Scholars have long argued that electoral systems

Transcription

1 Does Constituency Size Affect Elected Officials Trade Policy Preferences? David Karol University of California Berkeley Scholars have long argued that Presidents are less protectionist than Congress while Senators are less so than Representatives due to their larger constituencies. Yet until now this theory has escaped scrutiny. I argue that it is based on a misguided view of trade policy as distributive politics. I show via a series of tests that the theory is untenable. Unlike their differences in constituency size, the pro-trade leanings of the Presidency and Senate are postwar phenomena. Even now state size is unrelated to Senators votes on trade. In tests pooling legislators from both Houses, chamber membership predicts votes while constituency size generally does not. Senators are even less protectionist than Representatives with identical constituencies. Scholars have long argued that electoral systems affect policy outcomes. A familiar claim is that the division of a polity into many constituencies makes officials parochial and reliant on special interests. One widely accepted implication of this theory is that large constituencies insulate officials from protectionist lobbies and promote freer trade. The claim is simple; larger constituencies are more diverse, meaning that a smaller share of local firms and residents will benefit from a given tariff. Thus, scholars argue, elected officials from larger constituencies will feel less pressure to protect any particular sector than their colleagues from smaller ones. Legislators elected from small constituencies will join protectionist coalitions for the sake of sectors that are locally prominent, much as they vote for pork-barrel bills as long as their pet projects are funded. The smaller the constituency, the narrower the interest that may determine the legislator s vote; parochialism produces protectionism. Thus branches of government elected on different bases within the same polity may represent interests differently and diverge on trade policy as a result. The U.S. case seemingly supports this theory. Presidents are consistently less protectionist than the Senate. The Senate is in turn less protectionist than the House. These institutional disagreements over trade policy are longstanding and do not stem from any particular configuration of partisan control of the White House and Capitol Hill. In 1997 Clinton s fasttrack authority bid won 69 Senators votes, yet died in the House, despite the similar size of the Republican majorities in the two chambers. In 1999 a steel quota bill won almost two-thirds support in the House, but only 43 Senators favored even debating the plan which Clinton had promised to veto. In 2002 two-thirds of the Senate voted to give Bush Trade Promotion Authority even though Democrats were in the majority. Yet the result in the GOP-controlled House was long in doubt, and he prevailed there by only one vote. Similarly, in 2005 the battle for approval of the DR-CAFTA Free Trade Agreement was far tougher in the House than it was in the Senate. Such gaps matter, even when they are bridged; the tough fights in the House on NAFTA, TPA, and DR-CAFTA led to concessions protecting key industries to secure majorities in that chamber. In addition, the assumption that the branches of government have predictable relative positions on trade issues is built into the framework of U.S. policy. Since 1934 the largely successful effort to reduce barriers to trade has proceeded via Congressional delegation of authority to the President, a strategy that presumes that he will be at least as supportive of continued liberalization as Congress. Yet despite the persistence and importance of intercameral and interbranch differences on trade policy, their causes are poorly understood. Claims that The Journal of Politics, Vol. 69, No. 2, May 2007, pp Southern Political Science Association ISSN

2 484 david karol variation in constituency size explains them stem from a view of trade policy as distributive or pork-barrel politics. This understanding is widely shared, but based on dubious premises. These include the assumptions that there is a clear, constituencyinduced preference on trade policy that elected officials will represent regardless of party affiliation, that this position is determined chiefly by sectoral concerns, and that all of the politically relevant sectors are protectionist. Each of these premises is dubious. Combined, they grossly mischaracterize trade politics in the United States. However, the dominant view, bolstered by the differences observed between the positions of the President, Senate, and House, has escaped empirical scrutiny until now. In fact the theory that support for protectionism is inversely related to constituency size finds no support from close study of Congressional behavior. Below I show that, unlike differences in constituency size, intercameral and interbranch gaps on trade issues are recent, state size fails to predict votes on trade in the Senate, and Senators are more freetrading than Representatives controlling for population, even when their constituencies are identical. Thus the theory that larger constituencies render elected officials less protectionist has no empirical support. The intercameral and interbranch differences evident on trade issues are real and important, but the standard view as to their source can no longer be accepted. The findings presented here highlight the need for a better explanation for interbranch and intercameral differences on trade policy. Existing Theory Scholars have long explained interbranch policy conflict by asserting that the President s national constituency makes him more attuned to the public interest, while Congress represents special interests (Miller 1993; Moe 1990; Sundquist 1981; Truman 1951; Wilson 1908). Relatedly, students of political economy often identify free trade with the public interest that protectionist lobbies subvert. Thus they use constituency size to explain why Presidents are less protectionist than Congresses (Baldwin 1985; Haggard 1988; Krasner 1978; Lake 1988; Lohmann and O Halloran 1994; Milner and Rosendorff 1996; Pastor 1980; Schiller 1999). 1 Scholars also assert that constituency 1 McGillivray (2004, 77), however, notes that the Electoral College gives Presidents an incentive to favor some parts of the country at the expense of others on trade policy. size differences account for the fact that the Senate is less protectionist than the House (Bailey, Goldstein, and Weingast 1997; Fortier and Ornstein 2003; Gilligan 1997; Irwin and Kroszner 1999; Keech and Pak 1995; Magee, Brock and Young 1989; Rogowski 1987, 2002). Baldwin holds that Since senators generally represent more populous and industrially diversified political units than House members, it is less likely that the proportion of workers employed in import injured industries will be so high that a senator is forced to adopt a protectionist posture (1985,16). Baker explains Senators support for NAFTA by noting that they represent larger and more diverse constituencies than House members (1995, 228). Rogowski finds it almost self-evident that large districts offer insulation from sectoral pressure (1987, 208). Yet the theory that smaller constituencies foster protectionism rests on dubious claims that trade policy is best understood as distributive or porkbarrel politics. In distributive analyses (Austen-Smith and Wright 1994; Lohmann and O Halloran 1994; Lowi 1964; Nielson 2003; Rogowski 1987; Schaatschneider 1935; Truman 1951), each district seeks tariffs on its products. Larger districts will be more diverse and gain less from any particular tariff. Thus a chamber elected from large constituencies will be less protectionist than one chosen from smaller districts. Since the same amount of pork may make a bigger impression in a smaller district the claim that an inverse relationship exists between constituency size and orientation toward distributive politics is plausible. 2 Yet trade politics in the contemporary United States arenot marked by diffuse costs and concentrated benefits flowing only to protectionists. Lobbies exist on both sides of the issue, making porkbarrel analogies problematic. 3 Most trade scholars assume that some lobbies will seek freer trade, even if this policy is underprovided due to the disorganization of consumers. Rogowski (1989) holds that owners of a locally abundant factor of production, be it land, labor, or capital, will seek liberalization. Gilligan 2 Lee and Oppenheimer (1999) find that Senators from smaller states are more likely to serve on constituency committees and focus on bringing home particularized benefits than their largestate colleagues, sometimes by holding out on key bills until they receive a side payment for their constituents. 3 Note that Lowi (1964) argued that already in the postwar years trade policy had ceased to be distributive while Pincus doubts that trade politics were primarily distributive even in the antebellum period since Southern planters and Northern merchants were active in opposition to tariffs (1977, 181).

3 does constituency size affect elected officialsʼ trade policy preferences? 485 (1997), Magee, Brock, and Young (1989) and Irwin and Kroszner (1999) argue that both labor and capital in a given industry will agree on trade issues, with some sectors favoring liberalization. Hiscox (2002) detects both factor- and sector-based cleavages. Karol (2000), Shoch (2001), and Milner and Judkins (2004) find that party remains crucial and may mediate factor effects as Democrats give more weight to the concerns of unions and Republicans to those of business. This may explain the frequent disagreement on trade between Senators of different parties from the same state. While such Senators share a geographical constituency (Fenno 1978), in practice they will be responsive to different groups when forced to choose between them on trade policy. We need not settle these debates. Simply recognizing that lobbies oppose as well as seek protectionist policies implies that, unlike a highway bill, a trade measure cannot have something for everyone. Thus pork-barrel models fundamentally misrepresent trade politics. If a given protectionist lobby will be strongest in certain small districts the same should be true of pro-trade groups. A chamber with smaller constituencies should thus include not only high-demanders of various tariffs, but staunch free traders as well. Given the diversity of larger constituencies, a chamber elected from sizeable districts should include more cross-pressured members and fewer sure votes for either side. The distribution of preferences in the two chambers should have different variances, but comparable medians. Within a body elected from constituencies of varying sizes like the Senate, those from small states should be more extreme, while those from larger ones should be more moderate. 4 It is true that a distributive approach does not necessarily imply that all local interests are protectionist. Yet to the extent that localized pro-trade groups such as export-oriented firms are taken into account the theory that small constituencies promote protectionism is undermined. In practice scholars claiming that smaller constituencies inhibit freer trade overwhelmingly assume parochialism promotes protectionism and ignore pro-trade lobbies. 5 4 Bailey and Brady (1998) find voting on trade issues less partisan among Senators from homogenous states. Local concerns pull them away from the party lines hewed to by their colleagues from more diverse states. The claim that diversity does not produce support for freer trade is consistent with their findings, although they do not make it. 5 Of course legislators are responsive to industries located in their constituencies and interests seeking protection are indeed better organized than those who benefit from trade, given the passivity of Not only are the premises underlying the dominant view shaky, but empirical support for it is underwhelming as well. No consensus exists in the cross-national studies of the effects of electoral institutions on trade policy. Rogowski (1987) and Nielson (2003) find that larger constituencies and proportional representation yield greater openness to trade, yet Mansfield and Busch (1995) and Rogowski and Kayser (2002) reach the opposite conclusion while Grofman and Gray (2000) report null findings. However, most scholars focus on the American case where the well-known relative positions of the Presidency, Senate, and House on trade issues seemingly bolster the dominant view. Yet even here there is room for doubt since there have been no attempts to test its logical implications at the microlevel, i.e., learning whether Senators trade positions are a function of state size or if Representatives who represent entire states vote the same way as the Senators with whom they share a constituency. Data and Tests Below I subject the prevailing perspective to a series of tests. First I consider the relative positions of Congress vis-à-vis the President and the House contrasted with the Senate on trade issues. A pattern rooted in institutional differences as old as the Constitution should be evident from the start. So I track interbranch and intercameral differences on trade policy since I then assess the effect of state population on trade policy voting in the Senate. Next I pool votes from both chambers when legislators in both Houses of Congress considered the same text. This permits a comparison of the effects of constituency size and office on key votes. Finally, I exploit a natural experiment by comparing Representatives and Senators with identical constituencies. If the traditional view that constituency size explains interbranch and intercameral differences on trade policy is correct Presidents should have been consistently less protectionist than Congress, and the Senate less so than the House throughout U.S. history. Within the Senate, those from larger states should be more free-trading than those from smaller ones. In a pooled sample of Senators and Representatives constituency size should predict legislators votes, consumers. However, it does not follow from this that the protectionist side of the debate will be relatively stronger in smaller constituencies.

4 486 david karol while chamber membership should not. Finally, Representatives-at-Large and Senators with whom they share identical constituencies should agree on trade issues. I turn first to interbranch differences. Since 1932 all presidents have indeed been more supportive of freer trade than the Congresses with which they served. Yet earlier in U.S. history, trade politics were highly partisan. Party affiliation trumped institutional position as a predictor of politicians trade policy stands. As a result, Presidents were often more protectionist than Congress, especially when the Chief Executive was a Republican (or a Whig) and Democrats controlled Capitol Hill (Bailey, Goldstein, and Weingast 1997; Schnietz 1994). Space limitations preclude discussion of every President s position on every trade bill. Yet working backward from FDR, it is evident that many Chief Executives took a more protectionist stand than one or both of the Houses of Congress with which they served. GOP Presidents Hoover and Taft vetoed liberalizing bills passed by Democratic Congresses. 6 Republican Benjamin Harrison opposed tariff reductions passed by the Democratic House, which died in the Senate. Although a Democrat, James Buchanan still had protectionist leanings. Congress refused his request for higher tariffs until Southern legislators left Washington (Klein 1962, , Stanwood (1903) Vol. II, ). Before the rise of the Republican Party the Whigs were the chief advocates of protectionism. From 1849 to 1853 two Whig Presidents, Zachary Taylor and Millard Fillmore, served with Democratic Congresses. In their annual messages to Congress these Chief Executives pleaded without success for tariff increases to protect industry. James Monroe did the same (Ammon 1971, 463, 512). In 1816 Congress passed a lower tariff than Madison had sought (Stanwood 1903, 139, 155). Concluding the list of relatively protectionist Presidents,Washington, who endorsed his Treasury Secretary Alexander Hamilton s protectionist Report on Manufactures, might also be cited. Of course earlier in U.S. history there were also Presidents more free-trading than the Congresses with which they served. Two prominent examples are 6 Hoover vetoes Dem bill: message to Cong cites 4 major objections. New York Times May 12, :8. Taft Vetoes Farmers Free List Bill. New York Times August 9,1911, 4: 3:2. HR Adopts Sen Amendment to Cotton Tariff Bill with Cuts in Steel, Chem other Duties. New York Times August 22, :3. Taft Vetoes Cotton Bill. New York Times August 23,1911. Taft to Veto Metals Tariff Bill. August 13, :1. Taft Warns Against Lifting Duty on Sugar. New York Times November 11,1912 9:1. FIGURE 1 Senate-House Difference Intercameral Differences in Support for Freer Trade in the U.S. Congress: Grover Cleveland who proposed greater tariff reductions than Congress would enact and Woodrow Wilson who vetoed the so-called Emergency Tariff bill shortly before leaving office. Simply put, there was no distinct Presidential position on trade policy for most of U.S. history. Rather, the free-trading Presidency is a modern phenomenon. Turning to an assessment of intercameral differences over time, Figure 1 shows the percentage of Senators supporting a liberalizing bill or opposing a protectionist one minus the analogous percentage of Representatives. Positive values for a given bill indicate that the Senate was more free-trading, negative ones that it was more protectionist. The chart runs from 1820, when recorded votes were first cast in both Houses on what is widely seen as a protectionist bill, until Some votes differ slightly in form, e.g., a cloture motion in the Senate versus a vote on passage in the House, but all concern the same text. I exclude cases in which one chamber passed a bill by voice vote, those in which trade is mixed with other salient issues and those on which more than 95% of both Houses agreed. Fifty-eight cases remain. All are listed in the appendix (online at org). Figure 1 shows that the Senate s status as the more free-trading chamber is a recent condition. Except for the case of the 1990 textile bill, the Senate has been less protectionist than the House since Yet before then the Senate-House difference was often negative, i.e., the Senate was sometimes the more protectionist

5 does constituency size affect elected officialsʼ trade policy preferences? 487 body. 7 Lake holds that in the nineteenth century, The Senate was traditionally more protectionist than the House, even though economic interests tend to be more concentrated in the latter (1988,46). The Senate dealt some of the greatest blows to low tariff forces, including the frustration of Grover Cleveland s tariff reform plans in 1894 and the bidding up of the Smoot-Hawley Tariff in Historically, party affiliation strongly predicted politicians stand on trade and institutional location was a secondary factor at best. Depending on party control, the Presidency, the Senate, or the House could all be the protectionists stronghold (Bailey, Goldstein, and Weingast 1997; Schnietz 1994). Thus the Senate s relatively free-trading stand is, like the Presidency s, a postwar development. The consistent relative positions of the Presidency and the Senate vis-à-vis the House on trade issues evident in recent decades thus postdate the constituency size differences said to produce them by more than 150 years. This chronology alone calls into question the prevailing view. Yet perhaps the importance of constituency size has increased for some reason in recent decades and now really does affect elected officials positions on trade issues. To explore that possibility I examine Senate voting during the postwar era. Prevailing theory strongly implies that Senators from larger states will be less protectionist. Yet this ubiquitous claim has gone untested until now. Table 1 displays logistic regression estimates of the effect of state population on 16 important Senate roll calls. I focus on Senators because the variation in the size of their constituency is great while differences in size among districts in the House of Representatives are minimal. These cases comprise the votes on trade issues included in the entire Congressional Quarterly Key Votes series from 1945 to I code the votes so that a positive coefficient means that as state size increases so does the probability that Senators will vote for freer trade. In order to avoid giving too much weight to a handful of states, I use the log of population size. 8 This does not alter the results substantially. 7 Arguably, Senators did not have larger constituencies than Representatives before 1913, as they were elected by state legislators. Yet by this logic Senators from states with more than the median number of electoral votes have larger constituencies than the President. In any case, 1913 and the years prior to it, when some states began electing Senators directly de facto, is not a breakpoint in the time series. 8 Results of analyses using unlogged state population differed very little from those reported here for the logged variable. As Table 1 shows, the relationship between state size and Senators votes on trade issues is significant in only four of sixteen cases and in two of these the sign is wrong according to the dominant view. In 1958 those from larger states were more likely to oppose allowing Congress to override the Tariff Commission and in 1994 they were more apt to support U.S. accession to the W.T.O. Yet Senators from larger states were also more likely to support the Textile-Copper- Footwear tariff bill in 1985 and to oppose granting China Most-Favored Nation status in In the 12 remaining cases state size was not a significant predictor of Senators votes on trade issues. These results undercut the theory that larger constituencies orient officials toward freer trade. To ensure that such a relationship is not obscured by intervening variables I present multivariate analyses of the same votes below including controls for Senators party, coded as 1 for Republicans and 0 otherwise, the weight of merchandise and farm exports in a state s economy, the strength of unions in its labor force and the percentage employed in key importcompeting sectors: Textiles and Metals. Trade politics has historically been seen as a venue for sectional conflict (Trubowitz 1998) so I include a dummy variable, South, indicating whether the Senator represented a Southern state. Finally, since term length could explain House-Senate differences, with Senators temporarily shielded from voters being more freetrading, I add a dummy variable, Term Up, indicating whether a Senator s term expires at the election following a given vote. 9 Table 2 summarizes the results of this investigation. The key finding is that state population is significant in even fewer cases than it was in the bivariate analyses. In the multivariate models population is no longer a significant predictor of opposition to a protectionist reform of the tariff commission in GOP, the party dummy, is the only variable that is significant in most of the cases. As Keech and Pak (1995), Karol (2000), and Shoch (2001) lead us to expect, until the early 1970s Republican Senators were more apt to be protectionist, while since then they have been relatively free-trading. Thus while the sign on the party variable has changed, it remains the most consistent predictor of Senators votes on trade issues. This is another reason to doubt the applicability of distributive models, since pork-barrel politics is generally considered a bipartisan affair. 9 On the 1949, 1955, 1962, and 1987 votes all Senators from one party voted the same way, making the model impossible to estimate. Thus I exclude these cases from Table 2.

6 488 david karol TABLE 1 The Effects of Logged State Population on U.S. Senators Support for Trade Liberalization, Year and Issue 1945 RTAA 1.54 (1.04) 1949 RTAA 1 (.96) 1955 RTAA 1.04 (1.1) 1958 RTAA 3.4 (1)* 1962 Trade Expansion Act.69 (.86) 1974 Trade Act -.15 (1) 1985 Textile/Copper/Footwear Bill (.90)* 1987 Omnibus Trade Bill -1.7 (.94) 1988 Omnibus Trade Bill (.87) 1991 MFN for China (.92)* 1993 NAFTA -.93 (.88) 1994 WTO/GATT 3.7 (1.1)* 1997 Fast Track -.62 (.9) 2000 PNTR for China (.92) 2002 Trade Promotion Authority -.31 (.93) 2005 CAFTA 1.47 (.87) (Standard Errors in Parentheses) * = p-value <.05. Logistic Regression Models Log of State Population Constant Chi-Square N -.57 (.62) -.55 (.59) (.69)* -.85 (.57) -.34 (.50) 1.36 (.56)*.68 (.46) -.22 (.47).26 (.45).90 (.46).95 (.47)* -.80 (.48) 1.09 (.48)* 1.32 (.5)* 1.0 (.49)* -.45 (.45) Other control variables measuring the importance of specific industries or the strength of exporters were each significant in a minority of cases. After the party dummy, the exports variable is the most successful, being significantly and positively associated with votes for liberalization in three cases and almost reaching that level in a fourth. The coefficient for the variable measuring whether a Senator s term expired at the end of the Congress in which the vote occurred was never significant, suggesting that term length is not a factor in intercameral gaps. Null findings were also consistently revealed for the South dummy variable, indicating no role for sectionalism once other variables are held constant. In short, constituency size is generally not a significant predictor of Senators votes on trade policy in bivariate or multivariate models, even in an era in which the Senate is more supportive of trade than the House. Yet the effect of constituency size may be different now as compared to earlier eras in a way not captured by the analyses shown in Tables 1 and 2 due to inadequate controls. To address this concern I pool recent votes in which both chambers voted on the same text and the gap between them was greatest. These cases offer the best chance for constituency size to emerge as a significant predictor of Congressional voting on trade issues. They also permit a comparison of the relative importance of constituency size and membership in the House or Senate in predicting legislators stands.

7 does constituency size affect elected officialsʼ trade policy preferences? 489 TABLE 2 Logistic Regression Models of U.S. Senators Support for Trade Liberalization on Key Votes Vote Log of State Population GOP Term Up Exports Unions Textiles Metals South Constant Chi-Sq. N 1945 RTAA (6.5) 1958 RTAA 4.1 (2.5) 1974 Trade Act -.4 (1.5) 1985 Textile Bill Omnibus Trade Bill -1 (2) 1991 MFN China -6.3 (2.2)* 1993 NAFTA -1.1 (1.2) 1994 WTO/GATT 5 (1.6)* 1997 Fast Track -.9 (1.2) 2000 China PNTR -.8 (1.3) 2002 Trade Promotion Authority.5 (1.5) 2005 CAFTA (.8)*.5 (.7) 1.3 (.6)* 1.7 (.6)* 5.2 (1)* 5.3 (1)* 1.1 (.5)*.5 (.5).9 (.5) -1.2 (.6)* 2.8 (.8)* 3 (.7)* -.3 (.6).2 (.6) -.2 (.6) -.8 (.6) -.3 (.8) -.1 (.8) -.1 (.5) -.1 (.5) -.6 (.5) -.7 (.5) -.3 (.6) 1.1 (.6) 9.8 (6) -.1 (2.4) 1.2 (1.5) 3.2 * 1.5 (2.4) 8.3 (2.7)* (1.3) 1 (1.8) 4 (1.8)* (2.4).9 (1.6).7 (1.2) -.6 (1.9) -3.2 (2.4) 1.2 (2.6) -1 (1.5).8 (.7).4 (.7) -.2 (.8) -1.5 (1.1) -.2 (.9) 6.5 (2)* 1.5 (1.1) 1.4 (1.3) -5.6 (1.8)* -3.2 (2) -3.3 (2.3) (1.5)* (1.9) (2.4)* 1.1 (2.2) -.1 (2.1) (1.5).2 (1.6) -4.6 (2.5).8 (2.2) 3.1 (1.2)* 1.4 (1.2) -1.3 (.9) -1.2 (.9) -1.0 (1.1) (.8) -1 (.7) 1.4 (.8) -.3 (.9) (2) (1.2)* -2.3 (4.2) 2.3 (2.1) -6.6 (2.4)* (2.9)* -4 (1.6) -.3 (1.5) (2.6) -.9 (3.3) 1.8 (1.1) * = p-value <.05 (Standard Errors in Parentheses)

8 490 david karol TABLE 3 Pooled Logistic Regression Estimates of Members of Congress s votes On Trade Bills with Large House-Senate Gaps, (Positive Coefficients Indicate Support for Trade Liberalization) Vote GOP Senator Population Constant P.R.E N 1992 China MFN I 2.9 (.3)* 2.2 (.5)* -.4 (.3) -.8 (3.6) China MFN II 2.8 (.3)* 2.8 (.3)* -.6 (.3)* 2 (3.5) Steel Quotas 2.6 (.2)*.9 (.5)!.2 (.2) -8.3 (3.2)* China PNTR 1.5 (.2)* 1.4 (.6)*.1 (.3) -3.5 (3.7) Trade Promotion Authority 3.5 (.3)* 1.2 (.6)*.01 (.3) -5.4 (3.4) Standard Errors in Parentheses * = p-value<.05 = p-value <.10. Table 3 displays the results of those analyses. Unsurprisingly, it shows that Republicans are less protectionist than Democrats in recent years. More importantly, it reveals that, controlling for constituency size, Senators are significantly less protectionist than Representatives. Constituency size itself (the natural log of the population of Senators and Representatives constituencies) is a significant predictor in only one case, a vote on Most-Favored Nation Status for China, on which it takes the wrong sign, i.e., those with larger constituencies were more apt to take the protectionist side. These results indicate that constituency size does not explain intercameral differences on trade issues. Rather, Senators as a class aremoresupportiveoftrade liberalization than Representatives, controlling for both party affiliation and constituency size. Yet the results presented above may always be questioned on grounds of omitted variable bias. Rather than make a necessarily incomplete attempt to show that such differences persist independent of constituency factors, I exploit a natural experiment. Some Senators and Representatives have identical constituencies. Nowadays this only occurs in the least populous states, e.g., Vermont or North Dakota, which send just one Representative-at-Large and two Senators to Washington. Yet until 1967 a handful of larger states also elected one or more Representatives-at- Large in addition to those chosen from districts. A focused comparison of the votes cast cases by Representatives-at-Large and Senators from the same states permits an assessment of the effects of House versus Senate membership independent of constituency factors. Since the number of such legislators is small at any given time I pool their votes on trade issues across multiple bills. As in Figure 1, I use only votes for which both houses considered the identical text. Since intercameral and interbranch differences are a postwar phenomenon I used only votes occurring from 1945 until the present. Table 4 shows the results of three logistic regressions. In all cases the coefficient of interest is for the dummy variable Senator. It reveals that Senators are significantly more likely to vote for freer-trade than Representatives even in this pooled sample. Since the sample includes only Senators and Representativesat-Large from the same states this means those serving in the Senate are more free-trading than those in the House, even controlling for constituency. As the second and third models presented show, this effect is actually even larger when party and state dummies are added to the analysis. In Models 2 and 3 I code the party variable as a one for Democrats from 1945 to 1970 and for Republicans since then with other MCs getting a 0. This coding reflects the fact that the GOP replaced the Democrats as the more pro-trade party in the early 1970s. In Model 3, which also includes state dummies, North Dakota is the excluded variable. Thus a chamber membership effect is evident even when Senators and Representatives have identical constituencies. This is further evidence that constituency factors do not explain the intercameral differences that have emerged on trade policy in the postwar era. 10 We have seen that intercameral and interbranch differences on trade issues postdate by more than 150 years the differences in constituency size said to explain them. We found that in bivariate and multivariate analyses that state size is generally not a significant predictor of Senators voting on trade issues, even in the postwar era in which the Senate is less protectionist than the House. We saw that in pooled regressions Senators were more supportive of freer trade than Representatives controlling for constituency size 10 By controlling for constituency this analysis also reveals that intercameral differences are not a product of the fact that different sectors are better represented in one chamber than the other, e.g., the stronger position of agriculture in the Senate.

9 does constituency size affect elected officialsʼ trade policy preferences? 491 TABLE 4 The Effect of Membership in House vs. Senate on the Trade Policy Votes of Members of Congress with Identical Constituencies: Senators and Representatives-at-Large from the Same States Compared, (Pooled Logistic Regression Estimates) Variables Model 1 Model 2 Model 3 Senator.48 (.20)*.62 (.22)*.65 (.23)* Party 1.83 (.21)* 1.58 (.28)* Alaska.18 (.47) Arizona 7.48 (29.9) Connecticut 1.01 (.92) Delaware 1.04 (.39)* Hawaii 8.0 (33.6) Illinois 8.0 (33.6) Montana.72 (.47) Nevada.10 (.83) New Mexico 7.48 (17.29) Ohio -.54 (1.28) South Dakota.19 (.39) Vermont -.02 (.37) Wyoming.80 (.46) Constant.16 (.16) -.80 (.21)* (.30)* Chi-Square Pct. Predicted Correctly Number * = p-value <.05 Standard errors in parentheses. and even when, in the case of the natural experiment of Senators compared with Representatives-at-Large, the two classes of legislators shared identical constituencies. The case is overwhelming that constituency size does not account for the notable differences in preferences among the House, Senate, and Presidency on trade issues evident in recent decades. Directions for Future Study While intercameral and interbranch gaps on trade issues cannot be explained by constituency size, they are real and important. Many hold that the liberalization of U.S. trade policy since 1934 stems in part from Congress s delegation of authority to the President (Bailey, Goldstein, and Weingast 1997; Destler 1995; Haggard 1988; O Halloran 1994; Pastor 1980). Yet this strategy can only succeed if Presidents remain at least as committed to freer trade as Congress, as they have been since the days of FDR. Identifying the cause of this pattern could help us predict whether it is apt to persist. In order to explain its postwar emergence we need a new theory in which an institutional characteristic other than constituency size interacts with a factor that varies over time. Space limitations do not permit the full presentation and testing of alternate theories here. Nevertheless, I will briefly note some possible explanations for the pattern revealed above that future research might explore. Term length is an obvious characteristic that varies across political institutions. Yet, it is not clear what other factor it would interact with to produce the postwar emergence of the differences we observe. In addition, if those were more (or less) protectionist Presidents would not be so much more consistently supportive of trade than the median Senator. Moreover, in the models reported in Table 2 the coefficient for the dummy variable Term Up was never significant; Senators up for reelection at the end of the Congress in which a bill was considered did not vote differently from those who were not. One alternate explanation might center on the role of campaign finance. Perhaps the ratio of contributions from protectionist groups as opposed to protrade ones is greater among Representatives than Senators and for Members of Congress generally as contrasted with the President. If such differences exist and arose in the early postwar period they could account for the findings presented above. However, it

10 492 david karol is unclear what institutional features would give rise to such a pattern of contributions. Another factor that merits attention is the number of elected officials in each institution. Collective action problems mount as we move from the Presidency, a chamber of one dominant elected official, to the Senate, to the House. Officials whose actions are more likely to be pivotal may be more apt to take positions in line with their own preferences while those with less influence might weigh political pressures more heavily. Thus when political pressures and policy goals conflict Presidents, and to a lesser extent Senators, should act on their policy goals more often than Representatives. Of course, the reduced collective action problem in the Presidency as contrasted with the Senate and especially the House is no newer than their differences in constituency size. Thus, it alone cannot explain a pattern that is relatively recent. One possibility is that the postwar marginalization of protectionism in elite opinion scholars (Bauer, Pool, and Dexter 1972; Destler 1995; Goldstein 1993) describe has made the number of actors in each branch of government newly relevant. This could be the case assuming that the true beliefs of most elected officials now reflect this dominant view. Even self-interested incumbents have reason to promote policies that they believe will promote prosperity. Thus in a climate in which freer trade is widely seen as beneficial, officials less burdened by collective action problems in resisting lobbies, i.e., Presidents and to a lesser extent Senators, might be less protectionist than Representatives in a way that was not true in earlier years when protectionism was more respectable (Karol 1999). These are only some of the possible explanations that future research might productively consider. A more satisfactory understanding of the patterns described above requires more study and evidence. Conclusion Intercameral and interbranch differences on trade policy are real and important yet their causes remain poorly understood. The trade policy strategy that has prevailed in the United States since 1934 Congressional delegation of authority to the President can only succeed if Chief Executives remain at least as supportive of continued liberalization as Congresses. A key implication of the findings presented here is that the larger constituency of the President is no guarantee that this institutional alignment of preferences will continue to exist. A series of tests revealed that the conventional wisdom does not survive close analysis of Presidents and legislators behavior on trade issues. The postwar emergence of intercameral and interbranch differences on trade issues means that longstanding differences in constituency size alone cannot explain them. Even in more recent decades when intercameral and interbranch differences are clearly evident, constituency size generally is not a significant predictor of Senators votes on trade issues. Differences between Senators and Representatives on trade policy persist in pooled regressions when constituency size is controlled for and, most strikingly, remain even when the analysis is limited to those Representatives and Senators who share identical constituencies. These results become less surprising when we recognize that the distributive models underlying the claim that larger constituencies reduce protectionism are fundamentally flawed. Small constituencies may encourage politicians to practice distributive politics, but the struggles to shape U.S. trade policy do not fall into that category. Far from being congenial porkbarreling log-rolls, votes on trade policy in Congress are marked by bitter conflict between lobbies and increasing partisanship. The leading business lobbies support trade liberalization while organized labor and environmental organizations oppose it. Since all of these groups have party ties, the fact that trade politics often have a partisan flavor should not be surprising. These groups presence in party coalitions means that Democratic and Republican Senators who represent the same state may receive different signals from constituents on trade policy in a way that they would not were the issue in question something like a highway bill. It is not that legislators are indifferent to their constituents or that the concentration of sectors with strong interests in trade policy in their constituencies does not affect MCs votes. Rather it is the case that in contemporary U.S. politics parochialism does not produce protectionism. Factor-based conflict as mediated by party explains MCs behavior far more than disagreement between different industrial sectors. To the extent that constituency effects do matter, they impel some legislators towards freer trade, much as they incline others to protectionism. Acknowledgments I presented earlier versions of this paper at the 2003 annual meetings of the Midwest Political Science Association, the American Political Science Associa-

11 does constituency size affect elected officialsʼ trade policy preferences? 493 tion, and the Center for the Study of Democratic Politics at Princeton. I thank Larry Bartels, Kathy Bawn, Ben Bishin, Paul Frymer, John Geer, Mark Kayser, Bruce Oppenheimer, Wendy Schiller, John Zaller, and three anonymous reviewers. This paper is dedicated to the memory of Nelson Polsby. Manuscript submitted 25 May 2006 Manuscript accepted for publication 3 September 2006 References Ammon, Harry James Monroe: The Quest for National Identity. New York: McGraw-Hill. Austen-Smith, David, and John R. Wright Counteractive Lobbying. American Journal of Political Science 38 (1): Bauer, Raymond A., Ithiel De Sola Pool, and Lewis Anthony Dexter American Business and Public Policy: The Politics of Foreign Trade. Chicago: Aldine Atherton Inc. Bailey, Michael A. and David W. Brady Heterogeneity and Representation: The Senate and Free Trade. American Journal of Political Science 42 (2): Bailey, Michael A. Judith Goldstein, and Barry A. Weingast The Institutional Roots of American Trade Policy: Politics, Coalitions and International Trade. World Politics 49 (3): Baker, Ross K House and Senate. NewYork:W.WNorton and Company. Baldwin, Robert E The Political Economy of U.S. Import Policy. Cambridge: MIT Press. Destler, I.M American Trade Politics. Washington, DC: Institute for International Economics. Fenno, Richard F Home Style: House Members in their Districts. Boston: Little, Brown. Fortier, John C., and Norman J. Ornstein President Bush: Legislative Strategist. In The George W. Bush Presidency: An Early Assessment, ed. Fred I. Greenstein. Baltimore: Johns Hopkins University Press, pp Gilligan, Michael J Empowering Exporters: Reciprocity, Delegation and Collective Action in American Trade Policy. Ann Arbor: University of Michigan Press. Goldstein, Judith Ideas, Interests and American Trade Policy. Ithaca, NY: Cornell University Press. Grofman, Bernard, and Mark Gray Geopolitical Influences on Trade Openness in Thirty-One Long-Term Democracies, Typescript. University of California-Irvine. Haggard, Stephan The Institutional Foundations of Hegemony: Explaining the Reciprocal Trade Agreements Act of International Organization 42: (1) Hiscox, Michael J International Trade and Political Conflict. Princeton: Princeton University Press. Irwin, Douglas A., and Randall S. Kroszner Interests, Institutions and Ideology in Securing Policy Change: The Republican Conversion to Trade Liberalization after Smoot-Hawley. Journal of Law and Economics 42 (2): Karol, David Protecting the Public Interest against The Interests : Presidents, Congresses and Interbranch Differences in Historical Perspective. Presented at the American Political Science Association, Atlanta. Karol, David Divided Government and U.S. Trade Policy: Much Ado About Nothing? International Organization 54 (4): Keech, William, and Kyoungsan Pak Partisanship, Institutions and Change in American Trade Politics. The Journal of Politics 57 (4): Klein, Phillip Shriver President James Buchanan. State College: Pennsylvania State University Press. Krasner, Stephen D United States Commercial and Monetary Policy: Unraveling the Paradox of External Strength and Internal Weakness. In Between Power and Plenty: The Foreign Economic Policies of Advanced Industrial States, ed. Peter J. Katzenstein. Madison: University of Wisconsin Press, pp Lake, David A The State and American Trade Strategy in the Pre-Hegemonic Era. International Organization 42 (1): Lee, Frances E., and Bruce I. Oppenheimer Sizing Up the Senate: The Unequal Consequences of Equal Representation. Chicago: University of Chicago Press. Lohmann, Susanne, and Sharyn O Halloran Divided Government and U.S. Trade Policy: Theory and Evidence. International Organization 48 (4): Lowi,Theodore J American Business, Public Policy: Case- Studies and Political Theory. World Politics 16 (4): Magee, Stephen P., William A Brock, and Leslie Young Black Hole Tariffs and Endogenous Policy Theory: Political Economy in General Equilibrium. New York: Cambridge University Press. Mansfield, Edward D., and Marc L. Busch The Political Economy of Nontariff Barriers: A Cross-National Analysis. International Organization 49 (4): McGillivray, Fiona Privileging Industry: The Comparative Politics of Trade and Industrial Policy. Princeton: Princeton University Press. Miller, Gary J Formal Theory and the Presidency. In Researching the Presidency, eds. George C. Edwards III, John H. Kessel and Bert A. Rockman. Pittsburgh: Pittsburgh University Press, pp Milner, Helen V., and B. Peter Rosendorff Trade Negotiations, Information and Domestic Politics: The Role of Domestic Groups. Economics and Politics 8 (2): Milner, Helen V., and Benjamin Judkins Partisanship, Trade Policy, and Globalization: Is There a Left Right Divide on Trade Policy? International Studies Quarterly 48 (1): Moe, Terry M Political Institutions: The Neglected Side of the Story. Journal of Law, Economics and Organization 6 (Special Issue): Nielson, Daniel L Supplying Trade Reform: Political Institutions and Liberalization in Middle-Income Presidential Democracies. American Journal of Political Science 47 (3): O Halloran, Sharyn Politics, Process and American Trade Policy. Ann Arbor: University of Michigan Press. Pastor, Robert Congress and the Politics of U.S. Foreign Economic Policy Berkeley: University of California Press. Pincus, Jonathan J Pressure Groups and Politics in Antebellum Tariffs. New York: Columbia University Press. Rogowski, Ronald Trade and the Variety of Democratic Institutions. International Organization 41 (2): Rogowski, Ronald Commerce and Coalitions: How Trade Affects Domestic Political Alignments. Princeton: Princeton University Press.

12 494 david karol Rogowski, Ronald Trade and Representation: How Diminishing Geographic Concentration Augments Protectionist Pressures in the U.S. House of Representatives. In Shaped by War and Trade: International Influences on American Political Development, eds. Ira Katznelson and Martin Shefter. Princeton: Princeton University Press, pp Rogowski, Ronald, and Mark A. Kayser Majoritarian Electoral Systems and Consumer Power: Price-Level Evidence from the OECD Countries. American Journal of Political Science 46 (3): Schaatschneider, E. E Politics, Pressure and the Tariff; A Study of Free Private Enterprise in Pressure Politics as Shown in the Revision of the Tariff. Hamden, CT: Archon Books. Schiller,Wendy J Trade Politics in the American Congress: A Study of the Interaction of Political Geography and Interest Group Behavior. Political Geography 18 (7): Schnietz, Karen E To Delegate or Not to Delegate: Congressional Institutional Choices in the Regulation of Foreign Trade: Ph.D. Dissertation. University of California, Berkeley. Shoch, James Trading Blows: Party Competition and U.S. Trade Policy in a Globalizing Era. Chapel Hill: University of North Carolina Press. Stanwood, Edward American Tariff Controversies in the 19th Century. New York: Houghton, Mifflin and Co. Sundquist, James L The Decline and Resurgence of Congress. Washington, DC: The Brookings Institution. Trubowitz, Peter Defining the National Interest: Conflict and Change in American Foreign Policy. Chicago: University of Chicago Press. Truman, David B The Governmental Process. NewYork: Knopf. Verdier, Daniel Democracy and International Trade: Britain, France and the United States, Princeton: Princeton University Press. Wilson, Thomas Woodrow Constitutional Government in the United States. New York: Columbia University Press. David Karol is assistant professor of political science, University of California-Berkeley, Berkeley, CA