Article contents
Railroads, American Growth, and the New Economic History: A Critique
Published online by Cambridge University Press: 03 February 2011
Extract
Robert Fogel and Albert Fishlow have attempted in recent studies to measure the net contribution of the railroad to American economic growth in two specific years: 1890 and 1859. Both their methods and conclusions have confused, if not dismayed, more conventional historians. This article has two objectives: The first is to provide a simplified guide to the analytical framework underlying their calculations and the second is to suggest to both old and new economic historians that the measurements employed by Fogel and Fishlow bear no direct relationship to what they claim to measure—the benefits to the nineteenth-century economy from the existence of the railroad.
- Type
- Articles
- Information
- Copyright
- Copyright © The Economic History Association 1968
References
1 Fogel, Robert William, Railroads and American Economic Growth: Essays in Econometric History (Baltimore: Johns Hopkins Press, 1964)Google Scholar.
2 Fishlow, Albert, American Railroads and the Transformation of the Ante-Bellum Economy (Cambridge: Harvard University Press, 1965)Google Scholar.
3 With this objective in mind, the discussion of more technical issues will be con-fined to footnotes.
4 The caution of historians has not entirely eradicated from their writings such words as “decisive” or “indispensable.” When used to connote the contribution of a given factor to aggregate growth, such words would seem to imply the sort of counterfactual speculation outlined above. If such is not the case, it is difficult to understand what they do imply, other than the unrestrained enthusiasm of the writer for his subject.
5 Fogel, , Railroads, p. 20Google Scholar. For three other possible definitions of social saving refected by Fogel, see ibid., pp. 20–21.
6 Fishlow, , American Railroads, p. 63Google Scholar.
7 If the subscripts c and w refer to water and wagon transportation, respectively, then
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20151017021501848-0502:S0022050700076075_eqnU1.gif?pub-status=live)
would be comprised of
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20151017021501848-0502:S0022050700076075_eqnU2.gif?pub-status=live)
where Rc indicates the ton-miles heretofore carried by railroads and now diverted to water transportation.
8 Fishlow actually calculates three measures of social saving. An appendix criticizing the other two is available from the writer on request.
9 Fogel, , Railroads, p. 19Google Scholar.
10 “…averages that cannot reliably be calculated until the linear programming problems are solved.” (Ibid., p. 38). These problems are never solved. However, the failure to find the least-cost solution is not serious, provided alternative procedures give estimates that exclude serious downward biases. (Without such an exclusion one has no assurance that the final measure is an upper bound.)
12 Ibid., p. 41.
13 Ibid., p. 38.
14 U.S. Congress, Senate, Preliminary Report of the Inland Waterways Commission (60th Cong., 1st Sess., 1908), Doc. 325, pp. 205-9 (hereafter cited as Waterways Commission). The page citation is apparently incorrect. What evidence is available in this report is discussed below.
15 Southwestern Freight Tariff, No. 9 (November 16, 1890).
16 Waterways Commission, p. 236. A footnote states: “From 1883 to 1899 canal rates are based on rates on wheat from Buffalo to New York.”
17 Fogel, , Railroads, p. £38Google Scholar.
18 Waterways Commission, p. 344. This calculation is based on the assumption that I bushel of wheat weighed 60 pounds. The distance by water from St. Louis to New Orleans (1161 miles) was obtained from U.S. Bureau of the Census, Eleventh Census of the United States, 1890. Transportation, II, 420-22.
19 Waterways Commission, p. 344.
20 For a definition of “intraregional,” see footnote 9.
21 Fogel, , Railroads, p. 70Google Scholar.
22 Ibid., p. 67.
23 Even Fogel admits (p. 55) that the final results contain a number of upward and downward biases, with the ultimate effect unclear.
24 Ibid., pp. 219-21.
25 Ibid., pp. 221-22.
26 Ibid., p. 223.
27 Ibid., p. 243.
28 Fishlow, , American Railroads, p. 32Google Scholar.
29 See Ibid., p. 92.
30 Ibid., p. 52.
31 Ibid., p. 32.
32 Ibid., p. 93.
33 Ibid., p. 73. Fishlow incorrectly gives the reference as Appendix B in Taylor's book. The relevant data may be found in Taylor, George Rogers, The Transportation Revolution, 1815-1860 (New York: Holt, Rinehart and Winston, 1951), Appendix A, p. 442Google Scholar.
34 Taylor, , Transportation Revolution, p. 133Google Scholar.
35 Ibid., p. 134.
36 MacGill, Caroline E.et al., History of Transportation in the United States Before 1860 (Washington, D.C.: Carnegie Institution, 1917), p. 223Google Scholar.
37 McPherson, Logan G., Railroad Freight Rates in Relation to the Industry and Commerce of the United States (New York: Henry Holt and Company, 1909), pp. 148–49Google Scholar.
38 Hunt's Merchants' Magazine, V (09 1841), 284Google Scholar.
39 Vose, George L., Handbook of Railroad Construction: For the Use of American Engineers (Boston: James Monroe and Company, 1857), p. 3Google Scholar; U.S. Congress, Senate, Andrews Report (32d. Cong., 1st Sess., 1853), Sen. Exec. Doc. 112, p. 380.
40 Vose, , Handbook, p. 3Google Scholar.
41 Development of Transportation Systems in the United States (Philadelphia: J. L. Ringwalt, 1888), p. 28Google Scholar.
42 Seaman, Ezra C., Essays on the Progress of Nations (New York: Charles Scribner, 1852), p. 363Google Scholar.
43 Fishlow, , American Railroads, p. 52Google Scholar.
44 A few words of explanation may be necessary for those unfamiliar with this terminology. Marginal cost is defined as the increment to total cost from carrying one more ton of wheat one mile. Before the railroad is built, the canal receives, per ton-mile of wheat moved, 20 cents, while the incremental cost of moving that wheat is only 5 cents. Thus, 15 cents has been earned to put toward profits and to defray such fixed costs as canal maintenance. For a further discussion of the differences between marginal costs, variable costs, and fixed costs see Samuelson, Paul A., Economics: An Introductory Analysis (4th ed.; New York: McGraw-Hill, 1958), ch. xxivGoogle Scholar.
45 This is not equivalent to assuming constant returns to scale, insofar as the existence of fixed costs will cause average total costs to fall continually, ultimately approaching the marginal cost curve asymptotically.
46 If the general price level has changed in the interim, further adjustments would be necessary to make the two prices comparable.
47 The reader familiar with the concept of consumer surplus will recognize the standard bias problems, depending on the elasticity of demand and which quantity one uses (the smaller shipments when prices were high or the larger shipments after prices have fallen). Such refinements can be ignored here insofar as this measure will have little bearing on subsequent analysis. For further discussion, see Fishlow, , American Railroads, pp. 23–32Google Scholar.
48 Bias problems also enter here. One may wish to consider only that proportion of traffic heretofore carried by canals and now carried by railroads ( = — [
— R]) or the total traffic now carried by railroads (R). In subsequent discussion, the latter formulation will be used to simplify the analysis.
49 The validity of this measure depends on the assumption that marginal costs remain constant for both carriers as wheat is shifted from railroads to canals.
50 If perfect competition prevails in those alternatives and resource shifts are small, then the value of resources saved will exactly equal the value of the output which they create in those alternatives (see Fishlow, , American Railroads, p. 26)Google Scholar.
51 The relevant interest rate would be that on riskless securities, insofar as the desired comparison is with what one could have been certain of making in the next best alternative. The necessity for including a depreciation allowance should now be clear. Suppose that $1000 is invested either in securities yielding 7 percent or in building a railroad that wears out completely in one year. If securities are bought, at the end of the year one would have $70 plus the original $1000 worth of assets. The relevant comparison with railroads therefore includes an expenditure on railroads to offset depreciation, thereby guaranteeing that one is contrasting two choices, both of which will keep the value of assets invested ($1000) intact.
52 Consider a case of only two alternatives—canals and wagons—denote4d by the subscipts c and w, respectively. Then
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20151017021501848-0502:S0022050700076075_eqnU3.gif?pub-status=live)
would represent
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20151017021501848-0502:S0022050700076075_eqnU4.gif?pub-status=live)
where
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20151017021501848-0502:S0022050700076075_eqnU5.gif?pub-status=live)
53 See Sec. I, par. 2.
54 See Sec. I, par. 2.
55 His one relevant observation is that “even if water rates in 1890 equalled marginal cost …” (Fogel, Railroads, pp. 27-28), but at no point is it explicitly stated that such was indeed the case.
56 Fishlow, , American Railroads, p. 29Google Scholar.
57 In any oligopoly the relationship between average revenue and marginal cost is indeterminate, barring further assumptions about the behavior of the firms. For examples of determinate solutions based upon unrealistic assumptions, see Chamberlin, Edward Hastings, The Theory of Monopolistic Competition (8th ed.; Cambridge: Harvard University Press, 1962), ch. iiiGoogle Scholar.
58 The increased competition from railways forced many canals to cut their prices, thereby passing on to the public some of the benefits achieved from earlier technological advances in transportation. The possibility that canals, if left unchallenged, would have charged monopoly prices is one reason why the earlier measure of “direct financial benefits”
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20151017021501848-0502:S0022050700076075_eqnU6.gif?pub-status=live)
cannot be equated to the Fogel-Fishlow measure
![](https://static.cambridge.org/binary/version/id/urn:cambridge.org:id:binary:20151017021501848-0502:S0022050700076075_eqnU7.gif?pub-status=live)
The greater the monopoly power, the more would
diverge from the price charged by alternative carriers (Pa) when railways were actually competing.
59 The same point has been stressed by Lebergott, Stanley, “United States Transport Advances and Externalities,” Journal Of Economic History, XXVI (12 1966), 439Google Scholar. This applies, of course, only to the extent that railways did compete directly with alternative carriers. When that competition was not severe, a close similarity between railroad prices and those of alternative carriers would have been unlikely. Even more unlikely, however, would have been an identity of price with marginal costs within all transportation companies.
60 “…using the 1890 rates [for water transportation] is equivalent to assuming that the marginal cost of water transportation was constant over the relevant range (Fogel, Railroads, p. 28). If water carriers now transport C ton-miles and are to acquire an additional Rc ton-miles as railways disappear, the relevant range through which marginal costs must be constant is from C to C + Rc.
61 Fogel, , Railroads, p. 28Google Scholar.
62 Ibid.
63 Ibid.
64 Ibid. No attention will be given to Fogel's suggestion that water transportation may have been a declining cost industry. Those interested in problems of- market structure will nevertheless be struck by the difficulty of achieving a stable equilibrium in an oligopoly where one firm has falling costs. One reservation might be noted. If boat building in 1890 was a highly competitive industry, most scale economies would have already been exploited, thereby removing the most obvious source for the pecuniary external economies hinted at by Fogel on p. 28.
65 One brief aside is made by Fishlow (p. 93) on the possibility of rising costs in wagon transportation under the additional burden of one billion ton-miles of freight. Difficulties thereby created are dismissed by the use of geometry that ap-pears to be incorrect (Fishlow's Fig. 4, p. 95). This becomes apparent when one adjusts the diagram for the existence of two markets, the one lacking access to railroads and the other ignoring wagon transportation as long as the railroads are available.
66 Fishlow, , American Railroads, pp. 33–34Google Scholar.
67 Ibid., p. 33.
68 Ibid., p. 64.
69 The data on freight rates used in Chart 1 refer to flour shipped, per barrel of 216 pounds. Comparable data were not available for the food products that dominated eastbound canal traffic, namely, wheat and corn. Such information as could be found does indicate a similar pattern of wildly gyrating freight charges (see Table 3).
70 See Table 2.
71 See Table 4.
72 “The locks on the Erie Canal from Troy to Montezuma are all double. From Montezuma to Buffalo the locks are all single, with the exception of the five combination locks at Lockport. The eastern section of the Erie Canal is taxed with the business of the lateral canals, including the Oswego, Black River, Chenango, Seneca and Cayuga, and Chemung, in addition to the movement through the main trunk from Buffalo.” Buffalo Board of Trade, Annual Statement of the Trade and Commerce of Buffalo for the Year Ending December 31, 1865 (Buffalo: Matthews and Warren, 1866), p. 17Google Scholar. For complaints of overcrowding on the Erie during the Civil War, see ibid.; also New York Chamber of Commerce, Sixth Annual Report for the Year 1863-64 (New York: John W. Ameriman, 1864), pp. 145-53; U.S. Congress, Senate, Memorial of the Chamber of Commerce of the State of New York Praying the Enlargement of the Erie and Oswego Canals (37th Cong., 3d Sess., 1862-63), Sen. Misc. Doc. 12; City of Chicago, Report of the Committee on Statistics, Submitted to the National Convention at Chicago, 06 2, 1863 (Chicago: Chicago Tribune Company, 1863), pp. 23–25Google Scholar.
73 Fishlow, , American Railroads, p. 32Google Scholar.
74 U.S., Bureau of the Census, Eleventh Census of the United States, 1890. Transportation, I, 593.
75 For a similar criticism of Fogel's work, see Nerlove, Marc, “Railroads and American Economic Growth,” Journal of Economic History, XXVI (03 1966), 112Google Scholar.
76 Fishlow himself makes no such claim. Indeed, much of his book is devoted to elaborating various other effects wrought b y the railroads through backward and forward linkages to the manufacturing and agrarian sectors of the economy.
77 “…the key to measuring the direct benefits of railroads lies in the identity of the reduction in financial cost with the reduction in real inputs required per unit of transport output.” (Fishlow, American Railroads, p. 23).
- 21
- Cited by