Created 12/28/1998
Go to
Brad DeLong's Home Page

Comment on Doug Irwin, "Did Late Nineteenth Century U.S. Tariffs Promote Infant Industries?: Evidence from the Tinplate Industry"

J. Bradford DeLong
U.C. Berkeley

Douglas Irwin examines the late-nineteenth century U.S. tinplate industry as a case study of the effects of protection. Were there dynamic economies of scale or of learning--"infant industry effects"--that were large enough to make the imposition by the McKinley tariff of high duties on imported tinplate a socially efficient policy for the late nineteenth century United States?

This is an interesting and useful paper, and Doug Irwin does a good job of wrestling with this particular--single case study--problem. But it is worth pointing out that the process by which Irwin arrives at this particular case carries a lot of information, a lot of information to the effect that this case is unusual--that any conclusions to be reached from this case cannot be generalized to the late nineteenth-century economy, and that even if Irwin had found the tinplate tariff to be economically efficient, such a finding would have done little or nothing to alter our standard--free trade--assessment of the consequences of America's late nineteenth century tariffs for economic development.

As Irwin rightly points out, the tinplate industry is one of the few places to examine the infant industry argument as applied to the late nineteenth century economy because the tinplate industry is one of the few industries that was genuinely an infant, and thus one of the few industries where you can apply the infant-industry presumption of extraordinarily valuable social learning as an economy learns to handle the technologies needed to produce a new product. Other American industries were already established by the late nineteenth century. You can argue that they were still subject to dynamic external learning-by-doing effects, but be careful! To the extent that the industry you are examining produces goods that are then used by other industries--produces capital goods or intermediate inputs--you must face the argument that the customer industries would have benefited from higher scale, and thus what you gain in dynamic external learning-by-doing upstream in the production process you lose in foregone dynamic external learning-by-doing downstream.

Outside of the very narrow, very young, truly infant-industry context, any argument for the positive effects of protection working through learning-by-doing has to explain why the particular industry protected exhibits stronger such effects than do its customers. And such an argument is rarely made. Thus in the modern literature we find such examples as, say, sociologist Peter Evans, who in his Embedded Autonomy praises Brazil's policy of excluding computer imports as a magnificent success because it enables Brazilian producers to acquire the capability to produce a VAX-level minicomputer for four times the world price at the end of the VAX's life as a viable commercial product. He simply does not see the negative impact on all of those industries--from programming to retail sales--that would have benefited from access to foreign computers and would have undergone similar processes of learning how to handle post-industrial technologies had they not been sacrificed to the development of the Brazilian hardware industry.

It is for this reason that most discussion of optimal tariffs in the study of U.S. economic history has focused on the pre-Civil War period--on the antebellum textile spinning and weaving industries and on the cotton tariff. For these industries are (or are close to) consumer-goods industries, hence there are no downstream producers whose external learning-by-doing effects can be disrupted, and these industries were truly infant.

And it is for this reason that the opening of Irwin's paper, in which he details the difficulty of finding an industry sufficiently infant for the optimal tariff infant industry argument to apply in late nineteenth century America, goes a long way to deconstructing any possible attempt to then use pro-protection or ambivalent conclusions from the tinplate industry to draw conclusions about the McKinley and other tariffs as a whole.

In fact, even the tinplate industry is not ideal from Irwin's perspective. Consider that leading member of the Tinplate Consumer's Association of the United States: the Standard Oil Company. If ever there was an organization dedicated to the ruthless exploitation of all possible economies of scale and of scope, it was Microsoft--sorry, it was the Standard Oil Company. You ship oil in metal barrels, after all, and sell it in tin cans. Irwin doesn't consider the drag imposed on dynamic learning-by-doing downstream as a result of the high tariff-driven spike in U.S. tinplate prices in the early 1890s.

Once Irwin turns to the detailed workings of the tariff, I have little to say in comment on the paper besides "I agree." The methodology is interesting, and the conclusions appear solid and robust. The McKinley tariff raised foreign producers' costs, but by no more than the growth of U.S. metallurgy was lowering U.S. producers' relative costs in the course of a single decade. All the McKinley tariff did was advance the trend of foreign relative to domestic costs by a decade. Moreover, the McKinley tariff's longevity was uncertain--much more uncertain than were the cost savings produced by the ongoing course of industrial development.

Hence the conclusion: the McKinley tariff at most shaved a decade off of the time needed for the U.S. to develop a tinplate industry. And--at least as Irwin calculates the welfare economics--the McKinley tariff was higher than would have been desirable to compensate for the higher-than-market tariff-driven costs of the metal inputs needed to produce domestic tinplate.

Professor of Economics J. Bradford DeLong, 601 Evans Hall, #3880
University of California at Berkeley
Berkeley, CA 94720-3880
(510) 643-4027 phone (510) 642-6615 fax

This document:

Search This Website