«Tariffs and the Great Depression Revisited Mario J. Crucini James Kahn Staff Report no. 172 September 2003 This paper presents preliminary findings ...»
Federal Reserve Bank of New York
Tariffs and the Great Depression Revisited
Mario J. Crucini
Staff Report no. 172
This paper presents preliminary findings and is being distributed to economists
and other interested readers solely to stimulate discussion and elicit comments.
The views expressed in the paper are those of the authors and are not necessarily
reflective of views at the Federal Reserve Bank of New York or the Federal
Reserve System. Any errors or omissions are the responsibility of the authors.
Tariffs and the Great Depression Revisited Mario J. Crucini and James Kahn Federal Reserve Bank of New York Staff Reports, no. 172 September 2003 JEL classification: E3, F4, N1 Abstract Drawing on recent business cycle research on the Great Depression, we return to an argument we advanced in a 1996 article in the Journal of Monetary Economics—the argument that features of the Hawley-Smoot tariffs could have done more to decrease economic activity than is customarily believed, though not enough to account for the severe decline of the early 1930s. Here we reformulate our argument in a business cycle
accounting framework that apportions fluctuations between three types of “wedges”:
(productive) inefficiency, the consumption-leisure margin, and intertemporal inefficiency.
Tariff increases in our model correspond primarily to productive inefficiency in a prototype one-sector model. Moreover, the wedge implied by tariffs during the Depression correlates well with the overall measure of productive inefficiency. Our model fails to produce a labor wedge of any consequence—persuasive evidence that factors other than tariffs also contributed significantly to the severity of the Depression.
Crucini: Department of Economics, Vanderbilt University (e-mail:
email@example.com); Kahn: Domestic Research Function, Research and Market Analysis Group, Federal Reserve Bank of New York (e-mail: firstname.lastname@example.org). The views expressed in this paper are those of the authors and do not necessarily reflect the position of the Federal Reserve Bank of New York or the Federal Reserve System. Portions of this manuscript are reprinted from our 1996 Journal of Monetary Economics article, “Tariffs and Aggregate Economic Activity: Lessons from the Great Depression.” 1 Introduction In our 1996 Journal of Monetary Economics paper, we made the following
1. Eﬀective tariﬀ rates during the 1930s were higher than their apparent nominal rates because of deﬂation.
2. Because of the importance of material inputs in traded goods, the impact a given tariﬀ rate could be magniﬁed because of the impact on productive eﬃciency.
3. There was substantial retaliation from foreign countries in their tariﬀ rates.
4. Consequently, even a neoclassical equilibrium model with ﬂexible prices and no other distortions suggests that tariﬀ increases of the order of magnitude that took place in the 1930s could have resulted in substantial declines in output.
5. Though large enough to look like a modest recession, these model-calibrated output declines are only on the order of one-tenth the magnitude of the actual declines that occurred during the Great Depression.
Since this paper appeared in print, some new tools for business cycle analysis have emerged. In a series of papers (Hall, 1997; Mulligan, 2002a,b; Chari, Kehoe, and McGrattan, 2002, hereafter referred to as CKM; Gali, Gertler and Lopez-Salido, 2001), movements in output and employment have been decomposed into three sources, which amount to deviations from equilibrium conditions. The three conditions are an aggregate resource constraint, a static optimality condition relating consumption and leisure, and an intertemporal condition relating capital accumulation and expected consumption growth. It should be emphasized that this decomposition is really just an accounting framework.
It does not oﬀer a deeper explanation of the fundamental causes of ﬂuctuations, but the results of the accounting exercise may shed some light on what the causes could and could not be, and provide a set of stylized facts with which theories must be consistent. Thus, for example, Hall (1997) ﬁnds that most employment ﬂuctuations in postwar U.S. data appear to be accounted for by deviations in the static optimality condition relating the marginal product of labor (MPL) with the marginal rate of substitution (MRS) between consumption and leisure. This fact is consistent with any number of theories, and proposed candidates include preference shocks, distortions in labor markets resulting from taxes, unionization, rigid prices and wages, and so on. But it is not consistent with theories of employment ﬂuctuations that result in no change in the ”wedge” between the MRS and MPL. On the other hand, both Hall and CKM ﬁnd that output ﬂuctuations are composed of a mix movements in both the MRS-MPL (or “labor”) and eﬃciency wedges.
A similar ﬁnding with respect to prewar employment has led Mulligan (2002a,b) to cast strong doubt on the role of tariﬀs in the Great Depression. Mulligan asserts that tariﬀs in the sort of model we proposed would result primarily in reductions in labor productivity, which in the accounting framework described above amount to a distortion in the resource constraint, or an eﬃciency wedge.
The idea is that the production ineﬃciency that results from the tariﬀs would show up as a decline in total factor productivity (TFP), and in the context of standard modeling assumptions would lead to very little change in aggregate employment. Moreover, Mulligan argues that such a decline in productivity is counterfactual for the 1930s.
In this paper we return to the argument we made in our 1996 paper in light of these more recent developments. We will show, ﬁrst, that indeed our model does imply that tariﬀ increases in our model correspond to an increased eﬃciency wedge in a prototype one-sector model. This would seem to support Mulligan’s view that tariﬀs were not an important factor in the Great Depression. In fact, however, it supports the argument in our paper that tariﬀs did indeed contribute, albeit to a modest (but non-negligible) degree. Even accepting Mulligan’s claim that the employment decline was entirely attributable to an increase in the labor wedge, the output decline was the result of increases in both the eﬃciency and labor wedges (as CKM conﬁrm in their section on the Great Depression). Since we only claim that tariﬀs are responsible for roughly 10 percent of the overall output decline, nothing we say contradicts in any way the importance of the labor wedge in contributing to the decline in both output and employment.
Mulligan’s second argument, that productivity did not decline in the 1930s, is potentially more damaging. It is, however, at the very least debatable.
Mulligan makes his argument on the basis of wage data. This is a reasonable thing to do under the null hypothesis of a ﬂexible price equilibrium. If the production technology is Cobb-Douglass with constant share parameters, then the wage, which must equal the marginal product of labor in equilibrium, is also proportional to the average product of labor. Since real wages did not show any decline in the 1930s, it follows that the average product of labor did not decline either.
The problem with this argument is that the more relevant measure of productivity, namely total factor productivity, in fact shows substantial declines—at least from 1929-1933—according to CKM (2002). Using wage data to infer productivity is problematic on two counts. First, there are distribution eﬀects—to the extent lower wage workers are disproportionately aﬀected by unemployment, the average wage may not be aﬀected. Of course, this problem presumably affects measured labor productivity as well. The second problem is that for whatever reason (sticky wages, labor hoarding) labor’s share of income is typically countercyclical, and indeed rises substantially during the 1929-33 period.
According to calculations by Casey Mulligan, for example, labor’s share of national income (excluding proprietors’ income) rose from 0.71 to 0.83 from 1929 to 1933.
If wages are sticky above market clearing levels then a decline in aggregate eﬃciency (from whatever source) should result in a larger quantitative impact than under ﬂexible prices. In fact, Perri and Quadrini (2002) use wage rigidities to amplify the impact of tariﬀs in their study of the Great Depression in Italy.
We conclude from our reading of the interwar productivity literature that a decline in TFP follows the peak-to-trough movements in output fairly well, with the quantitative magnitude of the swing and underlying economic reasons for the movement remaining the subject of ongoing debate. Moreover, the quantitative contribution of various shocks and their propagation mechanisms remain the subject of active business cycle research.
The plan of this paper is as follows. In the next section we will review the historical evidence. Then we will present the model of tariﬀs and economic activity from our 1996 paper, examining both the steady-state implications of permanent tariﬀ increases and business cycle implications for cyclical variation in tariﬀs. Next we will use the one-sector stochastic growth model as a prototype (as suggested in the CKM paper) to show how tariﬀs in our three-sector two-country model translate into wedges in the prototype model. We will then compare the implied wedges with the historical ones, and show that the impact of tariﬀs is both consistent with the historical evidence (i.e. they do not imply wedges that were nonexistent), and moreover are well correlated with the distortions evident in those data.
2 The historical context Our 1996 paper identiﬁed three historical facts that are essential to understanding why the macroeconomic eﬀects of tariﬀs in the Great Depression were potentially much larger than has previously been thought. First, tariﬀ levels increased both at home and abroad by a factor of at least three from 1928 to 1933, not just from statutory changes but also from the interaction of deﬂation and speciﬁc (as opposed to ad valorem) tariﬀs. The magnitude of the tariﬀ increases were too large to be “optimal tariﬀs,” even for a large economy such as the United States. Further, foreign retaliation tends to wipe out such gains leaving the U.S. and its major trading partners worse oﬀ. Second, the majority of imports into the United States were material inputs; as a result, tariﬀs introduced production distortions. Third, the tariﬀ changes were persistent so their eﬀects were propagated through changes in the stock of capital. In this section we review U.S. trading patterns and present a brief tariﬀ history.
2.1 Interwar trading patterns We begin with an examination of the volume and composition of trade between the U.S. and some of its major trading partners: Canada and Europe (consisting of France, Germany, Italy, and the United Kingdom).
The pattern of U.S. trade was quite diﬀerent during the interwar period than observed today. As Table 1 indicates, U.S. trade was heavily skewed toward non-manufactured goods. For every dollar of non-manufacture exported, the U.S. exported less than 50 cents of manufactures (imports were even more skewed). Thus, the U.S. trade balance shows no obvious pattern of specialization across manufactured versus non-manufactured goods. In contrast, France and Germany exported more than 2 dollars of manufacturers for every dollar of non-manufacture exported and imports are even more skewed in the opposite direction, favoring raw materials. Thus the industrialized countries of Europe did have a distinctive pattern of specialization which favored manufactured goods.
Canada’s exports were reasonably balanced across categories, but imports favored manufacturers very strongly. In terms of trading partners, Canada and the United Kingdom were the two most important sources and destinations for U.S. products. Canada’s geographic proximity was probably important as was the United Kingdom’s dominant position in world trade.
2.2 A brief tariﬀ history Much of the historical tariﬀ literature has focused on questions of political economy, most prominently in the U.S. case, by Frank Taussig (1931) and more recently in studies that focus on the Hawley—Smoot tariﬀs by Eichengreen (1989) and Irwin and Kroszner (1996): Why was such a bill passed at such a crucial time? Who beneﬁted (ex ante) and who lost? While the political origins of interwar tariﬀs are by now fairly well understood (as classic examples of political log—rolling), their macroeconomic impact is not, and this is the question on which we focus.
Many countries passed legislative increases just after World War I and again during the period from 1927 to 1932. Historians emphasize internal reasons for the escalation of tariﬀ levels following the war and emphasize international retaliation in the wake of the infamous Hawley—Smoot Tariﬀ Act during the 1930’s.1 1 Jones (1934) discusses the question of retaliation in detail.
Table 2 reports summary statistics for international tariﬀ indices computed as the ratio of customs duties to total imports (except for the U.S. where the ratio of customs duties to dutiable imports is also presented). Using total imports (to be consistent with data available from other countries) tariﬀs in the United States rose from the level of 13 percent during the 1920’s to 16.6 percent during the 1930’s, while those in most European countries more than tripled.
Comparing these numbers gives the impression that the U.S. bore the brunt of the tariﬀ escalation. On a U.S. trade-weighted basis, however, things look more symmetric with foreign tariﬀs rates rising from 9.9 percent to 19.9. These numbers reﬂect the more modest increases in tariﬀs imposed by Canada and the U.K. (from all sources) and the fact that these two countries account for a considerable fraction of U.S. exports. While these estimates provide a useful starting point, they are reasons to interpret them with caution.