Railroads and American Economic Growth: Essays in Econometric History

Robert W. Fogel, Railroads and American Economic Growth: Essays in Econometric History. Baltimore: Johns Hopkins Press, 1964. xv + 296 pp.

Review Essay by Lance Davis, Division of Humanities and Social Sciences, California Institute of Technology. led@hss.caltech.edu

For those of us who lived through the exciting days of the “cliometric revolution,” the publication of Robert Fogel’s Railroads and American Economic Growth represented a very major milestone – it was as if we now had proof that we had left the bumpy and unpaved dirt road of the first few years and could see ahead a straight and well-paved highway into the future. (See note 1.) The roots of “clio” clearly lay in the 1956 publication of Cary Brown’s “Fiscal Policy in the Thirties: A Reappraisal” and, a few months later, in Alfred Conrad and John Meyer’s initial presentation of “The Economics of Slavery in the Ante-Bellum South.” Brown showed that, unlike the findings of the then-current historiography, government economic policy during the 1930’s was not an example of President Roosevelt’s imaginative application of the modern tools of Keynesian fiscal policy; and Conrad and Meyer demonstrated that, despite nearly a century of traditional historiography, ante-bellum slavery was profitable and, at least by implication, that, if the goal was to eliminate slavery before the 1940’s, the Civil War was not an extremely costly and totally unnecessary enterprise. However, these findings – findings that have been well substantiated by later research – while convincing to the small cadre of “converted,” were still not generally accepted by the historical profession. Thus, cliometrics did not really begin to flower until the publication of Robert Fogel’s study of the impact of railroads on American growth in the nineteenth century. Not only did it generate a spate of parallel studies (of Russia, Mexico, Brazil, England, and Scotland, to cite only five), but much more importantly, it provided a methodological foundation for the systematic study of economic history and long-term economic growth.

Despite the attention that had been paid to the construction of the Erie Canal, given the role of the national market in underwriting this country’s rise to become, economically at least, the richest nation in the world, and, given the speed with which rails came to dominate the transport network that provided the basis for that national market, it is not surprising that historians had concluded that railroads were the indispensable and driving force behind American growth in the nineteenth century. To the best of my knowledge, before the first annual Cliometric Conference (a conference held at Purdue University in 1960), few economic historians, neither those traditionally nor those cliometrically inclined doubted this fundamental tenant of American development. (See note 2.) Moreover, although some cliometricians may have been aware of the concept of social savings – a concept that was closely related to the economic literature on cost/benefit analysis – none had attempted to measure the savings attached to any specific legal or technical innovation. (Fogel had touched on a similar concept in The Union Pacific Railroad (1960), but his first published paper dealing specifically with social savings was still almost two years in the future – “A Quantitative Approach to the Study of Railroads in American Economic Growth” (1962).)

With its publication, Railroads proved once and for all that economic history, while still depending on the product of scholars “slugging it out in the archives,” could benefit mightily from the careful application of economic theory and econometrics. On the one hand, although the work immediately generated substantial controversy, and even today one might quibble about a few days or a few months, in the long run, there has been little question about the book’s major conclusion – that the level of per capita income achieved by January 1, 1890 would have been reached by March 31, 1890, if railroads had never been invented. Moreover, Fogel’s work also indicated that there was no other industry that was likely to have been more important than the railroads; and, thus, if not railroads, no other industry could have played the role that historiography attributed to the rails. On the other hand, the evidence is overwhelming that, since the publication and subsequent debate over Railroads, almost all economic history has been written by scholars who have either been trained in economics or who have found it necessary to acquire (either formally or informally) those basic economic and econometric skills. What, then, in addition to the central importance of the subject, made this such a path-breaking work? As the title suggests, the book is actually a collection of four interrelated, but really distinct, substantive essays: “The Interregional Distribution of Agricultural Products,” “The Intraregional Distribution of Agricultural Products,” “Railroads and the ‘Take-off’ Thesis: The American Case” and “The Position of Rails in the Market for American Iron, 1840-1860: A Reconstruction.” Any attempt at evaluating the contribution of the book rests on the evaluation of the methods and findings of the four.

If Fogel had limited his work to the last two essays – the two that in many ways were the most central to the then intense discussions of the “Axiom of Indispensability,” the work would have been important; but it would never have had anywhere near the impact that it actually did. In the third essay, “The Takeoff,” Fogel, although not addressing the question of whether or not there was in fact a “takeoff” between 1843 and 1860, in order to operationalize his argument, chooses the first of W.W. Rostow’s criteria for a “leading industry”: in this case, what impact did the railroads have on the “change in the percentage distribution of output among the various industries?” Then, drawing on the best available data – data reported by Robert Gallman in his seminal (1960) study of commodity output – Fogel finds that the impact of the railroads on that percentage distribution was minimal. In the case of iron, railroads, except at the end of the period, accounted for only a minor fraction of the output change (overall, including the later period, it was still only 17 percent); for coal, it was less than 5 percent; for lumber, barely 5 percent; in the case of transport equipment only 25 percent (only half of the change accounted for by vehicles drawn by animals); and for machinery it was less than 1 percent. Thus, for all manufacturing, the railroads accounted for less than 3 percent of the change – hardly a ringing endorsement for what was purported to be a “leading industry.”

In his more detailed examination of the impact of railroads on the development of the iron industry (an attempt to assess the importance of railroads to industrialization because of their alleged “backward linkages”), Fogel found it necessary to produce a new series on pig iron output between 1840 and 1860 and to revise the estimates of the consumption of railroads to account for imports and recycled rails as well as changes in the weight of rails. These new estimates represented a major contribution to our understanding of the industrial history of the period. Fogel’s primary interest, however, was not on the production of the new series, but on estimating the importance of the railroads in the development of the iron industry. His results, again, indicate that railroads did not dominate the development of the iron industry in the two decades before the Civil War. In fact, his conclusions strongly support Douglass North’s conclusion that, from the point of view of backward linkages, it would be as sensible to talk about an iron stove theory of the development of the iron industry as a railroad theory.

In these two essays Fogel demonstrates a command of what had heretofore been the best of traditional economic history, but in neither chapter are there any major methodological breakthroughs – merely a carefully constructed series of new estimates and the demonstration of an ability to bring those estimates to bear on important issues. In the first and second of the four substantive chapters – the estimate of the social savings from the interregional and from the intraregional distribution of agricultural products – Fogel’s methodological innovations do, however, play a central role. First, in both essays, he attempts to explicate and to provide estimates of the appropriate counterfactual – what the world would have been like had there been no railroads. Although historians have long employed counterfactual arguments – sometimes it seems without realizing it – to most historians the idea of an explicit counterfactual was still a very foreign notion in the early 1960s. Second, in both chapters Fogel employs the concept of social savings (the difference in social costs between the real and the counterfactual worlds) to provide a measure of the value of the introduction of the railroad. The concept of social savings is itself an important research tool; but, from a methodological point of view, it is equally important that the measure was defined operationally, so that Fogel’s calculations could be tested against alternative estimates and against possible alternative definitions. As an aside, however, it is interesting to note that, although the two studies are very very important from the view point of methodological innovation, from the point of view of traditional economic history, they are not as strong as the third and fourth substantive essays. In the second substantive essay – the social savings arising from the intraregional distribution of agricultural commodities – Fogel begins by noting that the substitution of rail for water was more rapid in the intraregional than in the interregional distribution of agricultural commodities, and, that, since the distances to be shipped in the intraregional case were only a third as great for rail as for water transport, one would expect that the social savings from the innovation would be greater. To estimate those savings he proposes two measures: alpha (a direct measure of the cost differences with and without the railroads) and beta (an indirect measure based on the difference in the value of the land that would have been economically productive with railroads and the lesser number of acres and, thus, the lesser value of land that would have been economically productive in the absence of those railroads).

Fogel then estimates alpha for a sample of counties in the North Atlantic region and concludes that the direct costs (alpha) would amount to a loss of 2.5% of GNP, and that adjustment for excluded indirect costs (alpha-2) would have increased that figure to 2.8% of GNP. Neither estimate, however, includes the potential savings that would have resulted from the construction of additional canals and better roads. He admits that the North Atlantic region may not provide an adequate representation of the entire country, but he argues that it would be too expensive and difficult to extend this direct measure of savings to the rest of the country.

As an alternative, Fogel suggests that, since water transport was available for about 76% of the land value in the U.S., since, in the absence of railroads, 75% of the loss of land value would be in the four states of Illinois, Iowa, Nebraska, and Kansas, and since all of the lost land could be brought into production with only a small extension of the canal network, a measure based on the difference in the value of arable land provides an equally good measure of social savings. He concludes that the cost of the direct loss of arable land from the absence of railroads (beta) would amount to 1.8% of GNP, and that the total loss – the sum of direct and indirect costs (beta-2) – would amount to 2.1% of GNP. Again, however, beta-2 does include the potential savings that would result from additional canals and better roads. Making further adjustments for the unbuilt canals and better roads, Fogel provides two estimates for the social savings from intraregional trade: alpha-3 equal to 1.2% of GNP and beta-3 equal to 1.0% It was, however, Fogel’s estimates of the social savings generated by railroads in interregional shipping (the first substantive essay), that really touched off the methodological revolution. As in the second essay, the use of explicit counterfactuals and the innovation of the concept (as well as his estimates) of the social savings broke new ground. In this case, however, there were also other very important methodological innovations.

Fogel begins with an operational definition of interregional distribution: “the process of shipping commodities from the primary markets of the Midwest to the secondary markets of the East and South.” While there were good estimates of agricultural production and agricultural exports, there were no data on the method and routes of shipment that were used to move agricultural commodities from producing areas to the points of domestic and foreign consumption; and it is here that Fogel introduces his single most significant innovation. He focuses of four commodities (wheat, corn, beef, and pork) – commodities that together represented 42 percent of agricultural income. He, first, estimates the export surplus at ten primary markets in the west and the consumption in the almost 200 deficit trading areas in the East and South (exports are attributed to the port from which they were shipped). The potential rail and water shipping routes from West to East were easily identified, and the costs of rail and water shipment were well known. To simplify the problem, Fogel focuses on a sample of 30 of the 825 potential routes between pairs of cities in the West and the East. Since the actual choice of routes is unknown, he very imaginatively suggests a linear programming model to estimate the routes – with and without railroads – that would have been selected had the shippers been guided by cost minimization. He then estimates the costs of the inferred shipments, costs estimated both with and without rails. Since there were also additional costs of water transport (lost cargoes, transshipment expenses, extra wagon haulage, time lost because of slower speed and because the canals and rivers froze, and the capital costs of the canals that were not included in the water rates), Fogel adjusts his original cost differentials to account for these additional expenses. His result is an estimate of the social savings in interregional shipment resulting from the innovation of railroads of six-tenths of one percent of GNP, a figure that would have increased to only 1.3%, had he assumed that rail rates were zero.

In this chapter Fogel made four important innovations that were to have a major impact of the nature of research in economic history: (1) the operational definition of social savings; (2) the use of an explicit counterfactual; (3) the use of a formal economic model to estimate what costs would have been had the decisions been made by economic man; and (4) his choice, when it was necessary to make assumptions about the actual world, of assumptions that were biased against his central findings. (See note 3.) Even more than his estimates of interregional social savings, the work in this essay completely changed the way economic historians would do business in the future. There is, however, one blemish in the story. Professor Fogel never actually solved the linear programming problem; his choice of routes was based on what he assumed the solution would have been.

Notes:

1. To give you some feeling about that first decade, one might note that the term “cliometrics” was coined by my then colleague at Purdue, Stanley Reiter – he had been toying around with questions raised by a new discipline that he called “theometrics” (for example, “how many angels can dance on the head of a pin?); and, in his joking way, he suggested that the work in quantitative history seemed to be drawn from similar academic stream.

2. Bob Fogel and, perhaps, Douglass North and Al Fishlow, were the major exceptions. Fogel, himself, has said that he began his investigation fully believing that it would confirm the importance of the railroads. Fishlow (1965) reached conclusions for the antebellum period very similar to those Fogel reached about the latter part of the nineteenth century. Not long before this, North (1961, p. 164) wrote, “While the value added of rails was approximately $6.5 million in 1860 and roughly equals to the value added of bar iron, it was dwarfed by the value added of the polyglot classification of iron castings, which was $21 million in 1860. Indeed, the value added in stove making alone was equal to that of iron rails.”

3. For example, Fogel made no adjustment for changes in non-rail transport that might have been made had there been no railroads: he holds both origins and destinations fixed despite the fact that there would almost certainly have been some such adjustments in the absence of railroads; and he assumes that, in the absence of railroads, water rates would be constant rather than declining as might have been the case had canal builders exploited potential economies of scale.

References:

E. Cary Brown. 1956. “Fiscal Policy in the Thirties: A Reappraisal,” American Economic Review, 46 (December).

Alfred Conrad and John Meyer. 1958. “The Economics of Slavery in the Ante-Bellum South” Journal of Political Economy, 66 (April). This paper was first presented at the meeting of the Economic History Association in 1956.

Albert Fishlow. 1965. American Railroads and the Transformation of the American Economy. Cambridge, MA: Harvard University Press.

Robert Fogel. 1960. The Union Pacific Railroad: A Case Study of Premature Enterprise. Baltimore: Johns Hopkins Press.

Robert Fogel. 1962. “A Quantitative Approach to the Study of Railroads in American Economic Growth: A Report of Some Preliminary Findings,” Journal of Economic History, 22 (June).

Robert E. Gallman. 1960. “Commodity Output in the United States,” in Conference on Income and Wealth, Trends in the American Economy in the Nineteenth Century, 24, Studies in Income and Wealth. Princeton: Princeton University Press.

Douglass North. 1961. The Economic Growth of the United States 1790 to 1860 Englewood Cliffs, NJ: Prentice-Hall.