Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Assessing the impact of U.S. public opinion : congressional roll call votes on seven free trade agreements MacIsaac, Roderick 2007

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_2007-0493.pdf [ 2.12MB ]
Metadata
JSON: 831-1.0101115.json
JSON-LD: 831-1.0101115-ld.json
RDF/XML (Pretty): 831-1.0101115-rdf.xml
RDF/JSON: 831-1.0101115-rdf.json
Turtle: 831-1.0101115-turtle.txt
N-Triples: 831-1.0101115-rdf-ntriples.txt
Original Record: 831-1.0101115-source.json
Full Text
831-1.0101115-fulltext.txt
Citation
831-1.0101115.ris

Full Text

A S S E S S I N G T H E I M P A C T OF U.S. P U B L I C OPINION: CONGRESSIONAL ROLL C A L L VOTES ON SEVEN FREE TRADE AGREEMENTS  by RODERICK MACISAAC B . A . , Simon Fraser University, 2005  A THESIS S U B M I T T E D IN P A R T I A L F U L F I L L M E N T OF T H E REQUIREMENTS F O R T H E D E G R E E OF M A S T E R OF ARTS  in  T H E F A C U L T Y OF G R A D U A T E STUDIES  (Political Science)  T H E UNIVERSITY OF BRITISH C O L U M B I A August 2007 © Roderick Maclsaac, 2007  ABSTRACT  Even with the heterogeneous nature o f public attitudes on trade liberalization, public opinion still matters in the voting behavior o f Congress. This paper examines the impact o f public opinion on seven free trade agreements. The conclusion is that public attitudes play a significant role in the outcome o f roll call voting in the House o f Representatives. In accordance with Interest Group Theory, Democrats are highly sensitive to interests as expressed as public opinion and as campaign contributions. In accordance with Delegation Theory, on the other hand, House Republicans are consistently more supportive o f the President on trade policy, even when their constituents' views on free trade agreements tend towards popular opposition. Applying the same variables to the Senate, however, did not lead to the same conclusion. This suggests some fundamental representational differences in Congress. The most obvious structural difference as it applies to public opinion is the distinctive election cycles for each chamber.  11  T A B L E OF C O N T E N T S Abstract  ii  Table o f Contents  iii  List o f Tables  iv  List o f Figures  v  Acknowledgment  vi  Research Paper Opening Comments  1  Literature Review  3  Public Opinion  3  Fast-Track  6  Interest Group Theory  8  Delegation Theory  '.  10  Factor Mobility Theories  12  Concluding Remarks  15  Research Design  16  Questions and Hypotheses  17  Variables  19  Analysis  23  Closing Comments  31  List o f References  41  Appendix A : List o f Data Sources  45  Appendix B : Descriptive Statistics o f the Variables  47  iii  LIST OF T A B L E S Table  Page  1.  House roll call votes on F T A s in the U . S . Congress  34  2.  Senate roll call votes on F T A s in the U . S . Congress  35  3.  Predicted probabilities o f v o t i n g i n favor o f F T A s , b y political party, at differing levels o f public opinion  36  4.  Predicted probabilities for the conditional impact o f business P A C contributions, by political party, at differing levels o f public opinion 37  5.  Predicted probabilities for the conditional impact o f public opinion, by political party, at differing levels o f business P A C contributions  37  Predicted probabilities for the conditional impact o f labor P A C contributions, by political party, at differing levels o f public opinion  38  6.  7.  Predicted probabilities for the conditional impact o f public opinion, by political party, at differing levels o f labor P A C contributions  38  8.  Descriptive statistics for the House o f Representatives  47  9.  Descriptive statistics for the Senate  48  10. Top ten districts in favor of, and oppose to, F T A s  iv  49  LIST OF FIGURES Figure  Page  1.  Formal model for roll call votes  39  2.  Predictive probability plots of House roll call votes, by political party, at differing levels of public opinion  40  v  ACKNOWLEDGMENT  The basis o f this thesis began as a term paper for Prof. Paul Quirk. H i s knowledge o f U . S . politics has been o f significant assistance when starting this paper.  M y examiner, Prof. Angela O'Mahony, has been very helpful on matters o f political economy. M y supervisor, Prof. Fred Cutler, has been very helpful on all matters related to this paper.  vi  Opening Comments In the introduction to their book, Congress Resurgent, Randall Ripley and James Lindsay note that on foreign policy matters during the mid-twentieth century, U . S . congressional oversight o f presidential authority was lax when compared with domestic policy concerns (1993, 4). International trade issues had been regarded as primarily belonging to the executive. But when stiff import competition began to impact the profit margins o f domestic producers, from the 1970s onwards, congressional deference to the President began to decline. Global economic interdependence - as part o f the globalization process - began to erode the once solid division between foreign and domestic policymaking (Ripley and Lindsay 1993, 13-14). Trade liberalization, though, remains one o f the hallmarks o f U . S . foreign policy for both Democrat and Republican presidents alike. This has not gone unchallenged by political groups such as labor unions and environmentalists, who link this policy to economic and social exploitation at home and abroad. Sectoral concerns over textile and steel imports have since morphed into widespread public opposition to the trade liberalization policy itself, which crystallized during the 1992 debate over N A F T A . A s Kenneth Scheve and Matthew Slaughter point out, a significant proportion o f the American public has made a connection between trade liberalization, wage levels, and employment prospects in the U . S . (2001a, 93). Whether this connection is accurate is a moot point; the fact that politicians can excite the electorate's imagination over trade liberalization means that public opinion has become a variable to consider when explaining legislative outcomes in this policy domain.  1  The angst from many analysts who describe a popular backlash in their writings (Balint and Destler 1999; Baldwin and Magee 2000a; Biglaiser, Jackson, and, Peake 2004) demonstrates the point. President Clinton's failure to garner congressional approval to negotiate a bilateral free trade agreement with Chile in 1997, and his failure to renew the Fast-Track authorization law from a Republican-dominated Congress during the 1990s, was partially the result o f deep public skepticism o f further trade liberalization. To date, however, the explanations in the scholarly literature remain dominated by economic variables with little or no consideration given to how the American electorate influences congressional actions in this policy area. This paper seeks to address that omission by examining the impact o f public opinion on seven free trade agreements ( F T A s ) introduced and passed in the 108 and th  109 Congresses o f the U . S . The conclusion is that public opinion plays a significant th  role in the outcome o f roll call voting i n the House o f Representatives; and, along with the lobbying efforts o f political action committees (PACs), public opinion can play a leading role in determining congressional behavior. Applying the same variables to the Senate, however, did not lead to the same conclusion. This suggests some fundamental representational differences in Congress. The most obvious structural difference as it applies to public opinion is the distinctive election cycles for each chamber. O f the seven F T A s in this study, six are bilateral treaties between the U . S . and each o f Australia, Bahrain, Chile, Morocco, Oman, and Singapore. The seventh F T A is the multilateral treaty between the U . S . , Costa Rica, E l Salvador, Guatemala, Honduras, Nicaragua, and the Dominican Republic. Some o f the F T A s passed easily through Congress, while others - notably the Central American treaty - were only passed by slim  2  margins in the House. The established procedures for trade negotiations, generally referred to as Fast-Track, tend to restrict the ability o f Congress to thwart these proposals with each successive step towards a finalized agreement. Hence, there is an institutional bias in favor o f a F T A once the executive has obtained Fast-Track approval from Congress to negotiate the treaty.  Literature Review This section begins with a brief discussion on public opinion as a determinant in policy research. This is followed with a description o f Fast-Track, on its origins and its current procedures. Next comes a discussion o f trade policy theories.  Public Opinion A n important ideal in a liberal democracy is that voters should be able to articulate policy preferences to their government. There are plenty o f doubts, however, about the electorate's ability to effectively evaluate policy choices and the decisions made by their representatives (Brady, Canes-Wrone, and Cogan 2002). Anthony Downs's seminal contribution described a rational public having little incentive to inform themselves on policy when a single vote has such a small impact on the overall electoral result (1957). Perhaps this bias has lead many researchers to avoid including public opinion as a variable in their studies. In a recent review o f the literature on public policy, Paul Burstein and A p r i l Linton noted that less than 10 percent o f these studies have public opinion as a variable under consideration (2002, 395-396).  3  Larry Bartels asserts that regardless o f the electorate's knowledge o f policy, representatives do feel a moral obligation to reflect the views o f their constituents at some meaningful level (1991, 458). Suzanna De B o e f and James Stimson found that even with the high rates o f incumbency, House members remained highly responsive to public opinion (1995, 346). A t some level, congressmen do take voter attitudes into consideration when making their decisions. According to David Brady, Brandice CanesWrone, and John Cogan, one reason is that incumbency is not exogenous; electoral safety depends on the representatives' performance in Congress (2002, 135-136). Voters themselves may not be fully informed o f the policy choices, but local elites and especially electoral competitors in the districts w i l l highlight the voting records o f incumbents (Brady, Canes-Wrone, and Cogan 2002, 138). Since voters' knowledge on the issues vary, so does the congressional response to public opinion. House members may not take polling results at face value, says Bartels, giving that some constituents w i l l have strong opinions on an issue, while others would be indifferent (1991, 468). Some sort o f weighting o f the relevant affected groups within the district is likely done. This weighting is what Lawrence Jacobs, Eric Lawrence, Robert Shapiro, and Steven Smith found in a qualitative study on how congressional aides collect information about voter attitudes. Constituents' letters, phone calls, and person-to-person contacts are preferred over opinion polls. The intensity o f the support or opposition to an issue is given a disproportionate level o f attention from congressmen. Jacobs and his colleagues found that it is the congressional leadership, not individual representatives, who respond to public opinion polls. The leadership is seeking to improve their parties' respective  4  standings with the electorate, and to retain (or obtain) the coveted position o f being the majority party in Congress (Jacobs et al 1998). Issue salience is a key consideration here. Bartels showed that public opinion was the dominant variable in defense spending outcomes during the early years o f the Reagan Presidency, due to issue salience (1991). Most researchers concur with Bartels that i f public opinion becomes fixated on an issue, voter attitudes eclipse all other political considerations (Hinton-Andersson and Wood 1998; Jacobs et al 1998; Burstein and Linton 2002; Burstein 2003). A corollary to this belief is that congressmen w i l l not always know when an issue becomes salient with voters. A seemingly trivial issue at one point in time can become an overriding concern at another point. Congressmen cannot be absolutely certain o f the political consequences when they vote on a given policy (Brady, Canes-Wrone, and Cogan 2002). Another important consideration is the election cycles o f Congress. For the House, the the election cycles force members to be highly receptive to voter attitudes at all times (De B o e f and Stimson 1995). For the Senate, however, the staggered electoral cycles means that Senators have more flexibility i n responding to public opinion. Angela Hinton-Andersson and Dan W o o d found that constituency concerns only mattered for Senators when it came time for them to run for re-election (1998, 728). They also found (contrary to De B o e f s and Stimson's findings on House members) that the Senate's responsiveness to public opinion declined as rates o f incumbency increased (HintonAndersson and W o o d 1998, 726).  5  Fast-Track Public opinion can have little influence on congressional voting behavior on trade agreements i f congress defers to the President in this domain. But both Pietro N i v o l a and John Tierney indicate that congressional deference to the President on trade policy matters had ceased by the mid-1980s (Nivola 1993, 101; Tierney 1993, 101). The deference that developed after W W I I , though, had more to do with an absence o f opposition to trade liberalization. A n expanding national economy that sought export markets combined with a lack o f import competition from abroad produced a unique elite consensus. Certain economic sectors did gain protectionist measures from Congress, notably textile mills, but most industries were more concerned with exports than imports (Tierney 1993, 100). The success o f trade liberalization, however, proved to be the catalyst for opposition. Foreign suppliers to American manufacturing increased their market share o f the U . S . industrial output from 4.3 percent in 1965 to 13.5 percent in 1980. These foreign suppliers began drawing the ire o f domestic suppliers, such as the carbon-steel producers. Consumer demand for imports had also increased. The increased demand for Japanese cars, for instance, squeezed the U . S . automotive industry (Falke 1996; Moore 1996; Nelson 1996). The end result by the mid-1980s was a ballooning U . S . trade deficit and a visceral response from Congress. Throughout this period, however, the demand from American industry for new export markets did not abate. Today, Congressmen are faced with heterogeneous constituency demands for both a more, and a less, open trade policy. A s Dirk De Bievre  6  and Andreas Diir explain, this dilemma is complicated further by the uncertainty over which vested interests would gain a greater salience in a particular constituency at any one point in time (2005, 1274 and 1291). Sharyn O'Halloran believes this uncertainty is what drives political behavior on trade policy: Congress requires sufficient information on how future trade agreements w i l l impact their constituencies, i n order for them to reduce their electoral risk o f antagonizing nascent opposition to any particular trade proposal (1993, 295; 1994, 27 and 141). The result over time has been the creation o f Fast-Track procedures. Fast-Track commences when the President provides notification o f trade talks to the House Ways and Means Committee and the Senate Finance Committee. Congress has sixty days in which to disapprove o f the negotiations. Should this first difficult step succeed, Congress foregos the right to amend an undefined trade proposal in exchange for access to the negotiating process, and enhanced information gathering over how local economic interests w i l l be affected by trade liberalization. Congress can withdraw its support at any time during the trade talks (O'Halloran 1993; 1994). Once the finalized agreement is presented as legislation, however, only voting against the entire proposal can thwart the F T A . The most contentious aspect o f Fast-Track is the multi-staged public consultation process. These advisory panels allow certain motivated interest groups to address trade concerns and seek compensatory remedies. From the institutional point-of-view, both Congress and the President obtain valuable information on the likely opposition to the ongoing trade talks. Logrolling between the two branches o f government tends to dominate the Fast-Track process, with the President offering considerable concessions to  7  key economic interests. M a n y o f these concessions consist o f replacing tariffs with nontariff regulations that continue to impede trade flows (McGillivray 2004, 87-88). There are two theories on who gains the most by Fast-Track procedures. Interest group theorists believe that Congress acts as a counterproductive force that undermines the essence o f free trade arrangements. These theorists assume that congressmen are primarily utility maximizers attempting to advance their own interests, and those o f their constituents (Zhang and Laband 2005). Delegation theorists, on the other hand, assume that congressmen are risk avoiders that pass the blame for import competition to the President while offering token criticisms o f U . S . trade policy to their constituents (O'Halloran 1994). Delegation theorists believe that the President is securing the coalition necessary for executive policymaking.  Interest Group Theory Interest group theorists Daowei Zhang and David Laband analyze the positions taken by Senators over the long-standing softwood lumber dispute with Canada. They conclude that the interests o f lumber producers eclipse domestic U . S . consumer and homebuilder interests, even when the latter has mounted a serious lobbying effort to counter the position o f the domestic suppliers. Consumer and homebuilder interests were simply denied access to the advisory panels, when Fast-Track had been approved for the U.S.-Canada free trade negotiations in 1986 (Zhang and Laband 2005). Robert Baldwin and Christopher Magee examine the impact of P A C contributions on roll call votes to multilateral F T A s .  They predict that without any labor group  contributions, the probability o f G A T T (Uruguay Round) and N A F T A being approved  8  would have been 13 percent and 15 percent higher respectively; and without any business group contributions, the probability o f approval would have been 8 percent and 9 percent lower respectively (Baldwin and Magee 2000a). P A C contributions, though, do not always matter to congressional outcomes. Magee later teamed up with Eugene Beaulieu to show that U . S . labor groups have had an inconsistent record in affecting congressional outcomes to trade votes (Beaulieu and Magee 2004). Many American economists examining U . S . trade policy tend to be interest group theorists. Their focus is on capital-intensive, unionized-labor-force industries that are geographically concentrated. Michael Moore points to the steel industry's diminished effectiveness on lobbying Congress for protectionist measures as a result o f declining numbers o f blue-collar workers employed in Pennsylvania and Ohio (1996, 16 and 32). Michael Finger and Anne Harrison connect the collapse o f the textile m i l l lobby to the defeat o f many Southern Democrats during the 1994 mid-term elections (Finger and Harrison 1996, 49). Public opinion as a variable in shaping policy, however, is incorporated in few o f the interest group theorists' models. It can be implied from Zhang's and Laband's analysis that they substituted consumer and homebuilder interests as an indirect measure of what constituted public opinion. Baldwin and Magee gave little consideration to the impact o f public opinion. A s for Moore and Finger, they acknowledge public opinion mattered, but this fact was simply o f secondary importance to their analysis. Europeanbased trade policy researchers De Bievre and Diir point out that the efforts o f interest groups can be largely futile without public support. Should these groups mobilize public  9  opinion, though, the interaction effect between the two variables can be large and highly significant (De Bievre and Diir 2005, 1274). A question arises from the preceding statement, however, in whether public opinion actually reflects the local interest. It can be readily assumed that the general public has only a limited knowledge o f F T A s . These trade negotiations, and the legislative logrolling sessions that follow, are not open to the public. A s Fred Cutler points out, though, despite the level o f knowledge that the general public may have on a particular issue, they do have an intuitive sense o f what is in their own interest. The social and economic structures o f the communities in which the voters live, and the daily interactions with the people that live with them, have a significant impact on public opinion formation (Cutler, 2007). Self-identification with a particular community incorporates the idea that people understand what are the underlying interests o f that community.  Delegation Theory Delegation theorists assume that the President is less susceptible than Congress to interest group activity. The domination o f logrolling to secure passage o f F T A s is acknowledged to diminish the impact o f trade liberalization, but the overall objective - a more open trade policy - is never much in doubt. Fred Bergsten refers to this as the "one step backward, two steps forward" approach to trade policy (2002, 6). Andreas Falke believes that informed politicians know that the U . S . cannot shut out imports without severe economic consequences, and he laments the protectionist pressures that are placed on Congress and the President (1996, 281 and 289-290). Bergsten offer no such lament,  10  for he realizes that these pressures can and are be manipulated to force reticent countries to address U . S . trade irritants and open up their markets to American imports (2002, 9). Fast-Track procedures are designed to simultaneously address the heterogeneous constituency demands over U . S . trade policy. For American congressmen, Fast-Track has never been just about negotiating F T A s with other countries. The research focus for delegation theorists is how levels o f partisanship and ideological polarization impact decision-making. Richard Sherman pursues this approach with his analysis o f trade-weighted average tariffs using economic and party affiliation variables. He concludes that Republican presidents have been more protectionist than Democrat presidents since W W I I . The implication is that Republican presidents are building the necessary coalitions with a more protectionist Democratic Congress to secure trade agreements. Ephemeral protectionist measures, however, do not eclipse the long-term objective o f trade liberalization (Sherman 2002). Delegation theorists Glen Biglaiser, David Jackson, and Jeffrey Peake analyze the positions taken by Congress on the 1997 Chilean Fast-Track proposal and the 2001 presidential Trade-Promotional Authority. They use two sets o f indexed congressional roll call votes (one for Republicans and one for Democrats) as dependent variables with interest group positions, constituency descriptives, and deference to the President as independent variables. They conclude that the most important influence on Congress delegating trade negotiation authority was party affiliation. Constituency factors were important to Republicans in 1997, when combined with their antipathy towards Clinton. But when Bush became President, constituency descriptives ceased to be statistically significant for Republicans. Democrats were more sensitive to one constituency variable:  11  the ratio o f the blue-collar workers i n their district. B y implication, the prevalence o f blue-collar workers is used a proxy for public opinion in these districts, even i f that opinion is only latent. Democrats' opposition to trade liberalization was moderated by their loyalty to Clinton in 1997, but this effect disappeared in 2001 when Bush became President (Biglaiser, Jackson, and Peake 2004). Similar to the interest group theorists, delegation theorists have taken an economic determinants approach to analyzing trade policy. There is a presumed influence o f public opinion impacting presidential decision-making in Sherman's study, but he does not expand his model to include a public opinion variable. Biglaiser, Jackson, and Peake indirectly measure public opinion with their inclusion o f the ratio o f blue-collar workers as a variable, but this is unsatisfactory. These analysts assume that blue-collar workers, and only blue-collar workers, might be opposed to trade liberalization. They do note that Congress assesses public attitudes before voting on trade policy issues (Biglaiser, Jackson, and Peake 2004, 682). In the end, however, public opinion surveys were not added to their models.  Factor Mobility Theories A second stream o f literature on trade liberalization revolves around the economic concept o f factor mobility. The factors in question are labor and capital. The two main theories are Heckscher-Ohlin model and the Ricardo-Viner model. The distinction between the two competing models is whether or not workers and investment can be moved effortlessly and without significant cost to owners o f factors. If factors are highly mobile, then wage levels and returns on investment would be the same over the long-  12  term. This is the Heckscher-Ohlin model. If factor mobility is impaired, wage levels and returns on investments w i l l stay varied between sectors. This is the Ricardo-Viner model. It is assumed by many economic analysts, such as Scheve and Slaughter, that the Ricardo-Viner model applies i n the short-run outcome whereas the Heckscher-Ohlin model represents the long-run outcome (2001a, 48). Two aspects o f factor mobility impact international trade. First, under the assumption o f the Stolper-Samuelson Theorem, those who possess an abundant amount o f a factor are assumed to be able to withstand import competition more vigorously, and take advantage o f export opportunities more readily, as compared to those who possess a scarce amount o f a factor (Alt et al 1996, 692; Kaempfer and Marks 1993, 728). Second, under the assumption o f the Mundell Equivalency, factor mobility and trade flows are economic substitutes. This means that the more mobile the factor is between sectors, the less effective protectionist measures should be in benefiting a particular sector (Hiscox 2004, 254). Taken together, i f factor mobility is high, class-based characteristics would be the salient economic determinants on whether opinion would favor trade liberalization. If factor mobility is low, then sectoral interests ought to prevail when determining who would favor an open trade policy. Sheve and Slaughter subscribe to the Heckscher-Ohlin model o f factor mobility not just to international trade interdependence but also to many different aspects o f globalization. They found that income and education levels mattered in public opinion formation on trade policy matters. One o f the reasons for accepting the Heckscher-Ohlin model is that they found scant evidence that employment within an export-seeking, or  13  within an import-competing, sector mattered to public opinion formation with regards to trade policy (Sheve and Slaughter 2001a; 2001b). Michael Hiscox adheres to the Ricardo-Viner model o f factor mobility. He uses economic characteristics - agricultural, industrial, and capital ownership variables - to show long-term generational changes to U . S . factor mobility (Hiscox 2002a; 2002b). Hiscox believes factor mobility has declined in recent years, resulting in a less coherent trade policy being articulated by the U . S . federal government (2002b; 2004). The discrepancy in the economic analyses is attributed to the heterogeneous nature o f the U . S . economy, which has both export-seeking and import-competing sectors in the same regions. Added to this problem is the congressional districts themselves. District boundaries tend to cut through industrial areas, and are often composed o f a multitude o f very different communities. Marc Busch and Eric Reinhardt identify this problem. They measure geographically-concentrated economic clusters within the U . S . as a unit o f analysis. Busch and Reinhardt found public opinion formation based on sectoral interests to be highly significant within these economic clusters (Busch and Reinhardt 2000). Another researcher related to the factor mobility literature is Fiona M c G i l l i v r a y . She is rather unique in her trade policy research, in showing how political institutions do not accommodate a homogeneous set o f interests. The most obvious reason is that the House, the Senate, and the Office o f the President all have different electoral bases and political constraints in which these institutions operate (McGillivray 1997, 592-593; 2004, 77). M c G i l l i v r a y analyses conclude that the most important predictors to modeling  14  trade policy, with congressional districts as a unit o f analysis, are trade exposure, the unemployment rate, and the nature o f the party system (2004, 84 and 118; 1997, 603). Researchers on factor mobility sometimes take public opinion on trade more seriously as a variable, since they are looking for the underlying demographic characteristics to explain their models. The problem is trying to make the connection between factor mobility, public opinion, and congressional roll call voting. Busch and Reinhardt have resolved the issue by divorcing public opinion from political institutions. But political institutions in a liberal democracy are suppose to reflect public opinion.  Concluding Remarks Congressmen can be more than just utility maximizers. Representation, as it assumed by the economic analysts, is taken too literally to simply mean advancing the economic interests o f constituents. Representatives have many issues to consider when making political decisions. Congressmen may decide that even though a legislative decision adversely affects the interests o f their district, the action may serve to advance some other cause. One plausible scenario is that politicians w i l l vote in favor o f a F T A , even i f hurts their districts' short-term interests, in order to advance an ideological commitment on trade liberalization. This position would be more in line with Delegation Theory. A t the same time, though, congressmen must be mindful o f their own electoral prospects. The principal task in this paper is to validate or refute De Bievre's and Diir's assumption that public opinion combined with interest group activity has a powerful effect on legislative outcomes. Logrolling between politicians and interest groups,  15  however, is not openly observable to the public and is therefore not actually measurable (O'Halloran 1993, 303; 2004, 27). P A C contributions, as used in Baldwin's and Magee's analysis, represent the most plausible measure o f interest group activity. If the combination o f these two variables is significant, this would indicate Interest Group Theory is more relevant. The competing theories on factor mobility are interesting as they relate to public opinion formation. The underlying consideration on Fast-Track procedures, though, is that constituency demands are heterogeneous. This implies the Ricardo-Viner model would outweigh the Heckscher-Ohlin model with regards to roll call votes. A s Busch and Reinhardt point out, however, sectoral interests on trade policy are not easily translatable into congressional voting behavior. Only the Hechscher-Ohlin model w i l l be tested in this paper.  Research Design Given the heterogeneous nature o f public opinion on trade policy, is it plausible to think public opinion matters on such a complex issue? M u c h o f the F T A negotiations are outside o f the scrutiny o f the media and, perhaps, even among the less influential politicians. There are many facets o f trade policy for congressmen to consider, and the elaborate procedures for Fast-Track would lead causal observers o f Congress to say no to the preceding question. If voter attitudes did matter in a complex area such as trade liberalization, then policy researchers should think about adding public opinion as a determinant to consider in their own future studies.  16  Questions and Hypotheses The first question to ask is a simple one: Does public opinion have an impact on congressional roll call voting with regards to F T A s ? Economic variables such as income levels and trade exposure have found to be significant in the literature, as has interest group activity and party affiliation. But what is the impact when public opinion when incorporated with these other variables?  De Bievre and Diir suggest that public opinion  should have an effect on legislative outcomes. Hypothesis One: Congress takes public opinion into consideration when voting on F T A s . Public opinion influences congressmen to vote for F T A s when there is popular support for these agreements i n their district or state; and public opinion influences congressmen to vote against F T A s when there is popular opposition in their district or state.  Two more questions follow from Interest Group Theory. First, do P A C contributions have an impact on congressional roll call voting on F T A s ? From the literature, Baldwin and Magee have found P A C contributions to be significant, although this effect varies. Hypothesis T w o : Congress takes into account both business and labor P A C contributions when voting on F T A s . Business contributions w i l l have an impact on voting for these agreements; and labor contributions w i l l have an impact on voting against these agreements.  The third question is the interactive effect: Does the combination o f public opinion and P A C contributions have an impact on congressional roll call voting on FTAs?  Again, from the literature, the interaction should have a powerful effect on the  outcome to legislative actions.  17  Hypothesis Three: Congress, when confronted with a congruence o f popular support and business P A C contributions in favor o f F T A s , w i l l vote in favor o f F T A s  Hypothesis Four: Congress, when confronted with a congruence o f popular opposition and labor P A C contributions against F T A s , w i l l vote against F T A s .  Delegation Theory implies another question: Does Congress delegate authority to the President on formulating and implementing F T A s ? Biglaiser, Jackson, and Peake suggest that delegation to the executive only happens when the majority o f congressmen belong to the same party as the President. For the 108 and 109 Congresses, th  th  Republicans would delegate authority to the President, whereas the Democrats would not delegate authority. Hypothesis Five: Republicans w i l l vote for F T A s regardless o f public opinion in their district or state; Democrats w i l l vote for F T A s i f public opinion is favorable in their district or state, and Democrats w i l l oppose F T A s i f public opinion is opposed in their district or state.  The last question is with regards to the economic and social characteristics under Heckscher-Ohlin model o f factor mobility: D o income and education levels in a district or a state impact congressional roll call voting on F T A s ? These are class-based characteristics found to be significant by Scheve and Slaughter.  Hypothesis Six: Congress takes into account the class-based characteristics o f income and education levels when voting on F T A s .  18  Since the dependent variable in this study is an indicator variable, hypotheses one, two, and six w i l l be addressed using a binary logit regression model. Hypotheses three, four, and five w i l l be answered using simulation techniques o f this logit model to yield predicted probabilities for congressional voting behavior. The simulations w i l l only be done, though, i f the variables public opinion and/or P A C contributions are proven to be significant i n the logit regression model.  Variables The seven F T A s i n this study correspond to a time when the Republicans controlled the House, the Senate, and the Office o f the President. The intent o f studying this period is avoid the partisan complications that were experienced in the 1990s, when a Republican Congress help thwart the trade policy proposals o f a Democratic President. One o f the motivations for the Republicans hindering Clinton's political agenda was quite simply that he was the leader o f an opposing party. Republicans had no real desire to see a Democrat in the White House succeed on trade policy - even i f they agreed with the objective o f trade liberalization (Balint and Destler 1999, 23). Bush remained President for the 2003 to 2006 period, so partisan attitudes in Congress towards the executive should not have changed precipitously during this time frame. Moreover, the incumbency rates for the membership in Congress were very high. For the House, 98 percent o f those elected in 2004 were incumbents; for the Senate, 96 percent were incumbents (Center for Responsive Politics, The big picture). There are two dependent variables: one for the seven votes in the House and one for the seven votes in the Senate. The purpose o f having two models is to see what  19  variables influence the outcomes o f roll call votes in the two legislative bodies. For each dependent variable, the independent variables are constructed from the same sources using the exact same time frame, so that cross-chamber comparisons are relevant. The final vote on each F T A Implementation A c t is taken as the roll call vote. The principal source for the dependent variables is from Dr. Keith Poole's Voteview website. The data from this website has been confirmed with data from the Library o f Congress' Thomas website (Poole; U . S . Library o f Congress). The main independent variable is Public Opinion on F T A s . The data is taken from the 2004 National Annenberg Election Survey. The question used is as follows: "The federal government (is) negotiating more free trade agreements like N A F T A : do you favor or oppose the federal government doing this?" (Annenberg Public Policy Center). Survey respondents were asked to rate their position on a five-point scale from strongly favor to strongly oppose (coded 2 to -2). There were more than 64,000 valid responses over a period from 7 October 2003 to 19 September 19 2004; o f those respondents, 42 percent expressed their approval for the policy, 40 percent o f them expressed disapproval. There were more than one-hundred responses from a majority o f the congressional districts, and all but one district had more than fifty respondents to the trade question. Party identification is minimally related to public opinion on trade. Four-five percent o f Republican identifiers supported the policy, as compared to 40 percent o f Democrats and 41 percent o f independents. Thirty-seven percent o f Republican identifiers opposed the policy, as compared to 41 percent o f Democrats and 43 percent o f independents.  20  When it comes to complex issues such as trade policy, as Hiscox notes, it is very easy to elicit a desired response to a leading question that emphasizes the costs o f trade (such as outsourcing), as compared to a question that does not emphasize either the costs or the benefits (2006). Unfortunately, many surveys on trade policy matters deliberately or inadvertently asked leading questions that could bias the results. The trade question from the 2004 Annenburg survey, however, is sufficiently neutral in tone to prevent overtly biased results. The Public Opinion variable takes on the mean values for each district and state. For the House, the values range from -0.876 to 0.599. The full descriptive statistics, as well as the sources for all the variables listed in this paper, are found in the appendices. Thirteen percent o f respondents said they "don't know" whether they approved or disapproved o f the F T A policy (Annenberg Public Policy Center). Rather than remove nearly 9,000 responses from the analysis, this data was recoded as neutral (i.e. with those who said they did not approve or disapprove o f the policy). The purpose for the recoding is twofold. First, these respondents represent potential voters who would be indifferent to trade issues, but perhaps still highly motivated to cast their ballot at election time. These potential voters would be o f interest to office-seeking politicians. Second, because these respondents represent potential voters, the magnitude o f the support or opposition to trade liberalization has to be placed i n a proper context. There w i l l be a significant numbers o f voters in the population, come election time, that w i l l not have strong opinions either way. To ensure that adding the do-not-know responses does not bias the Public Opinion variable, a control variable called Lack o f Opinion has been added to the model. This  21  variable is the proportion o f do-not-know answers in each district and state. This procedure is an alternative proposal to listwise deletions as suggested in the companion book to the 2004 Annenberg survey (Romer et al 2006, 23). For the House, Lack o f Opinion ranges from from 4.1 percent to 25.9 percent. For the variable Party, Senator J i m Jeffords and Representative Bernard Sanders, both o f Vermont, are coded as Democrats for this study, even though congressional records have them listed as independents (U.S. Congress). Both men had been included in the Democrats' caucus during the 108 and 109 Congresses. There are two control th  th  variables to Party, each representing the incumbency rates for Republicans and Democrats. The mean incumbency rate for Republicans in the House is 10 years; for Democrats, 12 years. The mean incumbency rate for Republicans in the Senate is 16 years; for Democrats, 21 years. The P A C contribution variables are rounded to thousands o f U . S . dollars. Business P A C s contribute far more in funds than do Labor P A C s . The maximum value given to a House member from Business P A C s is $2,247 million; and from Labor P A C s , it is $791,000. There are six economic and social variables. Each o f these variables are expressed as the difference between the district's, or the state's, measurement and the national measurement. For instance, for the variable Unemployment, i f the rate o f unemployment in Alabama is 7.2 percent and the national rate is 6.9 percent, then the difference is 0.3 percent. This indicates that Alabama has a slightly higher unemployment rate as compared to the U . S . average. The income variables are Per Capita and Median Household, measured in dollars. The former represents a district's  22  economic output; the latter represents household income alone. The reason for the two variables is that a suburb with little industry would have a lower economic output, but a high household income. Conversely a working-class district may have significant industrial output, but have a low household income. Education is the percent difference in those constituents twenty-five years or older who have a university degree. Export Exposure is the percent difference in ratio o f a state's exports to its economic output as compared to the national ratio. Union Strength is the proportion o f the workforce sixteen years or older that belong to a union. Lastly, the data on roll call votes is stacked. This means there are a maximum o f seven votes recorded by each House member and each Senator. Each F T A , called B i l l , represents a fixed-effects cluster from which the individual characteristics o f that vote are calculated separately. Hence, there are seven intercepts that represent the baseline support for each F T A . The formal model for roll call votes is showed in Figure 1, on page 39.  Analysis The results o f the logit regression models are in Tables 1 and 2, found on pages 34 and 35 o f this paper. A s expected, public opinion in the congressional districts does matter to House members. Given the short-term election cycles o f just two years, this finding is not all that surprising. There is evidence to validate the hypothesis that House members do take public opinion into consideration when voting on F T A s . For the Senate model, however, Public Opinion does not matter. The problem is likely the staggered election cycle, in which only a third o f Senators are actually thinking about the  23  immediate impact o f their voting record. Even so, the Senate is suppose to be responsive to states' interests. Perhaps the lengthy years o f service have made Senators insensitive to public opinion. Does public opinion causes House members to vote for or against F T A s ? Public opinion formation is derived from a variety o f sources, especially locality. But incumbent House members would be leading opinion makers, influencing both voters and the local media to a particular position on issues such as trade liberalization. Busch and Reinhardt believe public opinion formation is derived independent o f either politicians or the media (2000). Mobilization o f public opinion on trade does happen regardless o f which congressional district that voters reside in, or what level o f political information they are acquiring from the media. Local interest, says Busch and Reinhardt, plays the dominate role in public opinion formation (2000, 714). It is certainly plausible to believe that public opinion causes House members to vote for or against F T A s . Still, there is an information feedback loop between the media, the politicians, and the general public that cannot be fully isolated to determine a conclusive causal relationship. Therefore, there is qualified support for the idea that public opinion causes legislative behavior on F T A s . The Lack o f Opinion control variable is statistically insignificant for both models, implying that the do-not-know responses are sufficiently random so as not to bias the regression results for Public Opinion. It is unexpected to see Party in the Senate model not be significant, since this is usually the strongest indicator o f roll call voting in many analyses on Congress. The problem is the incumbency variables. When these variables are removed from Senate  24  model, Party does becomes a strong indicator on how Senators vote on F T A s . Party is the only variable that becomes statistically significant in the Senate model, however, when the incumbency variables are removed. The length o f congressional service does not appear to matter for Republicans with regards to voting on F T A s . Veteran Democrats in the Senate, however, are more inclined to vote against F T A s , on average, as compared to their caucus as a whole. The House incumbency variable for the Democrats is also significant, and the coefficient is in the same direction as the Senate. Veteran Democrats tend to be skeptical o f Bush's policy on F T A s . The two main P A C contribution variables are highly significant in the House model, but not in the Senate model. A s with Public Opinion, this distinction likely has to do with the different election cycles for each chamber. The effects on the House model, although significant, are not very large. There is weak evidence to validate the hypothesis that House members do take P A C contributions into consideration when voting on F T A s . Do business P A C contributions cause House members to vote for F T A s ? D o labor P A C contributions cause House members to vote against F T A s ?  Are the  congressmen voting on these agreements because o f the contributions they are being given, or are the congressmen being given contributions for the way they vote? From the variance in P A C contributions, two observations can be noted. Contributions to congressional leaders are higher than those given to average House members. For instance, the then House Minority Leader Nancy Pelosi received $684,000 in business P A C contributions and $349,000 in labor P A C contributions during the 109 Congress th  25  (Center for Responsive Politics, Members o f Congress). House Democrats, on average, received only $305,000 and $152,000 respectively during this same period. The implication is that P A C s were attempting to garner influence from an established member o f the Democratic leadership. O n the other hand, for newly-elected Democrats to the 109 Congress, the average contribution was $338,000 from business P A C s and th  $195,000 from labor P A C s . These higher contribution averages are partially due to the freshmen campaigning in competitive electoral districts - but these freshmen have no congressional record. Clearly, P A C s contributions are not just about making donations based on the way politicians vote in Congress. The answer to the last question has to be both responses. There is weak support for the idea that P A C contributions cause House members to act on F T A s . A m o n g the economic and social variables, Unemployment and Export Exposure are influential. The higher the percent difference in the unemployment rate for the districts, as compared to the national rate, the more likely House members w i l l vote against F T A s . Conversely, the higher the percent difference in the export exposure, as compared to the national level, the more likely congressmen w i l l vote in favor o f F T A s . Although Unemployment is marginally significant in the Senate model, the direction o f the coefficient is intuitively the wrong way. This indicates there are more problems associated with trying to make cross-chamber comparisons. A s for the income and education variables, there is no evidence to validate the hypothesis that Congress does take class-based characteristics into account when voting on F T A s . A t this point, it should be clear that the logit regression model for the Senate did not reveal many details. Clearly, there are fundamental differences between the two  26  chambers, at least i n the way these institutions address Bush's policy on F T A s . This paper does not present any further interpretation o f the Senate model. To provide a better understanding o f the House regression results, simulations have be applied using the coefficient estimates to obtain the predicted probabilities for House roll call votes. The logit model assumptions still apply. One thousand simulations have been drawn. The general idea is to hold all but one o f the independent variables constant at their respective mean values in order to obtain the predicted probabilities for House roll call votes, as only one independent variable is adjusted along set points. To show the magnitude o f change that public opinion has on roll call votes, the simulations are first used to predict what happens when attitudes on F T A s change dramatically in the districts. Taking the predicted probability that House members would vote for F T A s when Public Opinion is set at the 9 5 percentile, 0.378, and subtracting it th  from the 5 percentile, -0.478, reveals a 13 percent predicted change in congressional th  voting behavior. For Republicans, however, the predicted change is only 3 percent; for Democrats, the predicted change is 24 percent. This result suggests that Republicans and Democrats may behave differently on F T A s . To examine this difference more throughly, the simulations are used to test whether Republicans delegate authority to President Bush and, conversely, whether Democrats do not delegate authority. In this particular case, Public Opinion has been adjusted along the five-point scale while the other variables are fixed at their respective mean values. T w o sets o f predicted probabilities have been obtained, one for each party. The results are i n Table 3, on page 36 o f this paper.  27  The first column represents the predicted probability that Republicans w i l l vote for F T A s , at a given level o f Public Opinion, in descending order o f support. The second column represents the predicted probability for the Democrats. The value o f most interest is the pivot point. This predicts where a majority in a caucus would vote against F T A s , given a certain level o f public opposition to trade liberalization. The values o f these pivot points are recorded in the last row o f the table. The pivot points are showed in Figure 2, on page 40, where the yeas and nays intersect. Democrats are predicted to be highly sensitive to public opinion. When there is public support for F T A s , the Democrats are predicted to vote for such agreements. This suggests that Democrats would not be opposed to Bush's policy o f trade liberalization simply because the President is a Republican. The pivot point is effectively at the neutral position for Public Opinion. This would suggest that Democrats do not delegate to the President; rather, they are more concerned about public opinion in their districts. Republicans are predicted to be far less responsive to even moderate levels o f public opposition. There is a near 100 percent probability that they would vote for F T A s i f public opinion is at least neutral to these agreements. Only when there is strong opposition in their districts would Republicans be compelled to vote against F T A s . The pivot point lies between strongly opposed and somewhat opposed. Clearly, the Republicans are predicted to ignore public opinion under all but the most extreme circumstances. This suggests that Republicans may be delegating authority to the President. Certainly, most Republicans have an ideological commitment to the President's policy objective on trade liberalization. Another plausible scenario, however,  28  is that Republicans are simply responding to business interests, which happens to be in line with the Republicans' ideological goal. To see i f business interests have an influence, further simulations are used to test the interactions between public opinion and P A C contributions. First, the simulations are used to predict the conditional impact o f P A C contributions at differing levels o f public opinion. Again, Public Opinion has been adjusted along the five-point scale while the other variables are fixed at their respective mean values. Second, the simulations are used to predict the conditional impact o f public opinion at differing levels o f P A C contributions. For both business and labor P A C s , the starting point is zero contributions. The scale o f the monetary values has been arbitrarily set based on a range o f values in the dataset. The monetary values roughly approximate the midpoints o f distinctive clusters in P A C contribution levels. T w o sets o f predicted probabilities have been obtained, one for each party. The results for the interactions between public opinion and business P A C s are in Tables 4 and 5, which are on page 37 o f this paper. The results for the interactions between public opinion and labor P A C s are in Tables 6 and 7, which are on page 38 o f this paper. Starting with Table 4, the Republicans are predicted to support F T A s with near certainty provided that public opinion is not opposed to Bush's trade policy. In fact, the coefficients for the interaction between public opinion and business P A C contributions are nearly the same for public opinion alone (in Table 3). Democrats, on the other hand, are far more likely to be positively affected by business P A C contributions. The probability that Democrats would vote for F T A s at the neutral position jumps from 50  29  percent when considering public opinion alone, to 91 percent when public opinion and business P A C contributions are combined. Looking at Table 5, Republicans are predicted with a high probability to vote for F T A s , regardless o f what level o f business P A C contributions are given to these congressmen. P A C s may donate large sums o f money to Republicans, but the predictions indicate that this money has little impact to these members' commitment to trade liberalization. For Democrats, the level o f contributions matters. Donation levels at $350,000 are predicted to produce 16 percent increase in the probability that Democrats would vote i n favor o f F T A s , as compared to when business P A C s make no campaign donations to these congressmen. With greater certainty, having eliminated the business interests explanation, it can be said that there is evidence to validate the hypothesis that Republicans are delegating authority to the President on trade liberalization. There is no evidence to validate the hypothesis that Republicans alter their positions when confronted with a congruence o f popular support and business P A C contributions in favor o f F T A s . There is evidence, however, to suggest that Democrats do alter their positions. Turning to the interactions between public opinion and labor P A C contributions, Table 6 indicates that both Republicans and Democrats are affected by donations from labor P A C s . When public opinion is opposed to F T A s , the probability that either party would vote i n favor o f these agreements fall precipitously when public opinion and labor contributions are combined. The effect is even more dramatic in Table 7. Both parties are predicted with high probability to vote for F T A s when there are no labor P A C contributions. But the probabilities begin to fall rapidly after a mere $25,000 donation.  30  For Republicans, however, it should be noted that their average labor P A C contribution is only $26,000. There is evidence to validate the hypothesis that House members alter their positions when confronted with a congruence o f popular opposition and labor P A C contributions against F T A s . In summary, taking the logit regressions and simulations together, the congressional behavior o f Democrats on F T A s tends to support Interest Group Theory. These congressmen are sensitive to public opinion, but their positions can be altered by both business and labor P A C s . Democrats do not delegate authority to the President on trade policy. The congressional behavior o f Republicans, on the other hand, tends towards Delegation Theory. These congressmen are not very responsive to public opinion on F T A s , and business P A C s have little influence over their position on trade liberalization. Republicans do delegate authority to the President on trade policy.  C l o s i n g Comments The passage o f seven trade agreements was never in any doubt. The elaborate procedures o f Fast-Track have an institutional bias in favor o f a F T A ' s passage once the initial hard step has been taken to negotiate under these rules. In accordance with Delegation Theory, House Republicans are consistently more supportive o f the President on trade policy, even when their constituents' views on F T A s tend towards popular opposition. N o such support is given to Bush from the House Democrats, and these members are sensitive to swings in public opinion. It should be noted, however, that the roll call votes are on government bills. Both the President and the congressional leadership had a strong influence in securing the  31  passage o f the F T A s into law. A s Jacobs and his colleagues have said, it is the congressional leadership - and not the individual members o f Congress - who respond to public opinion. Incumbency rates alter the voting behavior on F T A s for House Democrats only, with veteran Democrats being slightly less receptive to Bush's trade liberalization policy. O n the other hand, incumbency rates do not change the significance that public opinion has on House F T A votes. The results for the Senate, though, reveal that public opinion has no effect. A s noted by Hinton-Andersson and Wood, the combination o f incumbency and the staggered election cycle can make Senators impervious to public opinion. A s has been discovered i n other studies, P A C contributions have an impact on roll call votes. The degree o f influence varies, though, on both what type o f P A C is making the political contribution, and what political party is receiving the donation. A l s o , there are significant differences in the impact that P A C s have on Congress; P A C s have no influence in the Senate. Furthermore, P A C s have heterogeneous motives for providing campaign contributions. The influence o f P A C s , as suggested i n the House model, is relatively small when compared to public opinion. In accordance with Interest Group Theory, Democrats are highly sensitive to interests as expressed as public opinion and as campaign contributions. De Bievre and Diir are half correct to surmise that public opinion and interest group activity would prove to be a powerful interaction. In the case examined here, only House Democrats are sensitive to these influences. Looking at factor mobility, the class-based characteristics o f the districts do not alter the voting behavior on F T A s . Congressmen do not incorporate the income or the  32  education levels o f their constituencies when determining to vote for or against F T A s . McGillivray's variables on trade exposure and the unemployment rate, however, did prove significant for the House votes. The question first asked at the beginning o f the research design section can be answered: Given the heterogeneous nature o f public opinion on trade policy, is it plausible to think public opinion matters on such a complex issue? The answer is yes. Public opinion can matter under certain conditions, and those conditions can help researchers better interpret their results when analyzing trade policy. Researchers should think about adding public opinion as a determinant to consider i n their own future studies.  33  Table 1. House roll call votes on F T A s in the U . S . Congress Variable / Intercept  Pr ( Roll Call Vote = 1) Coefficient  Public Opinion Lack of Opinion Party Rep. Incumbency Dem. Incumbency Business P A C Labor P A C PO*Business P A C PO* Labor P A C Per Capita Median Household Unemployment Education Export Exposure Union Strength Australia Intercept Bahrain Intercept Chile Intercept C A DR Intercept Morocco Intercept Oman Intercept Singapore Intercept N Log Pseudo4ikelihood Wald Chi-Squared Pseudo R-Squared  1.374** 4.367 0.857** 0.021 -0.066**** 0.004**** —0 009**** 0.004** -0.012*** 0.000 0.000 —0 ]92*** 0.021 0.109*** 0.013 0.651 0.167 -0.804**** -2.018**** 0.178 —1 9]4**** -0.823****  (Robust Std. Error) (0.676) (3.178) (0.403) (0.021) (0.015) (0.001) (0.001) (0.002) (0.004) (0.000) (0.000) (0.060) (0.024) (0.036) (0.017) (0.500) (0.173) (0.180) (0.209) (0.160) (0.209) (0.175)  Odds Ratio 3.950 78.751 2.356 0.938 1.021 1.004 0.991 1.004 0.988 1.000 1.000 0.825 1.022 1.115 1.013  (Robust Std. Error) (2.670) (250.300) (0.950) (0.014) (0.022) (0.001) (0.001) (0.002) (0.004) (0.000) (0.000) (0.050) (0.024) (0.040) (0.017)  2830 - 1107.054 361.57**** 0.3985  * denotes p values <=0.1; ** p values <=0.05; *** p values <=0.01; **** p values <=0.001  34  Table 2. Senate roll call votes on F T A s i n the U . S . Congress Variable / Intercept Pr ( Roll Call Vote = 1) Coefficient Public Opinion Lack of Opinion Party Rep. Incumbency Dem. Incumbency Business PAC Labor PAC PO*Business PAC PO*Labor PAC Per Capita Median Household Unemployment Education Export Exposure Union Strength Australia Intercept Bahrain Intercept Chile Intercept CA DR Intercept Morocco Intercept Oman Intercept Singapore Intercept  0.422 -1.248 0.713 0.026 -0.052*** 0.001 -0.001 0.002* -0.008 0.000 0.000 0.321* 0.090 0.042 -0.033 2.310 NA -1.141*** f  —2 190****  0.266 — 1 572****  -1.275***  N Log Pseudo-likelihood Wald Chi-Squared Pseudo R-Squared  (Robust Std. Error) (2.006) (12.665) (0.726) (0.036) (0.017) (0.000) (0.001) (0.001) (0.005) (0.000) (0.000) (0.175) (0.074) (0.068) (0.043) (1.823) NA (0.417) (0.386) (0.414) (0.378) (0.435)  Odds Ratio 1.525 0.287 2.041 1.026 0.949 1.001 0.999 1.002 0.992 1.000 1.000 1.378 1.094 1.043 0.967  (Robust Std. Error) (3.060) (3.634) (1.481) (0.037) (0.017) (0.000) (0.001) (0.001) (0.005) (0.000) (0.000) (0.241) (0.081) (0.071) (0.042)  t  560 - 241.212 79.68**** 0.2760  * denotes p values <=0.1; ** p values <=0.05; *** p values <=0.01; **** p values <=0.001 t passed by unanimous consent  35  Table 3. Predicted probabilities o f voting i n favor o f F T A s , by political party, at differing levels o f public opinion; House roll call votes on F T A s i n the U . S . Congress Public Opinion  Republicans P r ( R o l l Call Vote = 1) Coefficient  Strongly favor Somewhat favor Neutral Somewhat oppose Strongly oppose Value of Pivot Point  0.999 0.996 0.963 0.645 0.182  (Standard Error) (0.001) (0.004) (0.185) (0.144) (0.173)  - 1.26  Democrats Pr( Roll Call Vote = 1) Coefficient 0.706 0.622 0.506 0.388 0.300 -0.05  36  (Standard Error) (0.163) (0.107) (0.042) (0.102) (0.161)  Table 4. Predicted probabilities for the conditional impact o f business P A C contributions, by political party, at differing levels o f public opinion; House roll call votes on F T A s in the U . S . Congress Public Opinion  Strongly favor Somewhat favor Neutral Somewhat oppose Strongly oppose  Conditional on Republican Pr ( Roll Call Vote = 1 )*  Conditional on Democrat Pr ( Roll Call Vote = 1 )*  Coefficient  (Standard Error)  Coefficient  0.999 0.999 0.983 0.732 0.209  (0.001) (0.002) (0.005) (0.135) (0.201)  0.998 0.988 0.914 0.529 0.155  (Standard Error) (0.005) (0.010) (0.030) (0.157) (0.153)  X mean values, in U.S. dollars, for business P A C contributions; Republicans ($457,000), Democrats ($265,000)  Table 5. Predicted probabilities for the conditional impact o f public opinion, by political party, at differing levels o f business P A C contributions; House roll call votes on F T A s i n the U . S . Congress Business P A C Contributions  Conditional on Republican Pr (Roll Call Vote = 1) §  U.S. $ thousands  Coefficient  0 125 350 600 1000  0.888 0.928 6.969 0.988 0.997  (Standard Error) (0.030) (0.019) (0.008) (0.004) (0.001)  Conditional on Democrat Pr( Roll Call Vote = 1) §  Coefficient 0.784 0.857 0.936 0.975 0.995  (Standard Error) (0.057) (0.043) (0.024) (0.012) (0.004)  § mean values, on a scale from (-2) to (2), for public opinion; Republican-held districts (-.0666), Democrat-held districts (-.0108)  37  Table 6. Predicted probabilities on the conditional impact o f labor P A C contributions, by political party, at differing levels o f public opinion; House roll call votes on F T A s in the U . S . Congress Public Opinion  Conditional on Republican Pr ( Roll Call Vote = 1)" Coefficient  Strongly favor Somewhat favor Neutral Somewhat oppose Strongly oppose  0.977 0.964 0.919 0.780 0.563  (Standard Error) (0.041) (0.028) (0.021) (0.117) (0.252)  Conditional on Democrat Pr ( Roll Call Vote = 1)" Coefficient 0.738 0.706 0.629 0.518 0.435  (Standard Error) (0.224) (0.146) (0.061) (0.175) (0.268)  || mean values, in U.S. dollars, for labor P A C contributions; Republicans ($26,000), Democrats ($149,000)  Table 7. Predicted probabilities for the conditional impact o f public opinion, by political party, at differing levels o f labor P A C contributions; House roll call votes on F T A s in the U . S . Congress Labor P A C Contributions  Conditional on Republican Pr ( Roll Call Vote = 1)*  U.S. $ thousands  Coefficient  0 25 125 250 400  0.928 0.914 0.829 0.647 0.378  (Standard Error) (0.019) (0.022) (0.047) (0.099) (0.138)  Conditional on Democrat P r ( R o l l Call Vote = 1)* Coefficient 0.855 0.827 0.676 0.423 0.177  (Standard Error) (0.044) (0.047) (0.058) (0.071) (0.063)  # mean values, on a scale from (-2) to (2), for public opinion; Republican-held districts (-.0666), Democrat-held districts (-.0108)  38  Pr ( R o l l Call = l | x )  T T  =  G [ 0 i * Public Opinion + p * Lack o f Opinion + 2  P 3 * Party + P 4 * Republican Incumbency + P 5 * Democrat Incumbency + P  6  p  8  * Business P A C + p 7 * Labor P A C + * Public Opinion * Business P A C + p 9 * Public Opinion * Labor P A C +  P 10 * Per Capita + P n * Median Household + P 12 * Unemployment + P 13 * Education + p 14 * Export Exposure + P 15 * U n i o n Strength + S P j * Bill ] + e  F i g u r e 1. Formal model for roll call votes t f The model is duplicated twice, once for the House and once for the Senate  39  Republican House Vote on F T A Implementation Acts  o.o public_opinion Nay  •  Yea  Democrat House Vote on F T A Implementation Acts  1.0  CO CD  "co 0.5 £1  9 CL  0.0 public_opinion Nay  Yea  Figure 2. Predictive probability plots of House roll call votes, by political party, at differing levels o f public opinion  40  LIST OF R E F E R E N C E S  A l t , James, Michael Gilligan, Dani Rodrik, and Ronald Rogowski. 1996. The political economy o f international trade: Enduring puzzles and an agenda for inquiry. Comparative Political Studies 29, no. 6 (December) 689-717. Annenberg Public Policy Center. 2006. National Annenberg Election Survey ( N A E S ) 2004: Cross sections; national rolling. C D - R O M . In Capturing campaign dynamics, 2000 and 2004: The National Annenberg Election Survey, ed. Daniel Romer, Daniel, Christopher Adasiewicz, Kathleen H a l l Jamieson, Kate Kenski, and Kenneth Winneg. Philadelphia: University o f Pennsylvania Press. Baldwin, Robert, and Christopher Magee. 2000a. Congressional trade votes: From NAFTA approval to fast-track defeat. Policy Analyses in International Economics (February 2000). Institute for International Economics: Washington, DC. Balint, Peter, and I . M . (Mac) Destler. 1999. The new politics of American trade: Trade, labor, and the environment. Policy Analyses in International Economics (October 1999). Institute for International Economics: Washington, D C . Bartels, Larry. 1991. Constituency opinion and congressional policy making: The Reagan defense buildup. American Political Science Review 85, no. 2 (June): 457-474. Beaulieu, Eugene, and Christopher Magee. 2004. Four simple tests o f campaign contributions and trade policy preferences. Economics and Politics 16, no. 2 (July): 163-187. Bergsten, Fred. 2002. A renaissance for U . S . trade policy? Foreign Affairs 81, no. 6 (November-December): 86-98. Biglaiser, Glen, David Jackson, and Jeffrey Peake. 2004. Back on track: Support for presidential trade authority in the House o f Representatives. American Politics Research 32, no. 6 (November): 679-697. Brady, David, Brandice Canes-Wrone, and John Cogan. 2002. Out o f step, out o f office: Electoral accountability and House members' voting. American Political Science Review 96, no. 1 (March): 127-140. Burstein, Paul, and A p r i l Linton. 2002. The impact o f political parties, interest groups, and social movement organizations on public policy: Some recent evidence and theoretical concerns. Social Forces 81, no. 2 (December): 381-408. Burstein, Paul. 2003. The impact o f public opinion on public policy: A review and an agenda. Political Research Quarterly 56, no. 1 (March): 29-40.  41  Busch, Marc, and Eric Reinhardt. 2000. Geography, international trade, and political mobilization i n U S industries. American Journal of Political Science 44, no. 4 (October): 703-719. Center for Responsive Politics. OpenSecrets. Members o f Congress. http://www.opensecrets.org/politicians/index.asp (last accessed July 2007). . OpenSecrets. The big picture: Reelection rates over the years. http://www.opensecrets.org/bigpicture/reelect.asp?cycle=2004 (last accessed August 2007). Cutler, Fred. 2007. Context and attitude formation: Social interaction, default information, or local interests? Political Geography 26, no. 5 (June): 575-600. De Bievre, Dirk, and Andreas Diir. 2005. Constituency interests and delegation in European and American trade policy. Comparative Political Studies 38, no. 10 (December): 1271-1296. De Boef, Suzanna, and James Stimson. 1995. The dynamic structure o f congressional elections. Journal of Politics 57, no. 3 (August): 630-648. Downs, Anthony. 1957. An Economic Theory of Democracy. N e w York: Harper. Quoted in Brady, David, Brandice Canes-Wrone, and John Cogan. 2002. Out o f step, out o f office: Electoral accountability and House members' voting. American Political Science Review 96, no. 1 (March): 127. Falke, Andreas. 1996. American trade policy after the end o f the C o l d War. In American impasse: US domestic and foreign policy after the Cold War, ed. Michael Minkenberg and Herbert Dittgen, 264-297. Pittsburgh: University o f Pittsburgh Press. Finger, Michael, and Anne Harrison. 1996. Import protection for U S textiles and apparel: Viewed from the domestic perspective. In Political economy of trade protection, ed. Anne Krueger, 43-50. Pittsburgh: University o f Pittsburgh Press. Hinton-Andersson, Angela, and Dan Wood. 1998. The dynamics o f senatorial representation, 1952-1991. Journal of Politics 60, no. 3 (August): 705-736. Hiscox, Michael. 2002a. Commerce, coalitions, and factor mobility: Evidence from congressional votes on trade legislation. American Political Science Review 96, no. 3 (September): 593-608. . 2002b. International trade and political conflict: Commerce, coalitions, mobility. Princeton, N J : Princeton University Press.  42  and  . 2004. International capital mobility and trade politics: Capital flows, political coalitions, and lobbying. Economics and Politics 16, no. 3 (November): 253-285. . 2006. Through a glass and darkly: Attitudes toward international trade and the curious effects o f issue framing. International Organization 60, no. 3 (summer): 755-780. Jacobs, Lawrence, Eric Lawrence, Robert Shapiro, and Steven Smith. 1998. Congressional leadership on public opinion. Political Science Quarterly 113, no. 1, (spring): 21-41. Kaempfer, W i l l i a m , and Stephen Marks. 1993. The expected effects o f trade liberalisation: Evidence from U S congressional action on fast-track authority. World Economy 16, no. 6 (November): 725-740. M c G i l l i v r a y , Fiona. 1997. Party discipline as a determinant o f the endogenous formation of tariffs. American Journal of Political Science 41, no. 2, (April): 584-607. . 2004. Privileging industry: The comparative politics of trade and policy. Princeton, N J : Princeton University Press.  industrial  Moore, Michael. 1996. The rise and fall o f big steel's influence on U S trade policy. In Political economy of trade protection, ed. Anne Krueger, 15-34. Pittsburgh: University o f Pittsburgh Press. Nelson, Douglas. 1996. M a k i n g sense o f the 1981 automobile V E R : Economics, politics, and the political economy o f protection. In Political economy of trade protection, ed. Anne Krueger, 35-42. Pittsburgh: University o f Pittsburgh Press. N i v o l a , Pietro. 1993. Regulating unfair trade. Washington, D C : Brookings Institution. O'Halloran, Sharyn. 1993. Congress and foreign trade policy. In Congress resurgent: Foreign and defense policy on Capital Hill, ed. Randall Ripley and James Lindsay, 283-303. A n n Arbor, M I : University o f Michigan Press. . 1994. Politics, process, and American trade policy. A n n Arbor, M I : University of Michigan Press. Poole, Keith. Voteview: University o f California, San Diego. Nominate data, roll call data, and software. R o l l call data, http://www.voteview.com (last accessed June 2007). Ripley, Randall, and James Lindsay, eds. 1993. Congress resurgent: Foreign and defense policy on Capital Hill. A n n Arbor, M I : University o f Michigan Press.  43  Romer, Daniel, Christopher Adasiewicz, Kathleen H a l l Jamieson, Kate Kenski, and Kenneth Winneg. 2006. Capturing campaign dynamics, 2000 and 2004: The National Annenberg Election Survey. Philadelphia: University o f Pennsylvania Press. Scheve, Kenneth, and Matthew Slaughter. 2001a. Globalization and the perceptions of American workers, Globalization Balance Sheet Series (March 2001). Institute for International Economics: Washington, D C . . 2001b. What determines individual trade-policy preferences? Journal of International Economics 54, no. 2 (August): 267.292. Sherman, Richard. 2002. Delegation, ratification, and U . S . trade policy: W h y divided government causes lower tariffs. Comparative Political Studies 35, no. 10 (December): 1171-1197. Tierney, John. 1993. Interest group involvement in congressional foreign and defense policy. In Congress resurgent: Foreign and defense policy on Capital Hill, ed. Randall Ripley and James Lindsay, 283-303. A n n Arbor, M I : University o f Michigan Press. U . S . Congress. Biographical directory o f the United States Congress, 1774 to present. http://bioguide.congress.gov/biosearch/biosearch.asp (last accessed July 2007). U . S . Library o f Congress: Thomas home. R o l l call votes, http://thomas.loc.gov/home/ rollcallvotes.html (last accessed June 2007). Zhang, Daowei, and David Laband. 2005. From Senators to the President: Solve the lumber problem or else. Public Choice 123, nos. 3-4 (June): 393-410.  44  A P P E N D I X A : LIST OF D A T A SOURCES  For the variables Public Opinion and Lack o f Opinion, the mean value for the district or the state was taken from a C D - R O M produced by the Annenberg Public Policy Center. There is no public opinion data for either Hawaii or Alaska. Annenberg Public Policy Center. 2006. National Annenberg Election Survey ( N A E S ) 2004: Cross sections; national rolling. C D - R O M . In Capturing campaign dynamics, 2000 and 2004: The National Annenberg Election Survey, ed. Daniel Romer, Daniel, Christopher Adasiewicz, Kathleen H a l l Jamieson, Kate Kenski, and Kenneth Winneg. Philadelphia: University o f Pennsylvania Press.  For the P A C contribution variables, the actual values are taken from the OpenSecrets website. O n the rare occasion when negative values were given, these values were recoded as zero. For Senators John Kerry and John Edwards, the P A C contributions towards their respective presidential campaigns in 2004 were omitted from the Senate dataset. Center for Responsive Politics. OpenSecrets. Members o f Congress. http://www.opensecrets.org/politicians/index.asp (last accessed July 2007).  For the dependent variables and the party affliation variables, the actual values are taken from the Voteview website. Poole, Keith. Voteview: University o f California, San Diego. Nominate data, roll call data, and software. R o l l call data, http://www.voteview.com (last accessed June 2007).  For the Union Strength variable, the actual values are taken from the Labor Statistics' website. U . S . Bureau o f Labor Statistics. Labor force statistics from the current population survey. Household data series: U n i o n membership tables, http://www.bls.gov/ webapps/legacy/cpslutabl.htm (last accessed July 2007).  For the Education, Export Exposure, Median Household, Per Capita, and Unemployment variables, the actual values are taken from the Census Bureau's website. U . S . Bureau o f the Census. American community profiles. FactFinder: datasets. Data profiles. http://factfinder.census.gov/home/saff/main.html?_lang=en (last accessed July 2007).  45  -. American community profiles. FactFinder: datasets. Detailed tables. http://factfinder.census.gov/home/saff/main.html?_lang=en (last accessed July 2007). -. Foreign trade statistics. State by 6-digit H S code and top countries: Total (state) exports, http://www.census.gov/foreign-trade/statistics/state/data/index.html (last accessed July 2007).  A final note with respect to the data for Texas. Due to mid-decade redistricting in this state, some o f the measurements for the public opinion and demographic variables may be unreliable. A s a result, the House roll call votes for Texas in the 108 Congress has been omitted from the regression analysis. th  46  A P P E N D I X B : DESCRIPTIVE STATISTICS OF T H E V A R I A B L E S  This appendix consists o f three tables on most o f the main variables tested in this paper. Party has been omitted. Table 8 contains the descriptive statistics for the House; Table 9 contains the descriptives for the Senate. Table 10 contains the top ten list for the strongest intensity in the Public Opinion variable, both for and against F T A s .  Table 8. Descriptive statistics for the House o f Representatives Variable  Mean  Range  Public Opinion  0.405  1.475  -0.876  0.599  -0.478  0.378  Lack of Opinion  0.133  0.218  0.041  0.259  0.081  0.190  Rep. Incumbency  10  35  1  36  1  25  Dem. Incumbency  12  51  1  52  1  29  367  2247  0  2247  44  909  Labor P A C  84  791  0  791  0  255  Per Capita  0  51170  -14249  36921  -8906  12819  Median Household  0  75768  -27224  48544  -14922  26949  Unemployment  0  13.7  -3.3  10.4  -2,5  3.7  Education  0  56.2  -20.8  35.4  -13.1  18.4  Export Exposure  0  17.3  -6.3  11.0  ^4.0  5.7  Union Strength  0  23.8  -10.2  13.6  -8.0  11.7  Business P A C  Minimum  47  Maximum  5 Percentile th  95 Percentile th  Table 9. Descriptive statistics for the Senate Variable  Public Opinion  Mean  Range  Minimum  Maximum  5 Percentile th  95* Percentile  -0.099  0.717  -0.470  0.248  -0.377  0.173  0.136  0.061  0.110  0.171  0.111  0.161  Rep. Incumbency  16  41  1  42  2  32  Dem. Incumbency  21  53  1  54  3  . 45  1242  5686  0  5686  10  2768  139  1015  0  1015  0  480  Per Capita  0  15978  -7064  8914  -5710  6074  Median Household  0  28734  -13304  15430  -11243  14699  Unemployment  0  5.5  -3.1  2.4  -2.1  2.0  Education  0  20.0  -10.3  9.7  -8.3  7.7  Export Exposure  0  17.3  -6.3  11.0  ^1.5  5.9  Union Strength  0  23.8  -10.2  13.6  -7.9  9.4  Lack of Opinion  Business P A C Labor P A C  48  Table 10. Top ten districts in favor of, and oppose to, F T A s ; Public Opinion variable District  Mean  Incumbent Party  Geographic or Metropolitan Area  A Major Population Area  1  Virginia 11th  0.599  Republican  Washington, D C  Fairfax  2  Texas 32nd  0.565  Republican  Dallas  Irving  3  New Jersey 13th  0.542  Democrat  New York City  Hoboken  4  Virginia 8th  0.527  Democrat  Washington, D C  Arlington  5  Texas 29th  0.525  Democrat  Houston  Pasadena  6  Texas 15th  0.523  Democrat  Southern State  Edinburg  7  Texas 25th  0.505  Democrat  Southern State  Austin  8  New York 18th  0.496  Democrat  New York City  White Plains  9  Virginia 10th  0.492  Republican  Washington, D C  Fairfax  10  California 43rd  0.474  Democrat  Los Angeles  San Bernardino  1  N Carolina 10th  - 0.876  Republican  Western State  Hickory  2  N Carolina 11th  -0.708  Republican  Western State  Asheville  3  Mississippi 3rd  - 0.675  Republican  Central State  Jackson  4  N Carolina 5th  -0.655  Republican  Western State  Statesville  5  N Carolina 2nd  - 0.648  Democrat  Central State  Raleigh  6  Ohio 9th  - 0.644  Democrat  Lake Erie Region  Toledo  7  Tennessee 4th  -0.621  Democrat  Central State  Franklin  8  Michigan 5th  - 0.600  Democrat  Eastern State  Flint  9  Kentucky 5th  -0.582  Republican  Eastern State  Corbin  10  Ohio 16th  -0.564  Republican  Eastern State  Canton  49  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0101115/manifest

Comment

Related Items