The Cotton-Grain Coalition and the RTAA Thomas Oatley Department of Political Science University of North Carolina at Chapel Hill December 2005 Draft Manuscript, Comments Welcome Abstract Why did American trade policy shift from protectionism to negotiated liberalization in the early 1930s? The consensus view is that the initial 1934 legislation reflected the Democratic congressional majority. Its persistence over time reflects the growth of export-oriented manufacturing. This paper challenges this consensus by suggesting that the initial shift, as well as its persistence through the onset of World War II, was caused by a coalition realignment. Trade policy was determined by the coalitions formed among three legislative blocks. The northern industrialized states formed an import-competing protectionist block. Southern cotton states comprised an export-oriented liberal block. The third, and pivotal, block was composed of the western grain belt states. Between the late 19th century and the early 1930s, the grain belt aligned itself in a protectionist coalition with import-competing manufacturing industries. In the early 1930s, the grain belt abandoned this coalition to join southern cotton in support of negotiated liberalization. This coalition realignment pushed American trade policy on a liberal path and kept it on this path through the onset of World War II. A. Introduction No single policy change has received more attention from students of international political economy than the American shift from protectionism to reciprocal trade liberalization in the 1930s. It is not difficult to understand why this policy shift has received so much attention. For one thing, the policy shift was sharp and dramatic. In 1930 Congress passed legislation, the infamous Smoot-Hawley Act, which raised the average tariff to an historic high. Only four years later Congress amended this legislation with the Reciprocal Trade Agreements Act which empowered the executive to reduce tariffs by as much as fifty percent in exchange for equivalent reductions by foreign governments. For another, the American policy shift provided the foundation for the global economic system established after World War II. Indeed, it is hard to imagine that anything like the General Agreement on Tariffs and Trade would have been created after World War II had the United States not adopted a more liberal trade policy in the preceding years. The American shift has been so heavily studied, therefore, because it represents a dramatic policy change with far-reaching international consequences. Scholars examining this policy shift have converged around common answers to two distinct questions. First, why did the United States abandon its traditional protectionist policy in favor of reciprocal liberalization in 1934? Second, why was this policy shift so durable, that is, why has the policy of reciprocal liberalization persisted for more than seventy years? There is a fairly broad consensus that partisanship, embodied in the majority status of the traditionally free-trade oriented Democratic Party, explains the initial passage of the RTAA in 1934. There is also fairly broad consensus that the policy of reciprocal liberalization has persisted for so long because of the growing political influence of export-oriented manufacturing industries. Scholars disagree, however, over why exporters gained strength. Some scholars argue that growing exporter strength was caused by the crisis of the 1930s and the destruction of overseas competition in World War II (Ferguson 1984; Frieden 1988; Hiscox 1999). Others argue that it was caused by the RTAA itself, as this institutional change encouraged exporters to organize and pressure their legislators for trade-liberalizing agreements (Bailey, Goldstein, and Weingast 1997; Gilligan 1997). Still others suggest that both factors were important (Irwin and Kroszner 1999). These differences about causal mechanism aside, existing literature has increasingly converged around a two-stage explanation of this policy shift. The Democrat majority in Congress explains the initial passage of the RTAA, while the subsequent strengthening of export-oriented industry explains its persistence. This paper challenges this consensus by taking a detailed look at the initial adoption and subsequent extension of the RTAA through the lens of a coalition realignment model. In the coalition realignment model, policy is determined by the coalitions formed among three legislative groups. Each of the three groups is easily identified in early 20th century American trade politics. The northern industrialized states made up an import-competing protectionist block. Southern cotton states comprised an export-oriented liberal block. The third, and pivotal, group was composed of the western grain belt states. The coalition realignment model hypothesizes that the shift to reciprocal liberalization occurred when the pivotal group abandoned its traditional coalition with the protectionist northeast to form a new coalition with southern cotton. As we shall see, there is clear evidence of this realignment. 1 Between the late 19th century and the early 1930s, the western grain belt aligned itself in a protectionist coalition with import-competing manufacturing industries. In the early 1930s, however, the grain belt abandoned this coalition and joined southern cotton in a coalition that supported reciprocal liberalization. It was this coalition realignment that initially pushed American trade policy on a liberal path. And it was this cotton-grain coalition that kept American policy on this path through the onset of World War II. The paper is organized as follows. The first section briefly summarizes the standard partisan and export-sector explanations of the change in American trade policy and develops the coalition realignment model. The subsequent section presents empirical evidence to evaluate the standard account against the coalition realignment model. I first examine Senate votes on trade legislation between 1922 and 1940 to evaluate the explanatory power of the exportsector and coalition realignment models. This examination reveals clear evidence of the grain belt’s realignment in 1934. I then examine statistical and qualitative evidence to evaluate whether the grain belt’s realignment was partisan or policy-oriented. The final section summarizes the evidence and highlights the paper’s broader implications for our understanding of American trade policy. B. Explaining the Shift in American Trade Policy: The Rise of the Export-Oriented Sector or Coalition Realignment? Explanations of the liberal turn in American trade policy are dominated by what I will call the two-sector model (see e.g., Bailey, Goldstein, and Weingast 1997; Epstein and O'Halloran 1996; Ferguson 1984; Frieden 1988; Gilligan 1997; Hiscox 1999; Irwin and Kroszner 1999; Schnietz 2000). This approach models trade politics as competition between an import-competing sector that always prefers protection and an export-oriented sector that always prefers trade liberalization. Each sector’s trade policy preferences are a function of its position in the global economy. The import-competing sector loses from trade because it produces goods with processes that rely heavily on society’s scarce factor of production. The export-oriented sector gains from trade because it produces goods with processes that rely heavily upon society’s abundant factor. Changes in trade policy, therefore, are caused by changes of the relative size (or political influence) of the export-oriented and importcompeting sectors. In the context of American trade policy, the shift from protectionism to reciprocal liberalization was caused by the growth of the export-oriented sector relative to import-competing producers. Two variants of this approach, one structural and one institutional, have been developed to explain the shift of American trade policy. The structuralist variant focuses on the transformation of the American economy and of American manufacturing production more narrowly. Manufacturing activity grew in importance steadily from the late 19th century onward, thus transforming the United States from a largely agricultural to an overwhelmingly industrialized economy. No less important, the nature of manufacturing production itself changed dramatically. Traditional labor-intensive methods of production increasingly were replaced by capital-intensive forms, and production increasingly shifted out of small enterprises and in to large corporations. Economic change in turn shifted the balance of power in trade politics, pushing trade policy onto a more liberal path. 2 Structuralists attribute the timing of the policy change, as well as its persistence into the postwar era, to the impact of large exogenous shocks on the American competitive position. As Frieden (1988, 60) argues, “the crisis of the 1930s and the eventual destruction of most of America’s overseas competitors led to an ‘internationalist’ victory that allowed for the construction of the American-led post-World War II international political economy.”1 Hiscox further develops this argument when he argues that the RTAA survived the return of Republican congressional majorities after World War II because the Republican electoral base had been divided by “the dramatic exogenous effects of World War II on U.S. export and import-competing industries” (Hiscox 1999, 670). The institutional variant attributes the liberal turn in American trade policy to institutional innovation that encouraged exporters to organize and pressure legislators for trade liberalization (Bailey, Goldstein, and Weingast 1997; Gilligan 1997; Irwin and Kroszner 1999). Although exporters shared a common interest in trade liberalization, they could not readily overcome the collective action problem they faced in order to transform this common interest into a new policy. When Congress set tariffs unilaterally, as they did prior to the RTAA, import-competing producers had incentive to organize to capture the concentrated benefits high tariffs provide. Because the cost of protection is distributed across a large number of consumers, it is difficult for those who lose from protection to organize to oppose tariff increases. Consequently, legislators faced strong pressure from import-competing producers for high tariffs, and little countervailing pressure from exporters for low tariffs. This caused a protectionist bias in congressional tariff politics. The innovation in the tariff-setting process embodied in the RTAA reduced the severity of this protectionist bias by encouraging exporters to organize and lobby. Once Congress delegated to the executive the ability to set tariff levels through reciprocal agreements, the tariff reductions that the executive could extract from foreign governments offered a concentrated benefit to domestic exporters.2 The relatively small number of homogeneous firms in each benefited industry could now more easily overcome the collective action problem to organize and lobby Congress. While import-competing industries did not disappear, greater lobbying activity by export-oriented producers changed the balance of power within Congress. Thus, shifting from a unilateral legislated tariff to a bilateral bargained tariff changed the size of the politically relevant export-oriented and importcompeting sectors. Because exporters could now more easily overcome the collective action problem and lobby Congress, legislators faced stronger pressure to liberalize trade. While both variants of the two-sector model offer compelling explanations of the durability of reciprocal liberalization, neither variant’s central explanatory logic explains the initial shift Irwin and Kroszner (1999) combine the exogenous shock and institutional innovation arguments to suggest that the changes wrought by World War II caused Republicans to become more responsive to export interests and thus more willing to support the RTAA. 2 Gilligan (1997) argues that because the executive is elected by a larger constituency than a legislator, it is more likely to support a liberal trade policy. He does not, therefore, rely on economic change to account for the shift in American trade policy. As Hiscox (1999) points out, however, prior to 1934 American presidents were not more liberal than Congress. Republican presidents were elected on and subsequently signed into law protectionist legislation. One might suggest, therefore, that structural change may have caused the election of an executive that represented the growing export-oriented sector, thus merging the structural and institutional variants of the two-sector model. 1 3 in policy. The structural variant, as noted above, focuses on structural changes caused by World War II, more than ten years after Congress first passed the RTAA. The institutional variant cannot explain how exporters who needed delegation in order to be induced into Congress in numbers sufficient to influence policy were provided legislation that promoted trade liberalization. Because neither variant can readily account for the RTAA in 1934, both have relied heavily on partisanship to explain passage of the initial legislation. The partisan hypothesis suggests that the RTAA passed in 1934 because the Democrats acquired a congressional majority in the 1932 elections. Hiscox (1999), for example, argues that the RTAA passed in 1934 because of the “Democratic majority with a core constituency of interests that favored more liberal trade policies…” Similarly, Bailey, Goldstein, and Weingast (1997, 317-318), emphasize the traditional Democrat commitment to free trade, but note that in the midst of the Great Depression unilateral tariff reductions held little appeal. Thus, an innovative approach such as the RTAA was required.3 Existing research has thus converged around a two-part explanation for the liberal turn in American trade policy. Partisanship, and specifically the congressional majority the Democrats won in the 1932 elections, explains the initial passage of the RTAA in 1934. The subsequent strengthening of the political influence of export-oriented producers, caused either by the institutional change itself or by the destruction of overseas competition in World War II, explains the durability of this policy shift. In contrast, the coalition realignment model suggests that the liberal turn in trade policy was caused by a realignment of congressional trade policy coalitions. The coalition realignment model focuses on interaction between three sectors, I, E, and P (Adserà and Boix 2002). As in the two-sector model, I is the import-competing sector which loses from trade and therefore always prefers high tariffs and E is the export-oriented sector which gains from trade and therefore always prefers low tariffs. P is the pivotal sector and holds conditional trade policy preferences. P’s preferences are conditional because the impact of trade on its income is “subject to considerable variation over time” (Adserà and Boix 2002, 232). In the original model, P’s income is a function of the international business cycle. When global demand is booming, P’s income rises and it thus gains from trade. When the global economy is in recession, P’s income falls and it thus loses from trade (Adserà and Boix 2002, 232). Whether trade policy is liberal or protectionist depends upon whether P enters a coalition with I or instead joins a coalition with E. Which coalition P enters will be determined by one of two conditions. The first is the state of the global economy. If the global economy is booming, P joins E and trade policy is liberal. If the global economy is in recession, P joins I and trade policy turns protectionist. Alternatively, E can entice P into a liberal coalition by promising to create government programs that protect P from the risk it faces in an open economy. If a compensation mechanism is created, then P enters a coalition with E and remains a part of that coalition even if the global economy moves into recession. 3 Bailey, Goldstein, and Weingast appear to assume that partisanship accounts for the initial legislation, and are more concerned to explain why Democrats were motivated to shift from unilateral tariff reductions to reciprocal liberalization. That is, why did the Democrats innovate? They attribute the innovative impulse to the desire to craft a more durable liberal trade policy on the one hand and the impact of the Great Depression, which weakened support even among Democrats, for trade liberalization on the other. 4 Applied to American trade policy, the coalition realignment model hypothesizes that from the late 19th to the middle of the 20th century American trade politics were shaped by coalitions established by three principal groups. Import-competing manufacturing, based largely in the northeast and eastern Great Lakes, voted regularly for protection. Exportoriented producers, composed primarily of southern cotton states, voted consistently for trade liberalization. The pivotal group was composed chiefly of western grain belt states struggling to maintain steady incomes in the face of unstable (and generally declining) terms of trade against manufactured goods (see Frieden 1997). Whether American trade policy was liberal or protectionist depended upon whether the grain belt aligned itself with northeastern manufacturing or southern cotton. Until the early 1930s, the grain belt aligned itself with northeastern import-competing manufacturing and voted for protection. Consequently, trade policy was protectionist. When it became clear that protectionism would not stabilize farm incomes, the grain belt began searching for new policies and for new alignments through which to enact them. As a result, it abandoned its traditional coalition with northeastern manufacturing and entered a new coalition with southern cotton. The coalition realignment model further hypothesizes that this cotton-grain coalition was based on a simple exchange. Southern cotton agreed to support federal policies to stabilize farm incomes and in exchange, the grain belt agreed to support reciprocal trade liberalization. In contrast to the two-part explanation found in existing work, therefore, the coalition realignment model emphasizes a single factor behind the initial shift to and the subsequent persistence of reciprocal liberalization. It de-emphasizes partisanship as the cause of the initial legislation, and it de-emphasizes the growth of export-oriented manufacturing as the cause of the RTAA’s persistence. Instead, it attributes both the initial legislation and its subsequent extension to the emergence of a new coalition in American trade politics. And it suggests that this new coalition emerged as a result of new policies that stabilized the incomes of the pivotal grain belt. We turn now to evaluate these alternative explanations. C. Explaining the Shift in American Trade Policy: The Export Sector, Partisanship, or Coalition Realignment? I evaluate the two-sector and coalition realignment models in two steps. I first examine Senate votes on trade legislation from 1922 to 1940 to evaluate which approach best explains the pattern of votes. Senate votes clearly show the western grain-belt abandoning its historical protectionist alignment with northeastern import-competing manufacturing and aligning with southern export-oriented cotton in 1934. The analysis reveals few signs of export-oriented support for the initial policy shift or for its subsequent extension. The analysis thus suggests that the policy turn was caused by a coalition realignment. I then evaluate whether this cotton-grain coalition was partisan, i.e., merely reflected Democrat victories in grain-belt states in the 1932 election, or whether it was non-partisan and policy-oriented. The three tests I present, two statistical and one case study, provide substantial evidence that the cotton-grain coalition was policy-oriented rather than partisan. All of the evidence points to the same conclusion: the initial adoption of, as well as extensions of the RTAA through the onset of World War II were caused by a realignment of congressional trade coalitions. 5 1. The Growth of Export-Oriented Manufacturing or Coalition Realignment? I first test two hypotheses about the change in trade policy. The two-sector model hypothesizes that this policy change was caused by the growth of the export-oriented sector relative to the import competing sector. The coalition realignment model hypothesizes that the shift was caused by a realignment of the grain-belt. I evaluate these hypotheses by examining Senate votes on trade legislation between 1922 and 1940. I first present descriptive data on Senate votes on this trade legislation. This analysis provides initial evidence of a coalition realignment by revealing a regional realignment in Senate trade voting in 1934. I then present the results of statistical analysis that tests the relationship between the economic characteristics of each state and Senate trade votes. This analysis confirms the patterns revealed by the descriptive analysis. The RTAA passed in 1934 and was extended it in 1937 and 1940 because the grain-belt abandoned its traditional protectionist alignment with import-competing manufacturing industry and aligned itself with export-oriented southern cotton. a. Regional Alignments and Realignments in Senate Trade Votes, 1922-1949 Table 1 presents Senate votes for the 1922 and 1930 trade acts, the 1934 RTAA, and two critical amendments proposed in subsequent RTAA extensions. I focus on amendments rather than the final bill for the 1937 and 1940 votes because these amendments more accurately reflect the political struggle over trade policy than the final bill (Krehbiel and Woon 2005).4 Both amendments were non-trivial attempts to deny the executive the authority to negotiate reciprocal agreements without defeating the RTAA itself. The Pepper Amendment of 1937 sought to prevent tariff reductions on agricultural or horticultural commodities below the level that would “at least equalize the difference in the cost of production” between the foreign country and the United States.”5 The 1940 Pittman Amendment proposed that all RTAA agreements be submitted to the Senate for ratification by a two-thirds majority.6 Either amendment, if passed, would have ended the RTAA program, without expressly refusing to renew the authority. The vote on both amendments was substantially closer than the vote on the final bill. The Pepper Amendment was defeated by a 42 to 39 vote, while the final bill was passed by a 58 to 24 vote. 7 The Pittman 4Krehbiel and Woon (2005) develop a more general argument for focusing on amendments rather than final bill votes. 5The Pepper Amendment: “Provided, however, that in the negotiation of any new agreement or agreements, or in the renewal or extension of any existing agreement or agreements, under the authority of this act the tariffs or import duties upon all agricultural and horticultural products shall be maintained at a point which will at least equalize the difference in the cost of production thereof in the country or countries dealt with and the United States as determined by the United States Tariff Commission as of the date any such new or extended agreement becomes effective.” Congressional Record. 1937. Proceedings and Debates of the First Session of the SeventyFifth Congress of the United States of America. Volume 81 Part 2 February 15, 1937-March 17, 1937, page 1606. 6 The Pittman Amendment: “No foreign trade agreement hereafter entered into under section 1 of this act shall take effect until the Senate of the United States shall have advised and consented to its ratification, two-thirds of the Senators present concurring.” Congressional Record. 1940. Proceedings and Debates of the 76th Congress Third Session Volume 86 Part 4, March 27, 1940- April 17, 1940. page 3657. 7 The Pepper Amendment was actually passed on a first vote. It was then reconsidered after Senator Harrison asserted that he had voted for the amendment “by duress” (Congressional Record. 1937. Proceedings and Debates of the First Session of the Seventy-Fifth Congress of the United States of America. Volume 81 Part 2 February 15, 1937March 17, 1937, page 1611). The vote to reconsider the amendment passed 43 to 37, (Ibid., page 1611). The amendment itself was defeated in the second vote. 6 Amendment was defeated by a 44 to 41 vote, while the 1940 extension was passed by a 42 to 37 vote. Votes on these two amendments therefore provide a more accurate portrait of Senate voting patterns than the votes on the final bills. I group the votes by census region rather than states to simplify presentation and discussion. We look first at the two interwar tariff acts, the 1922 Fordney-McCumber and the 1930 Smoot-Hawley Acts. Because both acts are considered protectionist, they provide a useful baseline against which to identify the changes that shifted American trade policy onto a liberal path. A regional alignment is clearly evident in these two votes. Southern states were solidly liberal in both votes, voting by a very large majority against each of these tariff acts. In contrast, the northern industrialized regions (Middle Atlantic, North East, and East North Central) as well as the western regions are solidly protectionist, voting for both of these tariff acts by very large majorities. (Table 1 About Here) We turn now to the 1934 vote on the RTAA. The south remained solidly liberal, contributing 19 votes for the RTAA. In fact, in the South Atlantic region, only the northernmost states (Delaware and Maryland) and Carter Glass (Senator from Virginia) voted against the RTAA. We also see that the more competitive position of capital-intensive manufacturing has caused the industrial regions to become less solidly protectionist. Yet, the industrial states do not move solidly into the pro-liberalization bloc. Instead, they cast a majority vote against the RTAA. Moreover, the more industrialized states within these regions tended to oppose the RTAA (see table 2). In the Middle Atlantic region, Pennsylvania and New Jersey contributed four votes against the RTAA while New York contributed two in favor. New England states remain solidly protectionist, with only Massachusetts shifting into the liberal block. The biggest vote shift occurs in the East North Central region, and is most clearly evident in the agricultural states of Illinois and Wisconsin. All other states in the region split their votes. There is little sign that the industrialized states in 1934 had abandoned protectionism in favor of reciprocal liberalization. (Table 2 About Here) The most dramatic changes occur in the western regions. Whereas the western regions provided an average of 25 protectionist votes in 1922 and 1930 against only 12 liberal votes, this overwhelming majority for protectionism was completely reversed in 1934. In 1934 the western regions provided 29 votes for the RTAA and only 15 against. This shift of western votes is the first clear piece of evidence of a coalition realignment. Western states abandoned their historical protectionist alliance with the northeast and aligned themselves with the liberal south. This western realignment in turn provided the votes required to pass the RTAA in 1934. In fact, the 29 western votes for the RTAA combined with the 19 southern votes yielded 48 votes for the RTAA—exactly 50 percent of the Senate. Of course, this is not a direct test of the two-sector model, as both variants of this approach suggest that the RTAA passed in 1934 because of the Democrat majority. Exporter support for reciprocal liberalization emerged only after 1934. Moreover, the structuralist variant suggests that export support emerged after World War II, a period beyond the temporal scope of this study. We can evaluate the institutionalist variant, however, for it suggests that 7 exporter support emerged after the initial institutional change. Moreover, Gilligan (1997) found export-sector support in the 1937 vote to extend the RTAA. Yet, Senate votes on RTAA extension through 1940 provide little indication that Senators from industrial regions joined the RTAA block. Instead, the split evident in 1934 persists through 1940. Industrial regions voted against the 1937 amendment, and thus for the original intent of the RTAA, by a small majority of 13 to 10. Yet, even this small majority for the RTAA had evaporated by 1940, as the industrial regions voted for the 1940 Pittman amendment by a large majority. Nor do we see much evidence of support if we look at the individual state votes in 1937 and 1940 (table 3). In 1937, it was only four pro-RTAA votes cast by Illinois and Indiana that produced the small majority. In 1940, it was only New York and Indiana that prevented the anti-RTAA majority from being even larger. Senate votes on the RTAA through 1940 thus provide little evidence that manufacturing exporters were pressuring their Senators to support the RTAA. (Table 3 About Here) Descriptive analysis of Senate trade votes thus provides initial evidence that the RTAA passed in 1934 because of a realignment in congressional trade coalitions. Prior to 1934, western states aligned with the traditionally protectionist import competing industrial regions and voted for protection. In 1934, western states abandoned this traditional coalition and aligned themselves with the traditionally liberal southern states and voted for trade liberalization. Consequently, trade policy shifted from protectionism to the new policy of reciprocal liberalization. b. Statistical Analysis Statistical models of these same Senate trade votes confirm the pattern revealed by the descriptive analysis. I estimated logit models for each of the trade votes examined above. I coded a vote for trade liberalization (e.g., a vote against the Smoot-Hawley Act, for the 1934 RTAA and its extensions) as 1 and a vote for protectionism as 0. Previous statistical analysis of trade votes has regressed votes against estimates of the size of each state’s export-oriented and import-competing sectors (see e.g., Gilligan 1997; Irwin and Kroszner 1999). Such estimates are typically constructed by weighting each sector's share in total state tradeable output by the ratio of total U.S. exports and imports to national output in that sector. The resulting estimates are then used as proxies for the size of each state’s import-competing and export-oriented industries. This approach assumes that Senate trade votes align along the two-sector divide. Consequently, such measures do not allow me to evaluate the relative power of the twosector and coalition realignment models. Thus, I opted to employ direct measures of each state’s economic structure rather than estimates of each state’s import-competing and export-oriented sectors.8 Each independent variable thus measures a single industry’s percentage share of each state’s total tradeable output. I include such measures for eight 8 The economic data is the same as used by Irwin and Kroszner (1999) and I thank Doug Irwin for graciously providing this data in a machine readable format. State-level data on value added by industry are from the U.S. Department of Commerce, Bureau of the Census, Biennial Census of Manufactures, 1935 (1938). State-level data on mining income are from U.S. Department of Commerce, Bureau of Economic Analysis, State Personal Income (1984). 8 non-agricultural industries (transportation, iron and steel, machinery (including electrical), textiles and apparel, miscellaneous manufacturing, chemical, petroleum, and mining. The four agricultural variables (Cotton, Sugar, Wheat, and Corn) are dummy variables that take the value of 1 if the state in question is (a) predominantly an agricultural producer (data on state economic output indicate little non-agricultural traded goods production) and (b) the commodity in question was the most important crop for that state. In such cases, the state was coded 1. All other observations were coded 0. I relied on separate dummy variables rather than a single measure of agricultural production as a share of state output because there are substantial regional differences in agricultural production. Western production was concentrated in either wheat or corn, southern states grew cotton, while northeastern agriculture focused more heavily on dairy products, few of which were exported. It seemed more reasonable to assume that these different commodities were different than to assume that they were the same. The economic data are from 1935. I omit a measure of partisanship for now and focus only on state economic characteristics. I examine the partisan hypothesis in detail in the next section. All models are estimated with Stata’s logit procedure. I employed the cluster option to correct the standard errors for the lack of independence between observations from the same state. The results are presented in table 4. (Table 4 About Here) The test statistics indicate that the models fit the data relatively well. The log-likelihood ratios for all models are significant, and the pseudo R2 suggest that the models account for a reasonably large proportion of the “variance” present in Senate votes. Notice also that with only two exceptions, there are few stable patterns across time. Coefficients change in size, many also change sign, and there are few variables that consistently return statistically significant coefficients in all of the models. Cotton is one clear exception. Southern cotton provided a stable source of pro-liberalization votes throughout the entire period. Petroleum producing states provide the second exception. States in which petroleum was an important share of total output were solidly protectionist through the end of World War II. The models also indicate that states dominated by manufacturing production were neither solidly pro-liberalization nor staunchly protectionist. The absence of statistically significant coefficients on most of the manufacturing industry variables indicates that states in which these industries provided large shares of total output were divided on the RTAA. This may reflect an unwillingness of even the more competitive industries to open the American market in the midst of the Great Depression, or it might reflect the fact that importcompeting and export-oriented manufacturing production were both concentrated in a few industrial states. Forced to represent both sectors, Senators from these states were unable to move decisively in either direction.9 Nor does the analysis suggest that export-oriented manufacturing industries became more willing to support reciprocal liberalization as the decade progressed. Senators from states in 9 I also estimated versions of the model using an aggregate measure of manufacturing production rather than the individual industry measures. The aggregate measure also failed to return a statistically significant coefficient. 9 which the three industries that accounted for half of all manufactured exports during the pre-depression era—automobiles, iron and steel, and machinery—were no more likely to vote for trade liberalization in 1937 and 1940 than they had been in 1930 and 1934. Moreover, states in which the transportation industry was important were more likely to vote against than for reciprocal liberalization in 1937 and 1940.10 There is some indication that states in which the chemical industry as well as the textile and apparel industries were important were more willing to support reciprocal liberalization in 1937 than they had been in 1934, but only the chemical industry continues this support through 1940. Overall, therefore, the statistical models provide little evidence that the shift in policy was driven by export-oriented manufacturing, or that the policy change itself encouraged such producers to pressure their Senators to vote for its continuation. In contrast, the models provide substantial evidence of a coalition realignment. The models indicate quite clearly that the western farm and mining states abandoned their previous alignment with protectionist manufacturing industries to form a new coalition with proliberalization southern cotton in 1934. Initially, this realignment was driven by Senators from the wheat belt, which moved firmly into the pro-liberalization group in 1934. Once these states began to migrate out of the pro-liberalization block during the late 1930s, Senators from the Corn Belt took their place. An identical shift occurs in the western mining states, which also abandoned its protectionist alignment of the pre-New Deal era to join the proliberalization block for the 1934 and 1937 votes. The models also indicate that the realignment was not fully embraced by the grain-belt. While the wheat belt strongly supported the RTAA in 1934 it opposed its extension in 1937 and 1940 just as strongly. Their shift out of the pro-RTAA block was a consequence of the agreements that were negotiated, which the wheat belt viewed as highly disadvantageous to wheat and cattle ranchers.11 The corn-belt moved into the pro-RTAA block beginning only in 1937. Yet, in spite of these perturbations, the models provide fairly clear evidence that the grain-belt did in fact move away from its traditional alliance with northeastern protectionists and into the pro-liberalization camp in 1934. These results appear to be inconsistent with Gilligan’s (1997) results based on estimates of each state’s import-competing and export-oriented sectors. I think, however, that this inconsistency is more apparent than real. Gilligan (1997) finds that states with large exportoriented sectors were more likely to vote for trade liberalization after 1934 than they had been prior to 1934. The findings reported here suggest exactly this change. Votes by Senators representing western farm states were votes by export-oriented states. Until 1934, however, Senators from western farm states voted for protection, thereby splitting the vote of two of the largest American exporters—cotton and grain. Consequently, a measure of state export orientation would be unlikely to return a statistically significant coefficient for 10 I also estimated versions of the model in which these three export industries were aggregated into a single measure, and in which machinery and transportation were combined into a single measure. These aggregated measures also failed to return statistically significant coefficients. 11Senator Arthur Capper (Rep. Kansas) remarked in the Senate debate on the 1937 extension that he had supported the RTAA in 1934 in the hope that it would be beneficial to agriculture. In fact, he argued, it had harmed agriculture; imports of many commodities had risen sharply, while exports had not kept pace. Congressional Record. 1937. Proceedings and Debates of the First Session of the Seventy-Fifth Congress of the United States of America. Volume 81 Part 2 February 15, 1937-March 17, 1937, page 1511. 10 votes prior to 1934. In the 1934 and 1937 votes, however, (the two RTAA votes that Gilligan models) cotton and grain states both voted for trade liberalization. Consequently, a measure of state export-orientation would be likely to return a statistically significant coefficient. If so, then the results of my analysis are consistent with Gilligan’s findings. The broader point is that the two results are consistent because both highlight that the export-oriented sector became more willing to support trade liberalization in the 1930s than it had been previously. My results differ from Gilligan’s by suggesting that the cause of the export sector’s support for trade liberalization was not a general improvement of American manufacturing competitiveness, but the realignment of western farm states. The RTAA was passed and subsequently extended because Senators from western states, representing export-oriented agriculture, (and supported by western mining states), abandoned their historical alignment with import-competing manufacturing and entered a new liberalizing coalition with southern cotton. 2. Party or Policy? Evaluating the Cause of the West’s Realignment While the previous analysis highlights the emergence of the cotton-grain coalition and demonstrates that this coalition provided the votes needed to pass the RTAA, it does not explain why this coalition emerged. Partisanship is the explanation most commonly advanced to account for the shift from protectionism in 1930 to reciprocal liberalization in 1934 (see e.g., Bailey, Goldstein, and Weingast 1997; Gilligan 1997; Hiscox 1999; Pastor 1980; Schnietz 2000). The partisan hypothesis suggests that the western realignment identified above reflects the Democrat congressional majority won in 1932. And indeed, much of the Senate majority that the Democrats enjoyed in 1932 was acquired by winning seats west of the Mississippi. In fact, sixteen of the nineteen Senate seats that the Democrats picked up at the expense of Republicans between the 1930 Smoot-Hawley and the 1934 RTAA votes were in states west of the Mississippi.12 Moreover, eleven of these sixteen seats were in western farm states that had been staunchly Republican since the 19th century. Thus, partisanship is a plausible explanation of the formation of the cotton-grain coalition. The Democrats became the majority party because the Midwest elected Democrat Senators. The Democrats used their majority to enact their traditional low-tariff policy. The coalition realignment model emphasizes a policy-oriented alliance rather than partisanship. Southern cotton states and western farm states voted together for the RTAA not because they were Democrats, but because they joined forces to achieve mutually agreed policy objectives. This alliance might have been centered on a common interest in a more liberal trade policy, or it might have been based on a logroll in which western states voted for a liberal trade policy in exchange for southern votes on an issue (or issues) important to the west. In either case, the coalition was non-partisan and policy oriented. I present three tests of these two hypotheses. First, I include a Party variable in the statistical models reported above. If the results are robust to the inclusion of Party, we have evidence that the coalition realignment was an important cause of the shift of trade policy. Second, I 12 Between the Smoot-Hawley and RTAA votes, Democrats gained Senates seats at Republican expense in California, Colorado, Connecticut, Kansas, Indiana, Idaho, Illinois (2), Massachusetts, Nebraska, Nevada, Oklahoma, South Dakota, Utah (2), Washington, Wisconsin, and West Virginia. 11 correlate votes on issues known to be important to western farm states during the 1920s with the 1934 RTAA vote and a 1934 vote on silver purchases. If the correlation between votes in 1928 and 1934 is high, we have evidence that the coalition that voted to pass the RTAA was voting together long before its members belonged to the same party. This would support to the policy-oriented coalition hypothesis. Finally, I present case study evidence on the formation of the cotton-grain coalition. This evidence should reveal in detail whether a cotton-grain coalition was explicitly established as well as reveal the details about the central policy issues upon which it was based. a. Partisanship and the 1934 Vote on the RTAA Table 5 presents the re-estimated logit models, now controlling for party membership. Party is a dummy variable coded 1 if the Senator is a member of the Democratic Party and 0 if the Senator is a member of the Republican Party. The models indicate a somewhat surprising pattern that weakens a partisan explanation and strengthens a coalition realignment explanation. First, party membership had an important impact on the 1930 and 1934 votes. Moreover, the positive coefficients for Party on both votes indicate that Democrats were more likely to vote for a liberal trade policy than Republicans. In addition, inclusion of the Party variable substantially improves the explanatory power of the model as a whole relative to the models that incorporated only economic characteristics. The pseudo R2 is substantially larger in both instances. (Table 5 About Here) In contrast, Party did not have a statistically significant impact on the two votes to extend the RTAA. In fact, in these two votes Party does not approach conventional measures of statistical significance. This suggests that the 1937 and 1940 votes were not along party lines, and that both parties were deeply divided on the RTAA. Raw vote totals support this interpretation. Democrats split their vote, with 24 voting against and 34 voting for continuation of the RTAA in 1937. The split was similar in 1940, with 23 voting against and 39 voting for continuation.13 A majority of Republicans voted against continuation in both years, by 15 to 8 in 1937 and by 18 to 5 in 1937. Thus, while partisanship influenced the 1934 vote, it was not a critical factor in the extensions. Equally important, however, each of the individual components associated with the coalition realignment remain statistically significant. Cotton is a partial exception, as it is significant only in 1930 and 1940. In the 1934 vote this is most likely due to multicollinearity between Democrat and Cotton—all Senators from cotton states in 1934 were Democrats. In the 1937 vote it most likely reflects a split among the cotton states. The raw vote totals show that the cotton states voted for reciprocal trade by an unusually small 8 to 6 majority. The other patterns remain largely unchanged. Wheat returns a significant and positive coefficient in 1934 but not in 1930. The Corn Belt returns a significant coefficient in 1937 and 1940. Western mining also continues to return a statistically significant coefficient in the 1934 vote. Thus, the models continue to reveal evidence of the grain-belt’s realignment. In contrast, Democrats voted for the RTAA in 1934 by a margin of 55 to 5. Republicans voted against by a margin of 30 to 6. 13 12 The models also portray a manufacturing sector deeply divided over the RTAA. The transportation industry is revealed as a supporter of liberalization in the 1934. However, this support is transient, as these same states are more likely to oppose than support continued liberalization in 1937 and 1940. Miscellaneous manufacturing also opposes reciprocal liberalization in 1937. The chemical industry remains a consistent supporter of reciprocal liberalization, while the weak support of the textile and apparel industry is also evident in 1937. Thus, a few of the trends that the two-sector model hypothesizes emerge. Yet, the majority of the manufacturing industry variables are not statistically significant, and many that are statistically significant have negative signs. In short, even once we control for party membership we continue to see substantial evidence of a coalition realignment and few signs that export-oriented manufacturing industry supported the RTAA before World War II. As a second test of the partisanship and policy-oriented hypotheses I correlated votes on the 1934 RTAA with votes in the late 1920s on issues important to the western farm states. No issue was more important to western farm states during the 1920s than the McNary-Haugen movement. The McNary-Haugen movement emerged in the early 1920s as an attempt to secure government programs that would guarantee farm incomes equal to incomes in manufacturing (see Black 1928). It hoped to raise the purchasing power of farm incomes to the level that had prevailed during the ten relatively prosperous pre-war years (1904-1914). Parity would be maintained through a tariff that separated the high domestic price from the low world price. This higher domestic price would be supported by government purchases of production in excess of domestic consumption. The resulting surplus would be dumped in world markets at world prices. The gap between the domestic price and the world price would be financed by taxing each unit of the commodity (e.g., each bushel of wheat or corn) produced. McNary-Haugen legislation was introduced in Congress in 1924, 1926, 1927, and 1928. It failed to pass in 1924 and 1926. It then passed by small majorities in the House and the Senate in 1927 and 1928, but both bills were vetoed. The correlations between Senate votes on the 1927 and 1928 McNary-Haugen Bills on the one hand and the 1934 RTAA on the other are presented in table 6. Notice first the absence of a statistically significant correlation between the 1927 McNary-Haugen vote and the 1934 RTAA vote. In contrast, the correlation between the vote on the 1928 McNary-Haugen Act and the 1934 RTAA is large and statistically significant. This suggests that the cotton-grain coalition first emerged in 1928. The strength of the correlation between the 1928 and 1934 votes is remarkable. Eighty percent of the Senate seats that voted for the McNary-Haugen Act also voted for the RTAA. Sixty-five percent of those voting against McNary-Haugen also voted against the RTAA. The Senators who voted for both pieces of legislation came almost exclusively from southern and western states. The correlation between these two votes indicates that in 1928 Republican Senators from the grain belt voted alongside Democrats from cotton states. The cotton-grain alliance was therefore non-partisan and in place long before the 1932 election. (Table 6 About Here) As a further test I correlated the 1934 RTAA vote and the 1928 McNary-Haugen vote with another Senate vote that occurred in June of 1934. On June 4, 1934, only hours after having passed the RTAA, the Senate began considering legislation requiring the government to increase the proportion of silver relative to gold it held in its reserves. The Senate passed this 13 legislation on June 11, 1934 by a vote of 54 to 25.14 The correlations between the 1928 McNary-Haugen Act and the 1934 Silver Acts, as well as between the RTAA and the Silver Act are reported in table 7. The correlations among votes on all three bills are extremely high, providing further evidence of a policy-oriented coalition among southern cotton, western farmers, and western miners. The correlation between these votes also provides some insight into why western mining states were enticed away from their alliance with the protectionist block to support the RTAA in 1934. (Table 7 About Here) Statistical evidence thus provides substantial support for the assertion that the cotton-grain coalition was policy-oriented rather than partisan. Evidence of the western realignment found in the original logit models is robust to the inclusion of party membership. Moreover, the split in Democrat votes on the 1937 and 1940 extensions reveals that the Democrats were not universally supportive of the continuation of the RTAA. The correlations between Senate votes on McNary-Haugen legislation in 1928 and the RTAA and silver purchase legislation in 1934 demonstrates that the cotton-grain coalition was established well before the 1932 elections. These correlations also hint at the underlying policy bargain upon which the cotton-grain coalition was based. Southern cotton agreed to support the grain belt’s effort to secure legislation that would raise and stabilize farm incomes. In exchange, the grain-belt agreed to support the south’s bid for a more liberal trade policy. b. The Formation of the Cotton-Grain Coalition: Case Study Evidence A short case study tracing the emergence of the cotton-grain coalition can provide confirming evidence of, and fill in many of the details about, the coalition realignment that shifted trade policy. The cotton-grain coalition emerged in a two-stage process beginning in the late 1920s. The first step in its formation came in 1926-27 when southern cotton joined the grain belt to support McNary-Haugen legislation. The McNary-Haugen movement emerged in the early 1920s as commodity prices collapsed after the wartime boom. Initially, grain states responded to collapsing prices by supporting higher tariffs on agricultural commodities. They gained an Emergency Tariff Act in 1921, which raised tariffs on agricultural commodities, and these higher tariffs were made a permanent part of the schedule in the 1922 Fordney-McCumber Tariff Act. By 1923, farm organization leaders were recognizing that tariffs would not raise prices for the commodities that the U.S. exported. They began searching for measures that would “make the tariff effective” for agriculture. The McNary-Haugen program became the favored approach.15 The failure to pass McNary-Haugen legislation in 1924 and 1926 was due to the split between the two dominant agricultural sectors—western grain and southern cotton. Until “Silver Bill Passed by Senate, 54 to 25,” The New York Times, June 12, 1934, page 1. Peek is generally seen as the “father” of the farm parity movement. In 1921 he authored the pamphlet that spelled out the broad policy ideas (Equality for Agriculture). He campaigned throughout the 1920s on behalf of this approach. While initially a Republican, Peek grew disenchanted with the Republican Party’s unwillingness to embrace farm support and began to support the Democrats. He was appointed by Roosevelt to administer the AAA in 1933, and then in 1934 was appointed as Roosevelt’s special advisor on trade where he fought Cordell Hull to control the reciprocal trade policy that the RTAA made possible. Fite (1954) offers a compelling portrait of Peek as well as a detailed discussion of the McNary-Haugen movement. 14 15George 14 1927, southern cotton was unwilling to support the McNary-Haugen program. Its refusal to do so has been attributed to two considerations. On the one hand, about 50 percent of the cotton produced in the U.S. at that time was exported, compared to only 15 to 20 percent of American wheat production. Consequently, any system that financed the gap between domestic and world prices by taxing agricultural producers would provide substantially less benefit for cotton than it would provide for wheat (Bensel 1984). Cotton producers preferred instead an orderly marking arrangement in which the government purchased “surplus” cotton at an established floor price and sold when supply shortages arose. On the other hand, cotton prices held up well until 1926. In fact, in 1924 and early 1925 cotton was selling above the average price for the 1904-1914 period (Bensel 1984, 142; Fite 1954, 170). Consequently, cotton growers did not perceive an urgent need for price support mechanisms. Southern cotton had become more receptive to cooperation with the grain belt by late 1926. In part this more cooperative spirit reflected a sharp fall in cotton prices. A bumper crop in 1925 depressed cotton prices, and a second large harvest in 1926 added downward pressure. By 1926 cotton was selling at about half of its 1924 price.16 Falling prices led cotton farmers to reconsider the utility of the stabilization measures embodied in the McNary-Haugen program.17 In addition, as western states recognized that passing McNary-Haugen legislation required southern cotton’s support, they became increasingly willing to alter the details of the legislation to meet cotton’s concerns. Finally, it appears that the western states agreed to support southern efforts to promote the development of Muscle Shoals as a source of inexpensive fertilizer (Bensel 1984, 142-3). While the relative importance of these various factors remains unclear, what is certain is that in a meeting in November 1926 southern cotton agreed to cooperate with western grain in order to push the farm relief program through Congress. With the cotton-grain coalition in place, McNary-Haugen legislation passed through both houses of Congress in 1927 and 1928.18 The correlations reported above suggest that Southern support was moderate in 1927, and strengthened by 1928. Calvin Coolidge vetoed the first; Herbert Hoover vetoed the second. In neither case could the cotton-grain coalition override the presidential veto. Farm relief did not become a reality, therefore, until 1933. During the 1932 campaign Franklin D. Roosevelt met with farm organizations and promised to implement whatever policy they agreed to support. In exchange, the farm organizations used their influence in the grain belt to help the Democrats win the 1932 election. Once elected, Roosevelt moved quickly and in May 1933 Congress passed, and Roosevelt signed into law, the Agriculture Adjustment Act (AAA). Containing many of the central elements of the McNary-Haugen legislation, plus concerted efforts to limit production by requiring farmers to take some land out of production, the AAA passed through both houses of Congress on the strength of the The average cotton price in 1924 was 22.9 cents per pound. The price fell to 19.6 cents per pound in 1925 and then to 12.5 cents per pound in 1926. It recovered sharply in 1927 and then declined to less than 10 cents per pound by 1930 (Statistical Abstract of the United States, 1935, GPO, page 644. 17 “Southern farmers and farm leaders…began to say that if the McNary-Haugen Bill had been in operation, the calamity might have been avoided” (Fite 1954 170). 18 Three weeks after passing the 1928 legislation, Congress passed legislation “proposing a public corporation for the Muscle Shoals complex” (Bensel 1984, 146). 16 15 cotton-grain coalition (Campbell 1962, 66).19 By mid-1933, therefore, western grain had gained the protection that it had been seeking for more than ten years. The second step in the formation of the cotton-grain coalition immediately followed the enactment of the AAA. In this second step the grain-belt abandoned its traditional protectionist alliance with the northeast and voting with southern cotton for the RTAA. The shift of the grain belt out of the protectionist block and into the RTAA block is most clearly visible in the evolution of the American Farm Bureau Federation’s (AFBF) trade policy position. The AFBF was one of the largest and most influential agricultural associations in the nation. It had played a leading role in the McNary-Haugen movement and had been the central organization within which George Peek developed and lobbied for this program. Moreover, it believed that it had directly contributed to Roosevelt’s election in 1932 (by campaigning for him in Midwestern states) and it believed that it had shaped the specific details of the 1933 AAA (Campbell 1962, 63). The AFBF, therefore, embodied the central current of organized agriculture’s policy orientations. During the 1920s, the AFBF had supported a high tariff policy. It had pushed for the 1921 Emergency Tariff Act and the Fordney-McCumber duties; it had endorsed Hoover’s proposal for higher tariffs in 1929 and it had supported the high tariffs embodied in the Smoot-Hawley Act in 1930 (Campbell 1962, 141-142). Once the AAA was enacted in 1933, however, the AFBF began to move away from protectionism and toward support for reciprocal trade liberalization. In 1933, the AFBF board of directors did not strongly endorse the reciprocal trade policy then being discussed within the administration, but it did agree to focus its opposition on only those “reciprocal trade agreements that do not adequately safeguard the interests of agriculture” (Campbell 1962, 143). By early 1934 the AFBF had moved one step closer to endorsing fully the reciprocal trade program. In March 1934, the board of directors voted to support the RTAA “if and when the proper assurance is given by the Administration that the rightful interest of agriculture will be fully protected in the administration of such policy” (Campbell 1962, 143-4). Some evidence indicates that the Administration—or at least the State Department—provided such assurance to individual Senators, thereby encouraging the grain belt to vote for the RTAA in 1934.20 By the end of 1934, the board had fully endorsed the RTAA, stating that “we are in accord with the purpose of reciprocal trade treaties” (Benedict 1953, 342). The cotton-grain coalition was thus explicitly established in a two-step movement. During the late 1920s southern cotton joined the grain belt to support government policies that would raise and then stabilize farm incomes relative to manufactured goods. Once these 19 In the Senate, the South voted for the AAA by a 20 to 2 margin; the Midwest by a slightly smaller 15 to 7 majority. The northeast, in contrast, voted against the AAA 11 to 6 (Campbell 1962, 66 based on AFBF records). 20 Senator Joseph O’Mahoney (Dem., Wyoming) stated during Senate debate on the 1937 extension that he had visited the State Department prior to the 1934 vote and believed that he had been assured that agricultural commodities would not be a main focus for tariff reductions. It is unclear if he was unique in seeking such assurances, or whether other Senators from the grain belt also met with State Department officials. Congressional Record. Proceedings and Debates of the First Session of the Seventy-Fifth Congress of the United States of America. Volume 81 Part 2 February 15, 1937-March 17, 1937. 16 policies were in place, the grain belt abandoned its traditional protectionist alliance and joined southern cotton to support reciprocal trade liberalization. D. Conclusion The initial shift in American trade policy, as well as subsequent extensions of the RTAA through the onset of World War II was caused by a realignment in congressional trade politics. From the late 19th century until the early 1930s, the western grain belt aligned itself with northeastern import-competing manufacturing in support of high tariffs. In some respects the protectionist orientation of the grain belt made little sense. Grain producers were major exporters, and thus stood to gain from open foreign markets, and they were consumers rather than producers of manufactured goods, and thus would benefit from lower American tariffs on manufactured goods. Thus, it is surprising to find the grain belt favoring protectionism over a liberal trade policy in this period. In other respects, however, the protectionist alignment made sense for the grain belt. Support for protection was to a large extent based on the grain belt’s hope that higher tariffs on agricultural commodities would stabilize farm incomes in a world of falling commodity prices. With the grain belt firmly aligned with import-competing manufacturing, trade policy was staunchly protectionist. Then in 1934, the grain belt abandoned its traditional alliance with protectionist manufacturing to align with southern cotton in support of reciprocal trade liberalization. This realignment, and the cotton-grain coalition that it produced, was the reason why the RTAA was passed in 1934 and extended in 1937 and 1940. The grain belt’s realignment was in turn a product of the enactment of federal policies that stabilized farm incomes. The cotton-grain coalition first emerged in the late 1920s when the grain belt recognized that it needed the votes of southern cotton in order to pass McNaryHaugen legislation. And while the cotton-grain coalition was able to pass McNary-Haugen legislation in the late 1920s, it was not large enough to override presidential vetoes. Consequently, it was only in the first year of the Roosevelt administration that federal policies designed to stabilize farm incomes were enacted. Once such policies were in place, the grain belt became willing to abandon its traditional protectionist alliance and support southern cotton’s pursuit of a more liberal trade policy. The grain belt’s support for trade liberalization was always conditional, however. The grain belt did not support free trade, but viewed the RTAA as a means of opening foreign markets to surplus American agricultural production in exchange for opening the American market to foreign manufactured goods. When the early agreements that Cordell Hull negotiated appeared to provide an “unfavorable” exchange by opening the American market to foreign agricultural commodities in exchange for access to foreign markets for American manufacturing, the grain belt’s enthusiasm for reciprocal liberalization weakened substantially. The evidence presented here challenges us to rethink the consensus view of the political determinants of the shift in American trade policy. It is reasonably clear that we should consider jettisoning the partisan explanation of the initial 1934 policy shift altogether in favor of the coalition realignment explanation developed here. There is too little evidence to sustain the claim that the 1934 vote was simply a result of a Democrat majority pursuing its traditional free trade policy and too much evidence of the cotton-grain coalition to sustain a partisan explanation. We should be encouraged by this, as shifting from a partisan to the coalitional explanation enables us to replace an explanation based on non-economic 17 considerations (why were Democrats pro-trade, anyway?) with an explanation firmly grounded on economic interests. The Senate voted for the RTAA in 1934 because cotton and grain states both thought it would open foreign markets for the commodities they produced. That most of these Senators were also Democrats is not terribly important. Somewhat less clear is what the evidence presented here implies for the two variants of the two-sector model. On the one hand, this paper has not presented evidence on postwar trade politics. Consequently, it cannot draw conclusions about how well the two-sector model explains the evolution of trade policy after World War II. This is an important limitation, as both variants of the two-sector model emphasize changes in trade politics that unfold gradually. Yet, the evidence presented here does encourage some skepticism about the twosector model’s explanation for the early postwar period. There is little evidence that the industrialized states had become more supportive of the RTAA by the eve of World War II. A fresh look at the early postwar trade politics may be useful. The evidence must also make one wonder if it is merely coincidental that the two groups that passed and then sustained the RTAA prior to the war received the most significant exemptions from GATT-based trade liberalization. Agriculture was excluded from GATT entirely, while textile and apparel, located largely in the southern cotton states, was placed within a quota-based regime. Thus, the coalition realignment model may highlight the central political bargains at the base of American participation in postwar trade liberalization. Of course, this paper can only raise these questions. Answers to them will require additional research. 18 References Adserà, Alicia, and Carles Boix. 2002. Trade, Democracy, and the Size of the Public Sector: The Political Underpinnings of Openness. International Organization 56 (Spring):229262. Bailey, Michael, Judith Goldstein, and Barry Weingast. 1997. The Institutional Roots of American Trade Policy: Politics, Coalitions, and International Trade. World Politics 49 (April):309-39. Benedict, Murray R. 1953. Farm Policies of the United States 1790-1953. New York: Twentieth Century Fund. Bensel, Richard F. 1984. Sectionalism and American Political Development: 1880-1980. Madison: University of Wisconsin Press. Black, John D. 1928. The McNary-Haugen Movement. American Economic Review 18 (September):405-427. Campbell, Christiana McFadyen. 1962. The Farm Bureau and the New Deal. Urbana: University of Illinois Press. Epstein, David, and Sharyn O'Halloran. 1996. The Partisan Paradox and the U.S. Tariff, 1877-1934. International Organization 50 (Spring):301-324. Ferguson, Thomas. 1984. From Normalcy to New Deal: Industrial Structure, Party Competition, and American Public Policy in the Great Depression. International Organization 38 (Winter):41-94. Fite, Gilbert C. 1954. George N. Peek and the Fight for Farm Parity. Norman: University of Oklahoma Press. Frieden, Jeff. 1988. Sectoral Conflict and Foreign Economic Policy, 1914-1940. International Organization 42 (Winter):59-90. Frieden, Jeffry. 1997. Monetary Populism in Nineteenth Century America: An Open Economy Interpretation. Journal of Economic History 57 (June):367-95. Gilligan, Michael. 1997. Empowering Exporters: Reciprocity, Delegation, and Collective Action in American Trade Policy. Ann Arbor: University of Michigan Press. Hiscox, Michael. 1999. The Magic Bullett? The RTAA, Institutional Reform, and Trade Liberalization. International Organization 53 (Autumn):669-668. Irwin, Douglas A., and Randall S. Kroszner. 1999. Interests, Institutions, and Ideology in Securing Policy Change: The Republican Conversion to Trade Liberalization After Smoot-Hawley. Journal of Law and Economics 42 (October):643-673. Krehbiel, Keith, and Jonathan Woon. 2005. Selection Criteria for Roll Call Votes. Paper read at American Political Science Association, September, at Washington, D.C. Pastor, Robert. 1980. Congress and the Politics of U.S. Foreign Economic Policy, 1920-1976. Berkeley: University of California Press. Schnietz, Karen E. 2000. The Institutional Foundation of U.S. Trade Policy: Revisiting Explanations for the 1934 Reciprocal Trade Agreements Act. Journal of Policy History 12 (4):417-444. . 19 Table 1: Senate Votes on Trade Legislation, 1922-1949 (by Region) Region Votes 1922 Trade Act 1930 SmootHawley 1934 RTAA 1937 Pepper 1940 Pittman P L P L P L P L P L Middle Atlantic 6 6 0 3 1 4 2 2 2 2 4 North East 12 8 2 10 1 8 4 6 5 10 2 East North Central 10 5 0 7 2 3 7 2 6 3 5 Industrial State Totals 28 19 2 20 4 15 13 10 13 15 11 South Atlantic 16 3 8 6 8 5 11 5 8 3 9 East South Central 8 0 5 0 5 0 8 1 6 0 7 South Totals 24 3 13 6 13 5 19 6 14 3 16 Pacific 6 4 0 5 1 4 2 3 2 5 1 West South Central 8 3 3 3 4 2 6 4 2 1 7 West North Central 14 9 2 8 5 5 9 7 5 6 5 Mountain 16 9 5 10 4 4 12 9 6 11 4 West Totals 44 25 10 26 14 15 29 23 15 23 17 Vote Totals 96 47 25 52 31 35 61 39 42 41 44 20 Table 2: Liberal and Protectionist Votes in the Industrial States, 1934 RTAA Liberal (2 Votes For RTAA) Massachusetts Protectionist (1 Vote For RTAA) (1 Vote Against RTAA) New England Connecticut Connecticut New Hampshire New Hampshire (2 Votes Against RTAA) Maine Rhode Island Vermont Mid-Atlantic New York Illinois Wisconsin New Jersey Pennsylvania East North Central Indiana Indiana Michigan Michigan Ohio Ohio 21 Table 3: Liberal and Protectionist Votes in the Industrial States, 1937 and 1940 1937 Pepper Amendment Liberal (2 Votes For RTAA) Rhode Island Illinois Indiana Protectionist (1 Vote For RTAA) (1 Vote Against RTAA) New England Connecticut Connecticut New Hampshire New Hampshire Massachusetts Massachusetts Maine Mid-Atlantic New Jersey New York Pennsylvania Pennsylvania (2 Votes Against RTAA) Vermont East North Central Michigan Michigan Wisconsin Wisconsin 1940 Pittman Amendment Liberal (2 Votes For RTAA) Protectionist (1 Vote For RTAA) (1 Vote Against RTAA) New England Rhode Island Rhode Island Massachusetts Massachusetts New York Indiana Mid-Atlantic New Jersey New Jersey Pennsylvania Pennsylvania East North Central Michigan Michigan Illinois Wisconsin Ohio Ohio 22 (2 Votes Against RTAA) Vermont New Hampshire Maine Connecticut Table 4: Senate Votes on Trade Legislation, 1930-1940 1930 Cotton 1934 3.50 *** 3.11 *** 1.21 Sugar 0.72 -2.60 ** -2.65 *** 1937* 1940* 1.02 5.10 *** 0.74 -0.52 -1.81 * 1.13 0.85 Corn 1.00 1.13 Wheat 0.87 0.93 0.69 1.23 1.14 Western Mining 9.38 30.88 *** 16.88 * 7.42 7.71 7.57 9.23 6.42 -53.28 *** -57.37 *** 0.88 Petroleum Misc. Manufacturing 0.84 1.43 ** 20.30 17.22 2.07 12.89 13.41 Chemical Textile/Apparel Transportation Machinery 18.30 19.22 * 10.89 4.15 *** 1.56 -3.51 *** 0.97 3.20 *** 0.95 -3.08 *** -33.90 ** -10.27 17.02 -11.19 ** 5.33 13.74 11.10 20.98 31.75 *** 18.21 ** 10.17 8.38 0.47 -0.29 3.70 2.88 -3.08 -1.70 3.17 1.76 3.19 2.00 -1.52 4.42 0.46 -3.28 11.47 Iron and Steel 0.91 1.06 -14.98 ** 11.60 8.93 5.52 * 2.58 3.32 2.42 -9.75 *** -5.15 *** 9.39 8.69 8.65 -1.83 0.12 -4.54 6.73 6.44 7.35 -0.20 -0.56 -1.50 0.74 0.68 1.18 0.77 84 Observations 44.38 Wald chi-square 0 Prob > chi2 Log pseudolikelihood -35.29 0.36 Pseudo R-square 96 39.54 0.0001 -44.32 0.30 81 27.13 0.007 -40.97 0.27 85 32.53 0.001 -39.38 0.33 Constant *1937 vote is on the Pepper Amendment, Second Vote. 1940 vote is on the Pittman Amendment 23 4.92 -1.30 * Table 5: Senate Votes on Trade Legislation, 1930-1940, Controlling for Party Membership Party 1930 5.31 ** Cotton 2.10 ** Sugar -2.87 ** 1934 7.51 ** 2.14 1.01 3.18 1.03 2.05 0.61 1.71 -6.96 ** 1.30 3.42 Corn 2.20 0.51 Wheat 2.29 1.53 1.19 Western Mining 3.58 25.83 ** 1.37 1.33 3.32 *** 6.78 Petroleum -67.63 *** Misc. Manufacturing -27.20 * 27.35 Chemical 1937 1.20 0.83 -0.61 0.86 3.65 *** 1.56 -3.30 *** 27.08 1.00 0.87 -2.75 *** 1.04 8.93 16.32 0.94 2.92 *** 6.34 -30.89 ** 20.63 1.01 -1.70 * 1.20 10.79 29.76 0.84 4.66 *** 14.00 -53.72 * -31.63 1940 0.90 15.70 -12.80 ** 5.79 27.00 ** 13.42 9.35 20.45 16.44 ** 12.83 14.78 Textile/Apparel 0.99 -0.99 Transportation -2.10 3.83 2.93 3.54 1.76 Machinery 12.40 29.76 0.29 -1.87 4.75 5.08 6.30 ** 17.13 Iron and Steel -23.49 * 12.35 Constant Observations Wald chi-square Prob > chi2 Log pseudolikelihood Pseudo R-square 5.81 * 3.20 -7.83 ** 7.70 2.56 1.94 -4.37 ** 22.11 8.03 8.22 -21.86 -0.72 -5.12 15.09 -2.25 * 12.57 6.09 -11.12 -2.85 ** 6.59 -2.06 4.93 -1.77 ** 1.27 1.25 1.09 0.82 83 20.17 0.09 96 31.29 0.003 85 35.67 0.00 -21.88 0.60 -18.71 0.70 80 26.93 0.0127 39.8164 0.2816 24 -38.50 0.35 Table 6: Correlation of Senate Votes on McNary-Haugen Acts and the 1934 RTAA 1934 RTAA Against For 16 26 1927 McNary- Against 16 27 Haugen Act For 32 53 Total Pearson chi-square = 0.007 Pr = 0.933 Fisher's exact = 1.000 1934 RTAA Against For 15 8 1928 McNary- Against 11 39 Haugen Act For 26 47 Total Pearson chi-square = 12.83 Pr = 0.000 Fisher's exact = 0.001 Votes For 1928 McNary-Haugen and 1934 RTAA State Party State Party Alabama Democrat Nebraska Republican Alabama Democrat Nevada Republican Arizona Democrat Nevada Democrat Arizona Democrat New Mexico Republican Arkansas Democrat New York Democrat Florida Democrat New York Democrat Georgia Democrat North Carolina Democrat Illinois Republican North Carolina Democrat Indiana Republican Oklahoma Republican Iowa Republican Oklahoma Democrat Kansas Republican South Carolina Democrat Kansas Republican South Dakota Republican Kentucky Republican Tennessee Democrat Kentucky Democrat Tennessee Democrat Michigan Republican Texas Democrat Minnesota Republican Texas Democrat Mississippi Democrat Washington Democrat Mississippi Democrat Wisconsin Republican Missouri Democrat Wyoming Democrat Montana Democrat 25 Total 42 43 85 Total 23 50 73 Table 7: Correlation of Senate Votes on 1928 McNary-Haugen Act, the 1934 Silver Act, and the 1934 RTAA 1928 McNaryHaugen Act 1934 RTAA 1934 Silver Act Against For 11 6 Against 8 38 For 19 44 Total Pearson chi-square = 13.19 Pr = 0.000 Fisher's exact = 0.001 1934 Silver Act Against For 16 11 Against 9 43 For 25 54 Total Pearson chi-square = 14.46 Pr = 0.000 Fisher's exact = 0.000 26 Total 17 46 63 Total 27 52 79
© Copyright 2026 Paperzz