, Management and Appliances (New York: Scribner & Sons, 1892)Google Scholar; Baldwin Locomotive Works, History of the Baldwin Locomotive Worlds, 1831 to 1907 (Philadelphia: Edgell Co., 1907).
22 U. S. Statistics Bureau, Wheat Crops of the U. S., p. 7; U. S. Census Bureau, Historical Statistics, pp. 12, 13, 139; U. S. Statistics Bureau, Exports of Farm Products from the United States, 1851–1908, Bulletin No. 75 (1910), pp. 44, 46.
23 U. S. Statistics Bureau, Changes in the Rates of Charge of Railway and Other Transportation Services, Bulletin No. 15, rev. (1901), p. 14; U. S. Congress, House, Report of the Industrial Commission on Agriculture and Agricultural Labor, Doc. 179, 57th Cong., 1st Sess. (1901), X, 690–91.
24 See below, pp. 179–84, and Table 6.
25 U. S. Census Bureau, Historical Statistics, p. 139.
26 Gallman, Robert E., “Commodity Output in the United States,” Trends in the American Economy in the Nineteenth Century. Studies in Income and Wealth of the National Bureau of Economic Research, Vol. 24 (Princeton: Princeton Univ. Press, 1960), p. 47Google Scholar.
27 This statement is based on the assumption of the relative constancy of the saving and capital-output ratios over the range of national income being considered here.
28 U. S. Congress, House, Distribution of Farm Products, p. 142; below, p. 187.
29 The average wholesale price in Chicago of a bushel of wheat during 1890 was eightyseven cents. U. S. Census Bureau, Historical Statistics, p. 123.
30 Gallman, “Commodity Output,” pp. 46–48.
31 Ibid.
32 U. S. Congress, House, Monthly Summary of Commerce and Finance, Doc. 15, Part 7, 56th Cong., 1st Sess. (1900), pp. 2545–2636.
33 U. S. Congress, House, Distribution of Farm Products, p. 174.
34 Gallman, “Commodity Output,” pp. 46–48.
35 This estimate is based on data for 1899. U. S. Census Bureau, Twelfth Census of the United States, Agriculture, Part I, pp. clxxxii–clxxxiii.
It should be remembered that the East did not run the heavy deficits on dairy products that it did on grains and meats. New York and Pennsylvania were two of the three largest producers of dairy products. U. S. Congress, House, Distribution of Farm Products, pp. 268–69.
36 Cf. Robert W. Fogel, The Union Pacific Railroad: A Case in Premature Enterprise (Baltimore: The Johns Hopkins Press, 1960), especially pp. 81–85.
37 Preliminary calculations suggest that the Erie Canal was the only waterway on which a bottleneck might have arisen. In 1890 the Erie carried 3,200,000 tons. Capacity at the time was 10,000,000 tons. It is possible that without the railroad, the agricultural products shifted to the canal would have exceeded 7,000,000 tons, thus taxing the capacity of the Erie. On the other hand, it appears that the most the additional tonnage could have been was 10,000,000 tons. However, 13,000,000 tons was well below the capacity of the New York State Barge Canal. If the linear programming solution leads to shipments which exceed the capacity of the Erie, I will apply the cost figures of the Barge Canal, adequately adjusted for differences in the price level, etc., and find a new solution. The use of Barge Canal rates would further buttress the assumption of constant or declining marginal costs in water transportation.
38 As already noted, the estimate of the social saving in the interregional transportation of wheat, corn, pork, and beef requires a pair of linear programming models for each commodity. It might be thought that more than one pair of programs is required for each commodity. Wheat, for example, was carried east both as wheat and as flour. Pork was transported in an even wider variety of forms, including live swine, dressed pork, canned pork, mess pork, ham, bacon, etc. To die extent that each form of pork or wheat is considered a different product, one might be inclined to argue that a separate pair of models is required for each form. This costly complication can be avoided if (in the case, say, of wheat) the ratio of wheat to flour demanded in each secondary market was roughly the same, and with information on the quantities of wheat and flour shipped from each primary market. For then, knowing the wheat equivalent of a given quantity of flour as well as the cost of shipping each form, there is obviously some transportation cost at which X ton-miles of flour can be converted into Y ton-miles of wheat. The assumption of a constancy in the ratio of wheat to flour shipments tends to introduce an upward bias in die estimate of social saving. If the relative transportation costs of flour and wheat were the same by both forms of transportation, no bias would be introduced. However, the cost of shipping flour relative to the cost of shipping wheat was greater by water than by rail. Hence, in die absence of the railroad more wheat and less flour would have been shipped. However, the conversion of flour into a grain equivalent is based on the proportions of each that were actually shipped in 1890.
39 The pattern of trade is summarized in Johnson, Emory R. et al. , History of Foreign and Domestic Commerce of the United States (Washington: Carnegie Institution of Washington, 1915)Google Scholar, Parts I and II.
40 Table 1 includes only grain which was unloaded in the primary markets. Additional amounts were shipped through die primary markets without unloading. Obviously, these neglected amounts will eventually have to be included.
41 U. S. Foreign and Domestic Commerce Bureau, Atlas of Wholesale Grocery Territories, Domestic Commerce Series, No. 7 (1927).
42 U. S. Foreign and Domestic Commerce Bureau, Atlas of Wholesale Grocery Trading Areas, Market Research Series, No. 19 (1938).
43 U. S. Agriculture Dept., Yearbook, of Agriculture, 1923, p. 1140; U. S. Statistics Bureau, Wheat Crops of the U. S., p. 18.
44 Great Britain, Board of Trade, Cost of Living in American Towns (London: H. M. Stationery Office, 1911); reprinted in U. S. Congress, Senate, Doc. 22, 62nd Cong., 1st Sess. [ser. no. 6082].Google Scholar
The data needed to convert statistics on bread, cake, and macaroni consumption into a wheat equivalent were obtained from U. S. Experiment Stations Office, Bulletin, Nos. 35 (1896) and 156 (1905); cf. U. S. Agriculture Dept., Farmers Bulletin, Nos. 23, 1450; Panschar, William G. and Slater, Charles C., Baling in America (2 vols.; Evanston: Northwestern University Press, 1956).Google Scholar
A convenient review of various budget and diet studies is contained in Faith M. Williams and Carle C. Zimmerman, Studies of Family Living in the United States and Other Countries, U. S. Agriculture Dept., Miscellaneous Publications No. 223 (1935).
45 W. C. Funk, Value to Farm Families of Food, Fuel, and Use of House, U. S. Agriculture Dept., Bulletin No. 410 (1916), pp. 5, 18. Funk's figures were in consumption per equivalent adult, with children twelve and under counted as one half of an adult. Funk's data were converted to a per capita basis on the assumption that the average proportion of persons twelve and under in all the families in his sample residing in a particular state was the same as that state's proportion of persons twelve and under in the rural population in 1910.
The finding that average wheat consumption in the South was about as large as in the North requires some explanation. The letters, journals, and diaries of noted travellers usually stressed the importance of corn in the southern diet. There is no necessary contradiction between the data culled from the budget studies and the commentaries of distinguished observers. Corn was the dominant breadstuff in the southern diet. During the period in question, southerners probably consumed an annual average of about six bushels per capita; the rest of the nation used about a bushel per person. One can easily see why travellers would stress the unique element of the southern diet, while passing over the fact that wheat was also consumed in sizeable quantities. Historians have inferred that since the quantity of corn used was unusually large, the consumption of wheat must have been quite small. The budget studies suggest another interpretation: while some corn was substituted for wheat, even larger quantities were substituted for other commodities, especially dairy products. Average caloric intake in the South also appears to have been greater than in the North [Hawley, Edith, Economics of Food Consumption (New York: McGraw-Hill Book Co., 1932), p. 75Google Scholar].
A high rate of wheat consumption may have been characteristic of the South for the whole last half of the nineteenth century. Schmidt, in his series of studies on the grain trade, noted that the South imported an average of 10,000,000 bushels of wheat per year during the decade leading up to the Civil War. He estimated consumption of wheat for the year i860 at 4.5 bushels per capita in the South, and placed the national at 5.5 bushels [Schmidt, Louis B., “The Internal Grain Trade of the United States, 1850–1860.” Iowa Journal of History and Politics, XVIII, No. 1 (Jan. 1920), 101Google Scholar, 106].
46 Baltimore Corn and Flour Exchange, Annual Report, 1889, 1890.
47 Baltimore Corn and Flour Exchange, Annual Report, 1890–1894; U. S. Census Bureau, Eleventh Census of the United States, Population in the United States, Part I, Tables 8 and 89; cf. note 45, above.
48 These estimates are based on data for the years 1910–1914. The relevant figures are: horses, 27.811 bushels; hogs, 16.568; dairy cattle, 5.112; other cattle, 2.460; sheep, 0.413; poultry, 0.671. U. S. Agricultural Research Service, Consumption of Feed by Livestock, 1909–1956, Production Research Report No. 21 (Nov. 1958), pp. 28–31, 80.
49 U. S. Agriculture Dept., Consumption of Food in the United States, 1909–1952, Agricultural Handbook No. 52 (1953); U. S. Labor Bureau, Eighteenth Annual Report of the Commissioner of Labor, Cost of Living and Retail Food Prices (1904). Frederick Strauss and Louis H. Bean, Gross Farm Income and Indices of Farm Production and Prices in the United States, 7, U. S. Agriculture Dept., Technical Bulletin No. 703 (1940).
50 Funk, Value to Farm Families of Food, pp. 5, 20; U. S. Agriculture Dept., Livestock, and Meat Statistics, Statistical Bulletin No. 230 (July 1958), pp. 283, 284; U. S. Agriculture Dept., Consumption of Food in the U. S., p. 197. Two adjustments were made to the Agriculture Department data to make them comparable to Funk's: the per capita estimates were transformed to equivalent adult estimates on the assumption that the proportion of persons twelve years old and under was the same in 1913 as in 1914 (cf. note 45); edible offals were added to the Agriculture Department figures on beef and pork, since farm families generally consumed all parts of the animals they slaughtered [Zimmerman, Carle C., Consumption and Standards of Living (New York: D. Van Nostrand, 1936), pp. 81–82Google Scholar].
51 That actual rates declined between 1890 and 1910 is suggested by the fact that average freight revenue per ton-mile (adjusted for changes in the price level) declined by over a third (U. S. Census Bureau, Historical Statistics, p. 43). Of course, the decline in average revenue could have taken place even though actual rates were rising if there had been major changes in the composition of freight traffic. However, the available data suggest that the composition and pattern of traffic remained relatively stable during this period. Cf. data in U. S. Census Bureau, Eleventh Census of the United States: 1890, Transportation, Part I, and Interstate Commerce Commission, Twenty-Fourth Annual Report on the Statistics of Railways in the United States for the Year Ending June 30, 1911 (Washington: Govt. Printing Office, 1913).
52 See, for example, Jenks, “Railroads as an Economic Force,” pp. 12–13; and Moulton, Harold G., Waterways versus Railways (Boston: Houghton Mifflin, 1912), pp. 12–13Google Scholar, 33–38.
53 U. S. Congress, House, Distribution of Farm Products, p. 142.
54 See, for example, data in U. S. Statistics Bureau, Changes in Rates; U. S. Congress, Senate, Preliminary Report of the Inland Waterways Commission; and Louisville and Nashville Railroad, Southwestern Freight Tariff, No. 9 (Nov. 16, 1890).
55 The lake-and-canal rate differs from the all-water rate cited in the previous paragraph by 0.047 c e n t s because the former includes transshipment and insurance charges. The lake-andrail rate includes insurance but not transshipment costs, since the ex-lake rail rates included transshipping charges. Cf. notes to Table 4.
56 The water rate is the highest that prevailed during the 1890 season of navigation. U. S. Congress, Senate, Preliminary Report of the Inland Waterways Commission, p. 343. The rail rate is taken from Louisville and Nashville Railroad, Southwestern Freight Tariff, No. 10.
57 See above, Table 2.
58 A breakdown of this figure is given in Table 6.
59 The averages are simple arithmetic means.
Water distances between the points in the sample are an average of 70 per cent longer than rail distances. This suggests a somewhat greater degree of circuity in water transportation than was indicated by the study of the Bureau of Railway Economics, An Economic Survey of Inland Waterway Transportation in the United States, Special Series, No. 57 (Washington: Bureau of Railway Economics, 1930).
60 The adjustment was made in the following manner:
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20151019113355099-0892:S0022050700062719_eqnU2.gif?pub-status=live)
61 The cost of transshipping meat products appears to have been included in the water rate. U. S. Corporations Bureau, Report of the Commissioner of Corporations on Transportation by Water in the United States. Part III, Water Terminals (Washington: Govt. Printing Office, 1910), pp. 329–34.Google Scholar
62 The assumption that boats took an average of a month longer than trains to provide the same transportation service introduces an upward bias into the estimate of the cost of time. The minimum time required by express freight trains on the run from New York to Chicago in 1896 was seventy-five hours, indicating an average speed of twelve miles per hour (Joint Traffic Association, Proceedings of the Board of Managers, 1896, p. 627). About the same time, boats on the Great Lakes made the round trip from Duluth to Buffalo in nine days, indicating an average speed of over nine miles per hour [Ransdell, Joseph E., “Legislative Program Congress Should Adopt for Improvement of American Waterways,” Annals of the American Academy of Political and Social Science, XXXI, No. 1 (Jan. 1908), 38Google Scholar]. In 1912 the average speed of freight boats on rivers was about seven miles per hour. (U. S. Agriculture Dept., Bulletin No. 74, p. 36.) These facts suggest that the average time advantage of railroads in the interregional transportation of agricultural products was less than a week. Some observers argued that in the transportation of bulk items boats actually provided quicker service than trains (Ransdell, “Improvement of American Waterways,” p. 38).
63 To the Chicago values shown in Table 6, $2.83 per ton was added for wheat, $2.59 for corn, $4.00 for cattle and $5.00 for hogs. Dressed meats in Table 6 are quoted at the New York prices, so no further adjustment was necessary. U. S. Congress, Senate, Aldrich Report, I, 518–19, 526.
64 The cost of elevating and storing grain in Buffalo from November 10 to the opening of navigation (about five months) was two cents per bushel or .4 cents per bushel per month (Buffalo Merchants Exchange, Annual Report, 1890, pp. 88–89). At this rate, storage charges on the six months of additional inventory of wheat and corn would have amounted to $5,300,000. In 1910 cold storage rates were $4.96 per ton per month on beef and $4.70 on pork (U. S. Agriculture Dept., Statistical Bulletin No. 493, p. 44). At these rates, the additional storage charges on beef and pork would have been $31,000,000 in dollars of 1910 or $25,000,000 in dollars of 1890. However, cold storage would have been the most costly way of maintaining the additional inventory of meat. It would have been cheaper to store meat in the East by sending live animals to eastern feeders. In this case, the cost of storage would have been essentially the cost of shipping more feed but smaller animals.
65 U. S. Congress, Senate, Preliminary Report of the Inland Waterways Commission, pp. 180–81, 193–97. 202–3, 205–9; cf. U. S. Census Bureau, Transportation by Water, pp. 44–46.
66 The preceding calculation may be summarized as follows:
First approximation of social saving
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20151019113355099-0892:S0022050700062719_eqnU3.gif?pub-status=live)
67 How significant is a social saving of 1 per cent of national income? This question cannot be answered without making further assumptions as to how the economy would have adjusted to the absence of the railroad. One consequence of the absence of the railroad would have been a rise in the seaboard prices of agricultural commodities. This could have had a significant effect on the U. S. balance of trade. Similarly, a shift from railroad to water and wagon transportation may have been a shift from a more to a less capital-intensive activity. Such a change might have aggravated market problems in the capital goods sector. On the other hand, it might have increased the demand for labor.
If one abstracts from these essentially Keynesian issues of insufficient demand and focuses on the economy's production possibilities, it is possible to interpret the social saving in a fairly simple way. Assuming that the marginal aggregate savings and capital-output ratios would have been what in fact they were when national income was 99 per cent of the 1890 level, the absence of the interregional railroad would have retarded the development of the economy by about three months.
- 92
- Cited by