¹ Fogel, Robert William, Railroads and American Economic Growth: Essays in Econometric History (Baltimore, 1964)Google Scholar; Fishlow, Albert, Railroads and the Transformation of the Antebellum Economy (Cambridge, Mass., 1965)Google Scholar; Hawke, Gary, Railways and Economic Growth in England and Wales, 1840–1870 (London, 1970)Google Scholar; Vamplew, Wray, “Railways and the Transformation of the Scottish Economy,” Economic History Review, 2nd ser., 24 (Feb. 1971), 37–54CrossRefGoogle Scholar; McGreevey, William Paul, Economic History of Colombia, 1845–1930 (Cambridge, 1971), chap. 10Google Scholar; Metzer, Jacob, Some Economic Aspects of Railroad Development in Tsarist Russia (Ph.D. diss., Univ. of Chicago, 1972)Google Scholar. Somewhat higher savings were reported by Hurd, John III, “The Economic Impact of Railways in India, 1853–1947”(paper presented to the Economic History Workshop, Univ. of Chicago, Feb. 1976), p. 4Google Scholar. The existence of cheap water transport prior to railroad construction does not, of course, prevent savings from rising dramatically where railroads are constructed instead of canals or where potentially navigable rivers are left undredged. This was the point of the dispute between Fogel and Fishlow in the U.S. case. Jeffrey Williamson has argued recently that both authors have failed to measure fully the railroads' impact on the U.S. economy between 1870 and 1890 by ignoring certain indirect effects (like the impact on regional terms of trade and the spatial distribution of economic activity), and by omitting consideration of possible dynamic linkages between railroads and such variables as the rate of capital formation. See Williamson, Jeffrey, Late Nineteenth-Century American Development: A General Equilibrium History (Cambridge, 1974), chap. 9Google Scholar. Colin M. White has presented an excellent review of the debate and added to it with references to his own work on nineteenth-century Russian railways in “The Concept of Social Saving in Theory and Practice,” Economic History Review, 2nd ser., 29 (Feb. 1976), 82–101CrossRefGoogle Scholar. See also O'Brien, Patrick, The New Economic History of the Railways (New York, 1977)Google Scholar. The last word to date, and on most issues, is Fogel, , “Notes on the Social Saving Controversy,” his Presidential Address to The Economic History Association, in this Journal, 39 (March 1979), 1–54Google Scholar.

² Chapman, John Gresham, La constructión del Ferrocarril Mexicano (1837–1880) (Mexico, 1975)Google Scholar; Calderón, Francisco, La República restaurada: La vida económica, in the series edited by Villegas, Daniel Cosio, ed., Historia moderna de México (Mexico, 1955)Google Scholar, Part 3; Calderón, Francisco, “Los ferrocarriles,” in Villegas, Daniel Cosio, ed., Historia moderna de México: el Porfiriato: la vida econdmica, 2 vols. (Mexico, 1965), vol. 1, pp. 483–684Google Scholar; Pletcher, David, “The Building of the Mexican Railway,” Hispanic American Historical Review, 30 (Feb. 1950), 26–62CrossRefGoogle Scholar.

³ Lozano, Sergio Ortiz Hernán, Los ferrocarriles de México: ùna visión social y económica (Mexico, 1970), chaps. 2–3Google Scholar; Halsey, Frederic M., The Railways of Central and South America: A Manual Containing Statistics and Other Information Concerning the Important Railways of South and Central America, Mexico and the West Indies (New York, 1914)Google Scholar. For a sharp contrast, see Miller, Rory, “Railways and Economic Development in Central Peru, 1890–1930,” in Miller, Rory, Smith, Clifford T., and Fisher, John, eds., Social and Economic Change in Modern Peru (Liverpool: Centre for Latin American Studies, University of Liverpool Monograph Series No. 6, n.d.), pp. 27–52Google Scholar. Only Argentina and Brazil had comparably large railroad networks by 1910, and only the Argentine system extended over as large a portion of the national territory as that of Mexico.

⁴ Secretaría de Hacienda, Informe presentado al Presidente de la República por el Secretario de Hacienda y Criéito Público sobre los estudios y gestiones de la Secretaria de su cargo en asuntos de ferrocarriles (Mexico, 1903); John H. McNeely, The Railways of Mexico: A Study in Nationalization (El Paso, Texas, 1964).

⁵ Fishlow, Railroads, pp. 90–93; J. Boyd, Hayden and Walton, Gary M., “The Social Savings from Níneteenth-Century Rail Passenger Services,” Explorations in Economic History, 2nd ser., 9 (Spring 1972), 237–40Google Scholar; Hawke, Railways, p. 188. Fishlow's estimate places passenger savings at 31.1 percent of freight benefits; Hawke's passenger estimates run between 38.4 and 65.7 percent of freight savings; while Boyd and Walton's passenger benefits amount to 38.1 percent of Fogel's freight savings for the same year.

⁶ de Fomento, Secretaría, Memoria presentada al Congreso de la Unión por el Secretario de Estado y del Despacho de Fomento, Colonización, Industria y Comercio de la República Mexicana, General Carlos Pacheco, corresponde a los a˜os transcurridos de diciembre de 1877 a diciembre de 1882, 4 vols. (Mexico, 1885), vol. 2, p. 605Google Scholar.

⁷ Fomento, Memoria, vol. 2, contains reports from all federal highway directors on road conditions, traffic, transport costs, etc.

⁸ Calculated from ibid., in Coatsworth, John H., “The Impact of Railroads on the Economic Development of Mexico, 1877–1910” (Ph.D. diss., Univ. of Wisconsin, 1972), pp. 75–78Google Scholar.

⁹ Calculated from annual reports of all Mexican railroad companies on file in the Archivo Histórico of the Secretaría de Comunicaciones y Transportes, Mexico City (hereafter cited AHSCT, followed by file number).

¹⁰ Rail passenger fares, even for second-class travel, were quite high in relation to wages in Mexico (see below). Long-distance travel by commercial carrier probably increased more rapidly among upper-income earners than among the poor. It seems likely that some second-class passengers would have switched to stage travel as well as to riding if railroads had not existed. Since both mounted and stage travel were more costly than walking, the second-class savings estimate is exaggerated, because it assumes the least costly (and least comfortable) alternative for all second-class rail passengers. The first-class estimate is exaggerated because it assumes that stagecoach travel faced constant marginal costs over an enormous range of output.

¹¹ Stagecoach fares for fourteen major routes in 1876 are reported by Calderón, La República restaurada, pp. 604–06. The charge per passenger kilometer for each route was calculated on distances reported in Fomento, Memoria, vol. 2, passim. This charge varied between P$ 0.033 and P$ 0.083. The average for all routes was P$ 0.061. Boyd and Walton cite stagecoach rates varying from U.S. $0.05 to U.S. $0,135 per passenger mile for various dates in the nineteenth century. They employ a rate of U.S. $0.06 per mile for their 1890 estimate. At the 1890 exchange rate, this amounts to just over P$ 0.07 and about the same in pesos of 1877. This is about P$ 0.04 per passenger kilometer, or about 20 percent below Mexican stagecoach fares in the 1870s. Boyd and Walton, “The Social Savings,” pp. 243–44.

¹² Contemporary air passengers on business trips valued their time at approximately twice the U.S. hourly wage in manufacturing, according to Gronau, Reuben, The Value of Time in Passenger Transportation: The Demand for Air Travel (New York: National Bureau of Economic Research, Occasional Paper no. 109, 1970)Google Scholar. See also De Vany, Arthur, “The Revealed Value of Time in Air Travel,” Review of Economics and Statistics, 56 (Feb. 1974), 77–82CrossRefGoogle Scholar; and Boyd and Walton, “The Social Savings,” p. 245. Most studies of passengers using other modes of transport place the value of time much lower, at less than half the wage of the traveling population and well below the average wage in industry. See Hensher, David A. and Hotchkiss, Williard E., “Choice of Mode and the Value of Travel Time Savings for the Journey to Work,” Economic Record, 50 (March 1974), 94–112CrossRefGoogle Scholar; Kraft, John and Kraft, Arthur, “Empirical Estimation of the Value of Travel Time Using Multi-mode Choice Models,” Journal of Econometrics, 2 (Dec. 1974), 317–26CrossRefGoogle Scholar.

¹³ Existing historical data collections report only “average minimum daily wages,” unsuitable for our purposes. The estimates here employ the average wage paid to railway workers, based on data for the National Railways of Mexico, the government-controlled firm that employed more than half of all railway workers in 1910. According to its annual report for 1910 (Manuscript Version, AHSCT, 10/2329–1), the average wage (excluding top-management personnel) amounted to P$ 1.78 per day. This wage, adjusted by the Mexico City Wholesale Price Index (reproduced in Table 3), amounted to P$ 1.08, in pesos of 1900.

¹⁴ Calderón, La República restaurada, p. 603.

¹⁵ Boyd and Walton, “The Social Savings,” pp. 248–51.

¹⁶ GDP estimates in current pesos are taken from Solís, Leopoldo, “La evolución economica de Mexico a partir de la Revolución de 1910,” Demografia y Economia, 3 (1969), 12Google Scholar. The Solis estimates are deflated using the Mexico City Wholesale Price Index.

¹⁷ While the average journey of Mexican rail passengers in 1910 was 67 kilometers, that of U.S. passengers at a comparable stage of railroad development (the 1880s and 1890s) was approximately 25 miles, or less than 40 kilometers. In 1910 the average U.S. rail journey was 33.5 miles (56 kilometers); Hultgren, Thor, American Transportation in Prosperity and Depression (New York, 1948), p. 61Google Scholar. In Britain, the average journey covered 9 miles (14 kilometers); Hawke, Railways and Economic Growth, p. 51.

¹⁸ The 1910 average minimum daily wage in agriculture was P$ 0.26, according to El Colegio de México, Estadísticas económicas del Porfiriato: fuerza de trdbajo y actividad por sectores (Mexico, n.d.), p. 148. The second-class fare plus time cost per kilometer for rail travel amounted to P$ 0.02431, calculated from the data in Table 2. Thus, a 67-kilometer trip cost P$ 1.63 by rail, as opposed to P$ 2.41 on foot. One implication of these results is that only the relatively prosperous saved by taking the train. The second-class savings estimate uses P$ 1.08 per day as the assumed opportunity cost of time spent in traveling. If the average minimum daily agricultural wage (P$ 0.26) were used instead, the cost of walking 67 kilometers (in 2.3 days) would be only P$ 0.60 (versus P$ 1.63 by train) and the savings estimate would drop from a positive P$ 3.9 million to a negative P$ 4.5 million. Indeed, for any assumed opportunity cost of travel time below P$ 0.70 per day, the social savings estimate would be negative. Since it is not likely that more than a small minority of the Mexican population actually earned more than P$ 0.70 per day, very few indeed could actually have gained in pecuniary terms by using rail transportation. For most Mexicans the price of a second-class rail passenger ticket took more days' labor to pay for than would have been required to walk the same distance.

¹⁹ On prerail freight transport, see Coatsworth, “The Impact of Railroads,” chap. 4.

²⁰ This crude estimate is based on Fishlow's data, which indicate a maximum “feasible” capacity of somewhat less than 20,000 tons of freight per route mile for the United States in the 1850s; Fishlow, Railroads, p. 93. Applying this estimate to Mexico yields a required minimum of 167,000 route miles, or approximately 250,000 kilometers of highway to accommodate the 3.5 billion ton kilometers of freight shipped by. rail in 1910.

²¹ Reports of highway directors in Fomento, Memoria, vol. 2, passim, list dry-season rates for freight transport, which varied from P$ .058 to P$ .221 per ton kilometer. The average rate of the fourteen indicated was P$ .152. Using P$ 0.10 (more than 50 percent less than the average 1878–79 rate) exaggerates the efficiency of the prerail freight transport system and thus reduces the savings estimate.

²² Calderón, La República restaurada, pp. 596–601.

²³ It may also be observed that the railroad industry was highly competitive. The major trunk lines from the center of the country to the northern border and parallel lines from the interior to most major ports competed directly. This competition and increased government intervention in the industry might have driven tariffs slightly below marginal costs during the first decade of this century. This possibility suggests yet a further source of downward bias in the social savings estimate.

²⁴ Fogel employs this term in a discussion summarizing the results published in his book and after considerable debate on the nature of his estimates. See Fogel, Robert William, “Railroads and American Economic Growth,” in Fogel, and Engerman, Stanley L., eds., The Reinterpretation of American Economic History (New York, 1971), 196n.Google Scholar

²⁵ While the Durbin-Watson test yielded negative results at the 95 percent level (indicating a low level or absence of serial correlation), repeated manipulation of the form of the equation (eliminating the population variable, adding a time trend, fixing parameters, and the like) was undertaken in order to check for multicollinearity. The regression coefficients remained quite stable through every run, with the estimated price elasticity below 0.75 in every case.

²⁶ Wilson, George W., “Notes on the Elasticity of Demand for Freight Transportation,” Transportation Journal, 17 (Spring 1978), 11Google Scholar; Worcester, Dean Jr., “On Monopoly Welfare Losses: Comment,” American Economic Review, 65 (Dec. 1975), 1016, Table 1Google Scholar; Weiss, Leonard W., Case Studies in American Industry (New York, 1967)Google Scholar; Sloss, James, “The Demand for Intercity Motor Freight Transport: A Macroeconomic Analysis,” Journal of Business, 44 (Jan. 1971), 62–68CrossRefGoogle Scholar.

²⁷ See Stonham, P. E., “The Demand for Overseas Shipping in the Australian Export Trade,” Journal of Transport Economics and Policy, 3 (Sept. 1969), 333–49Google Scholar.

²⁸ For GDP estimates, see Solís, “La evolución,” p. 12. For population, see El Colegio de México, Estadísticas: fuerza de trabajo, p. 25.

²⁹ See El Colegio de México, Estadísticas: fuerza de trabajo, pp. 147–54.

³⁰ While substantial migration from the center to the north of the country took place during the Porfiriato, the railroads also played a role in moving labor to the sisal plantations of Yucatan. Plantation labor here, as well as in Campeche, Chiapas, and parts of Oaxaca, appears to have involved considerable coercion; forced laborers were transported to these areas from the center and north, especially from Sonora, where Yaqui Indians were captured in large numbers during a prolonged period of warfare. Railroads did stimulate henequen production (and demand for labor) in Yucatan, as they did the production of other plantation crops in the rest of the south. Migration was inhibited, however, by the absence of direct rail links (which existed to the north) and by the reputedly unhealthy climate of these areas.

³¹ The method employed for calculating the social rate of return is discussed in Mercer, Lloyd, “Rates of Return for Land Grant Railroads: The Central Pacific System,” this Journal, 30 (Sept. 1970), 602–26Google Scholar. But see also McClelland, Peter D., “Social Rates of Return on American Railroads in the Nineteenth Century,” Economic History Review, 2nd. ser., 25 (Aug. 1972), 471–88CrossRefGoogle Scholar.

³² Coatsworth, “The Impact of Railroads,” chap. 5.

³³ Mexican Central Railroad Company, Annual Report (ms. version), AHSCT, 10/3175–1.

³⁴ Mexican Central Railway Company Limited, Annual Report, 1907–1908, p. 21.

³⁵ Mexican National Railroad Company, Annual Reports, 1887, 1908.

³⁶ “A constant effort has been made to curtail, as much as possible, the purchase of material in the United States and Europe and, wherever practicable, to make purchases in Mexico,” the President of the Mexican Central reported to his stockholders (Mexican Central, Annual Report, 1894, p. 10). The Mexican National President reported a similar policy to his stockholders the year before (Mexican National, Annual Report, 1893, p. 8).

³⁷ Backward linkages to industrial development in the United States, Britain,.and Germany contrast sharply with the Mexican case; Fremdling, Rainer, “Railroads and German Economic Growth: A Leading Sector Analysis with a Comparison to the United States and Great Britain,” this Journal, 37 (Sept. 1977), 583–604Google Scholar. A parallel case is that of Peru, where backward linkages were also negligible; Miller, “Railways and Economic Development in Central Peru,” pp. 44–47. The Spanish case provides a further contrast; Tortella found that forward as well as backward linkages to the industrial sector were small, and he attributes the collapse of the Spanish banking system in the 1860s to government policies that diverted resources to railroad construction from more productive investment in other sectors of the economy; Casares, Gabriel Tortella, Los orígenes del capitalismo en España (Madrid, 1973), chaps. 3–5.Google Scholar

³⁸ Calculated from the 1910 annual reports of all Mexican railroads under federal concession, in AHSCT.

³⁹ In 1910, 4.1 percent of the employees of the National Railways were foreigners, two-thirds of whom were represented among the 5 percent highest paid. The 1,074 foreigners earned 14.7 percent of all wages paid by the company. Their average wage was P$ 6.49 per day; that of Mexican employees, P$ 1.58. The wage data are in the ms. version of the company's Annual Report, 1909–1910, AHSCT, 10/2329–1.

⁴⁰ In 1910 six major companies that accounted for more than 85 percent of all railroad earnings spent exactly one-third of their gross receipts for service payments on their bonded debt and dividend payments to stockholders. (The companies were the Interoceanic, the Mexican, the National Railways of Mexico, the Sonora, the Tehuantepec National, and the United of Yucatan.) The Mexican government earned P$ 1.4 million in dividends (it owned no bonds) from the National Railways in 1910. This estimate assumes that 95 percent of the remaining interest and dividend payments (P$ 17.2 million) left the country. The same percentage is applied to the interest and dividend payments of the rest of the railroad companies (estimated at one third of gross receipts). The annual reports of these six companies are in AHSCT. Data on the government's share of National Railways revenues are found in The Mexican Year Book: A Statistical, Financial, and Economic Annual, Compiled from Official and Other Returns, 1911 (New York, 1912), p. 167; and Nicolau d'Olwer, “Las inversiones extranjeras,” in Cosio Villegas, ed., Historia moderna de México: El Porfiriato, la vida económica, vol. 2, pp. 1074–76.

⁴¹ Coatsworth, “The Impact of Railroads,” chap. 5.

⁴² Export data are in de México, El Colegio, Estadístkas económicas del Porfiriato: comercio exterior, 1877–1911 (Mexico, 1960)Google Scholar, passim. In fiscal 1909–10, Mexican exports were valued at a total of P$ 265,727,730.

⁴³ Jenks, Leland H., “Railroads as an Economic Force in American Development,” this Journal, 4 (May 1944), 1–20Google Scholar; Chandler, Alfred D. Jr., The Visible Hand: The Managerial Revolution in American Business (Cambridge, 1977), chaps. 3–6.Google Scholar

⁴⁴ See Shafer, Robert J., Mexican Business Organizations: History and Analysis (Syracuse, N.Y., 1973)Google Scholar, chap. 2.

⁴⁵ Coatsworth, John H., “Railroads, Agrarian Protest and the Concentration of Landholding in the Early Porfiriato,” Hispanic American Historical Review, 5 (Feb. 1974)Google Scholar, passim.

⁴⁶ Published estimates of this difference have been exaggerated, however. My research indicates that domestic foodstuff production increased at about the same rate as population, while real exports grew at 5.8 percent between 1880/81 and 1910/11. See Coatsworth, John H., “Anotaciones sobre la producción de alimentos durante el Porfiriato,” Historia Mexicana, 26 (Oct.-Dec. 1976), 167–87.Google Scholar

⁴⁷ E.g., Moore, Barrington, The Social Origins of Dictatorship and Democracy: Landlord and Peasant in the Making of the Modern World (Boston, 1967)Google Scholar. For an application to the Mexican case, see Coatsworth, John H., “Orígenes del autoritarismo moderno en México,” Fōro Internacional, 16 (1975), 205–32Google Scholar