A coalition approach to trade policymaking in the United States: the fast-track authority fiasco of 1997 and the approval of Permanent Normal Trade Relations (PNTR) with China in 2000

This article proposes a framework centered on coalitions between the executive, the congress and interest groups for the analysis of the US trade policymaking process. Such a framework is focused on two main concepts, “willingness to bargain” and “pivotal actor”. The work advances two main hypotheses related to the ability of pro-free-trade actors to come to an agreement and anti-free-trade actors to effectively oppose trade liberalization. Two case studies are undertaken in order to check the plausibility of such hypotheses: the fast-track fiasco (1997) and the approval of Permanent Normal Trade Relations (PNTR) with China (2000). distance among actors’ preferences within the deadlock coalition, the more/less effective are the coalition’s efforts to oppose trade liberalization


Introduction
T he study of the domestic determinants of foreign policy preferences has consolidated as a reflection of the dissatisfaction with structural explanations (Waltz 1979;Krasner 1976). Those studies did not consider the need to take into account gains and losses at the domestic level, considering the state a unitary actor and excluding the entire political environment that surrounds the decision-making process within national borders. Nowadays, the study of the role played by domestic politics in the IR literature gained so much support that Bruce Bueno de Mesquita (2002: 1) drew attention to the, "urgency of refocusing our efforts on leaders and domestic affairs as the centerpiece for understanding the world of international relations".
However, there are still limitations in domestic politics analyses focused on domestic institutions and preferences (Alt and Copyright: • This is an open-access article distributed under the terms of a Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided that the original author and source are credited. • Este é um artigo publicado em acesso aberto e distribuído sob os termos da Licença de Atribuição Creative Commons, que permite uso irrestrito, distribuição e reprodução em qualquer meio, desde que o autor e a fonte originais sejam creditados. Gilligan 2000; Milner 1997). Most of the recent trade policy works are too focused on theoretical parsimony and generalization, relying more and more on very disaggregated variables (Cohen 2008). For instance, one of the most known approaches to the study of trade policy, to which the understanding of how openness to international trade affects domestic alignments is central, (Rogowski 1989;Alt and Gilligan 2000;Frieden and Rogowski 1996), does not take into consideration the preferences of the executive and the congress and focus solely on specific domestic civil society preferences. Although this model is focussed upon understanding the formation of domestic coalitions, it does not emphasize transversal coalitions among political actors of distinct levels, which, we claim, are crucial for understanding the "big picture" of a fragmented trade policymaking process.
In view of that, looking for new ways to analyze the domestic determinants of trade policy decisions is an opportunity to be seized. A promising path to be followed is to focus on domestic political coalitions among multilevel players. This work aims to do so by adapting a framework that uses elements of the decision-units model (Hermann 2001;Hagan et al 2001), to the realm of trade policy. To do that, we first make a brief presentation of the state-of-art of approaches dealing with coalitions in the IR and Political Science literature, and present the elements of our adapted coalition-based approach. We then present "plausibility probes" (George and Bennett 2005) of the hypotheses generated by using the case studies of Clinton's failed attempt to renew the fast-track authority in 1997 and the approval of Permanent Normal Trade Relations (PNTR) with China in 2000.

A coalition-based approach in a fragmented decision-making process
There is an intricate relation between the actors participating in the US trade policymaking process. According to Alice Rivlin -Chair of the US Office of Management and Budget between October 1994 and April 1996 -, as quoted by Destler (1996, 2), the trade decision-making process, "… is so fragmented and complicated that it is almost impossible to explain to the uninitiated how it is supposed to work, let alone how it does work". Cohen (2000), in turn, considers the US trade decision-making process extremely fragmented and complex. That occurs because, not only within civil society, but also within the government itself, there is a plethora of divergent preferences that cannot be all taken into account at the same time (Gibson 2000). Table 1 gives a clear dimension of that fragmentation by presenting the bureaucratic agencies with foreign economic policy departments. This enormous fragmentation contributed to the creation of the National Economic Council in 1993, an organism serving as an "honest broker" committed to find a common denominator among different agencies with stakes in economic policy (Destler 1996).
The institutional setting of USTR -United States Trade Representative -also contributes to the multitude of actors accessing the US trade policy-making process. USTR is the main US trade Given the fragmentation and complexity of the US trade policymaking process, we assume that in order for a given trade policy to be approved agreement, or at least acquiescence, among the most important actors of the decision-making process is necessary (Hagan et al 2001). In other words, it is necessary to seek explanations focused on the creation of coalitions among the main actors in the US trade decision-making process. Since all main actors may become a veto player, "[there is] no single group or actor with the political autonomy to commit the public resources (Hagan et al 2001, 170).
Approaches focused on coalitions are not new in the IR and Political Science literature (see Table 2) and one of the main approaches to the study of trade policy is actually centered on the study of coalitions, as pointed out in the introduction. However, some of these approaches are either not usual in trade policy works or are not able to account for coalitions among multilevel actors, in our case, interest groups, Congress and executive agencies. The Bureaucratic Politics Model (BPM) itself emphasizes the forging of coalitions at the bureaucratic level (Allison and Halperin 1972). These informal coalitions are a result of concessions and bargaining among the main actors involved in the policy-making process. However, the BPM does not take into consideration the importance of actors other than the bureaucracy in the process of forging coalitions for the formulation/approval of a given policy (Art 1973;Rosati 1981). Another approach focussed upon coalition-building processes is the iron triangles approach. The main assumption of the framework is that there is a relation among interest groups, bureaucracies, and legislative committees, each of them with the objective of consolidating and maximizing its own power. Although the approach is focused on multilevel actors, it is often considered a metaphor instead of a well-knit theoretical model (Vernon, Spar and Tobin 1991;Heclo 1990). The iron triangle literature has been constantly criticized by policy network authors. For Hugh Heclo (1990, 444), for instance, "the iron triangle concept is not so much wrong as it is disastrously incomplete". For Heclo, the US political system cannot be adequately studied through a rigid relation between interest groups, bureaucrats and Congress committee. In other words, "looking for the close triangles of control, we tend to miss the fairly open networks of people that increasingly impinge upon government" (Heclo 1990, 444).
A specific type of policy network approach is the Advocacy Coalition Framework (ACF). This analytical framework, according to Sabatier (1989), considers advocacy coalitions as a set of informal and institutional linkage among governmental and other actors, structured around shared beliefs in the process of formulation and implementation of public policies. In spite of ACF's potential, the timespan of the analysis limits its applicability, since the framework is more useful for medium and long-range processes. This reflects on the fact ACF does not take into consideration the possibility of "coalitions of convenience", those based on short-term bargains and concessions.
Differently from previous trade policy works, this paper seeks to understand the elements that influence the formation of minimum winning coalitions -the minimum size of a coalition in order for a given policy to be approved -in the specific realm of the US trade policy-making process. This approach goes beyond the analysis of coalitions within Congress and includes other "autonomous" actors within and outside government. Hagan et al (2001) deserve credit for advancing the use of the theory of coalitions in the realm of foreign policy, through development of the decision units approach (Hermann 2001). However, such a framework has yet to be adapted to specific subsets of foreign policies, including trade policy.
Unlike the policy networks approach, this framework is centered on the most resourceful actors of the decision-making process, essentially because we understand that they constitute the minimum possible set of actors that must be taken into account for a certain policy to be approved. Thus, this work avoids the empirical and methodological complications that arise when too fluid networks of actors are investigated. Moreover, this article considers that neither the Congress nor the executive are unitary actors, an assumption that pursues a different path than the one followed by classic trade policy works (see, for instance , Milner 1997;Lohman and O'Halloran 1994). Instead, it acknowledges the possibilities of divisions within bureaucracy (Department of Commerce and USTR, for instance) and intraparty divisions (liberal and moderate Democrats, internationalist and conservative Republicans).
The formation of coalitions depends on the degree of agreement between political actors. That is why the two main variables to be used here are the ones that explain agreement among "autonomous" and contentious actors in the policy process. In that sense, "willingness to bargain" evaluates if there are "serious a priori constraints on parties which make them hesitant to negotiate or strike bargains" (Dodd 1976apud Hagan et al 2001. This concept is of relevance given that seldom are decisions based on consensus, particularly because opponents will bargain their support in exchange of political or economic benefits in the form of pork-barrelsmetaphor for projects and favors for legislators' districts - (Evans 2004). When there is willingness to bargain, "advocates of a policy may be able to buy off a strong dissenter with concessions critical to them on another issue" (Hagan et al 2001, 176).
In opposition, the absence of willingness is the result of, "extreme distrust between parties, immediate competition for control of the government, opposition to agreements from factions within the coalition parties" (Hagan et al 2001:, 177). Extreme distrust and intense competition can make actors define a policy problem as a "zero-sum" issue. As to the question regarding when actors are willing to bargain, which influences the question as to when actors can offer such concessions, we initially focus on three main elements: (1) the pay-off of supporting a given policy; (2) political polarization and (3) level of opposition among distinct preferences in the civil society. Such elements as roughly drawn from recent US trade policy events (for instance , Destler 2005;Destler and Balint 1999;Mayer 1998;Cohen, Paul and Becker 2003).
The second element concerns the existence of a pivotal actor in the decision-making process. Classically, a pivotal actor assumes such a role, "when the absolute difference between the combined votes (weights) of members on his right and of members on his left is not greater than his own weight" (De Swaan 1973, 89, apud Hagan et al 2001. In terms of the adaption made by Hagan et al (2001), a pivotal actor can be understood as the one capable of tipping the balance in favor of one side or another, although it cannot dominate decision-making process. A pivotal actor can take many forms: a political party, the President, an interest group, etc. (Hagan et al 2001). If a pivotal actor does not take a specific side in the process, it may still act as a broker and mediate conflicting preferences (Çuhadar-Gürkaynak and Özkeçeci-Taner 2004). Such a concept begs the question: when does an agent assume the role of pivotal actor? We focus on three conditions in this work: (1) relevant impact in the policy process; (2) high costs of non-participation; (3) availability of resources to assume a pivotal role. We consider that these three elements offer a good range of conditions according to the relevance of an actor, its willingness to assume a pivotal role and ability to do so.
One of the premises of the minimum winning coalition literature (Riker 1962;Hagan et al. 2001) is that, in order for the government to keep its political power, it cannot spend too many resources on side payments for uncommitted (or opposing) parties. For that motive, if the preferences are too far apart because of lack of willingness to bargain or inexistence of a pivotal actor, the government will have difficulties assuring a minimum winning coalition, since its capacity of forging that coalition will fall short of encompassing all the necessary actors to get a trade bill formulated and approved.
In that regard, Illustration 1 represents the formation of minimum winning coalitions. As the dependent variable of this work is the approval of liberalizing policies, groups closer to the anti-free-trade edge will form a "deadlock coalition", and groups closer to the free-trade edge will define a possible minimum winning coalition in favor of trade liberalization. It is not the aim of this framework to predict actors' preferences, as done by works based on the Stolper-Samuelson and Ricardo-Viner models and by the bureaucratic politics model ("where you stand depends on where you sit" proposition). Thus, the position of each actor in the spatial models set out by illustration 1 is grounded on the recent US trade history. Following that rationale, AFL-CIO -American Federation of Labor and Congress of Industrial Organizations -, liberal Democrats (Dl) -the more traditional faction of the Democrat party -and the conservative Republicans are closer to the anti-free-trade edge, although for distinct reasons. Since the mid-1970, the first two took a more hesitant stance towards trade liberalization as a result of the perceived effect of free-trade on workers right (Shelton 1970), while conservative Republicans (Rc) usually are driven by nationalist arguments regarding the possible loss of autonomy by the US (Destler 2005). On the other hand, moderate Democrats (Dm) are closer to the free trade edge as a response to the need of bringing the Democrat party to a more centrist position (Hale 1995). Internationalist Republicans (Ri) are free-traders in response to their constituencies' preferences, including the Business Roundtable, for example, composed of highly competitive industries favorable of trade liberalization (Destler 2005).
In the Executive, USTR and the Department of Commerce are somewhat more hesitant to support trade liberalization, given that among their constituencies there are groups negatively affected by the relative loss of competitiveness of the US industry in the last decades. The Department of State, in turn, usually takes a position favorable of trade liberalization, in part because it does not want trade issues to endanger the foreign policy and security objectives of the US in its relationship with countries such as Japan and China (Cohen, Paul, and Becker 2003). In relation to the trade preferences of US Presidents, in the recent US trade history both Republican and Democrat presidents (Pd,r) have been positioning themselves in favor of trade liberalization, which is why we assume that their preferences do not change in response to their party affiliation. This is usually referred to as the liberal presidency thesis (Karol 2000).
Based on Hagan et al. (2001), and with the support of the illustration above, it is possible to extract the following hypothesis: The lower/higher the presence of willingness to bargain and the presence of a pivotal actor in favor of trade liberalization, the more difficult/easier to approve liberalization policies.
In the absence of willingness to bargain, a pivotal actor can be exemplified by illustration 1.a, in which the maximum size of the minimum winning coalition (pro-trade) does not encompass crucial political actors, such as moderate democrats and civil society groups, such as the Business Roundtable. On the other hand, Illustration 1.b exemplifies the effects of the presence of both variables in the formation of minimum winning coalitions, when all main actors fit the maximum size of such a coalition. The issue in question is more likely to result in failure when voted by Congress. In the same way the presence/absence of willingness to bargain/pivotal actor in favor of liberalization may increase/decrease the possibilities of reaching a minimum wining coalition, it is possible to hypothesize that:

Hypothesis 2:
The lower/higher the distance among actors' preferences within the deadlock coalition, the more/less effective are the coalition's efforts to oppose trade liberalization In other words, to the same extent that convergence of preferences in the pro-free-trade coalition is necessary for a coalition to be formed, actors' preferences in the deadlock coalition must also converge for it to be effective. Illustration 1.a presents such a convergence, while Illustration 1.b illustrates higher distance among preferences. This hypothesis adds a new element to the work of Hagan et al (2001) by presenting the prospect of a deadlock coalition, which opposes the formation of a minimum winning coalition in favor of trade liberalization.
As it is possible to note from illustration 1, our model presents the prospect of Republicans and Democrats as possible allies in the process of trade liberalization. As such, we offer an alternative to the argument that a divided government is an obstacle to trade liberalization (Milner 1997) given that a Republican majority in Congress, when associated with a Democrat president, may actually contribute to trade liberalization policies. This is one of the novelties of the way the current framework approaches Congress' preferences, based on constituency's preferences and not on incumbency (Shoch 1999) and based on the liberal presidency thesis (Karol 2000). By the same token, illustration 1 also presents the possibility of otherwise very distinct actors, such as conservative Republicans and liberal Democrats to form "coalition of conveniences" in order to fight liberalizing policies.
As to a more practical element concerning the analysis, while to our knowledge works focused on coalitions (see, for instance, Sabatier 1989) usually do not pay due attention to the divisions between the formulation stage -when a proposal is being shaped/negotiated -and voting stage -when a proposal has been finished and will be voted by the Congress -, we consider that to understand why a given policy is approved or not, investigating only the voting phase is not enough. As a timely and well-knit proposal is particularly important for the prospect of pivotal actors to take part in the voting stage, and for willingness to bargain to be present among parties, it is also necessary to investigate the formulation stage as an element affecting the formation of minimum winning coalitions.
Difficulties in the formulation stage reflect the fact that bureaucratic divisions tend to be much more acute the farther the expected voting date and, as such, without proper coordination -a pivotal actor taking the role of broker, for instance -these divisions may result in delays and in stalemates during the voting phase. This in part results from the President being more active in the debate only closer to the voting stage. Given, that bureaucratic divisions tend to fall apart close to the voting date as a result of the more active participation of the President, focusing on the voting stage solely would give a false impression of a united Executive throughout the entire policy process.

Coalitions in the United States trade policymaking process: two cases
Having presented the main components of the theoretical framework adapted from Hagan et al (2001), we now turn to applying it to two episodes during the Clinton Administration, which happened in similar contexts (Clinton administration) but with different outcomes.
(1) The fast-track authority is an instrument delegated by the Congress to the Executive which allows an agreement to be negotiated without the possibility of amendments during the voting phase. In 1997, the Clinton administration faced major hurdles trying to get the authority approved and ended up not being able to avoid the fiasco that came by the end of the year. Fast-track confers credibility to the US trade negotiating position, as it guarantees agility of the voting stage in the US Congress. While fast-track has been granted since its creation, in 1997 it faced unique challenges to be approved. Since Clinton became president, the debate surrounding the fast-track authority became increasingly polarized, particularly after the approval of NAFTA. Actually, in many ways fast-track had been considered a referendum on NAFTA (Bardwell 1997), and since the onset this association affected the political actors´ willingness to bargain, polarizing the debate. Thus, in spite of the interest in getting the instrument approved, the administration knew it would have a hard time getting fast-track passed Congress. As pointed out by a congressional adviser, NAFTA's political heritage was much bigger than its economic heritage, and made room for the rise of an anti-fast-track movement, "much more intense that anything we would have seen if there was not a NAFTA" (Bluestein 1997). In that context, the need for formulating an acceptable proposal for the Congress was reinforced.
Because of the rising polarization within Congress, since the fast-track debate became more salient after January 1997, the USTR repeatedly gave preference to its substance. In other words, USTR was not that interested in the amount of time it would be necessary to invest to come up with a proposal, as long as it was considered the best possible agreement (USTR 1997). However, inside the administration, many supporters considered that there was a very short window of opportunity regarding the formulation of fast-track, and that it should be voted by the Congress as soon as possible (Kolbe 1997a). Because of USTR's initial position, fast-track took longer to be presented to the Congress. This cautious position attracted criticisms in view of a possible lack of engagement of the administration (Kolbe 1997a).
In addition, part of the Executive considered that fast-track would incur large political costs to the President and bring few economic advantages, considering that there was no relevant trade initiative that justified the use of the authority in the close future (Administration 1997). There were, thus, "different degrees of enthusiasm within the administration" (Kolbe 1997b), which points to a relative division within the Executive between those favorable of getting fast-track approved, and the ones that considered it unnecessary and even dangerous at the moment. As it will be presented, such a division fell apart once the voting stage approached. This is, we consider, a reflection of the division between the formulation and voting stages of fast-track.
Clinton himself did not give a lot of attention to fast-track in the first semester of 1997 (Orszag et al 2006) and did not play a pivotal role during the formulation stage. In July 1997, it was clear that the active participation of the President would only come by the end of the year. More than a decision, this was a direct result of the very narrow legislative agenda, given that, along with the fast-track, the administration was negotiating the 1998 Budget Plan and the renewal of the most-favored nation status to China. In that context, dealing with all these fronts was "more than the Congress could take" (Orgaz et al 2006(Orgaz et al , 1012. Moreover, for most Republicans, fast-track was not a priority in mid-1997. Clinton, in turn, had to focus in only one or two aspects at a time (Simendinger 1997). Some also said that the officials allocated to the fast-track discussion were not influential enough to call the attention of higher-ranked officials (Orszag et al 2006). In spite of all the difficulties, Destler's thesis (1996) is that the lack of an effective coordination by the National Economic Council (NEC) was the main factor that contributed to the absence of an acceptable and timely proposal. Be that as it may, it is possible to notice difficulties regarding the emergence of a pivotal actor to promote the fast-track authority.
Although the USTR participated as a broker in the process, the fact that the President did not assume the role of pivotal actor in the formulation, and that there was a lower degree of willingness to bargain, put into question the capacity of USTR to forge a common ground between parties not interested in offering concessions on fast-track. This context of low willingness to bargain was also the result of Clinton's victory in the 1996 elections. After then, the Republican party was unwilling to give more trade victories to the Democratic party. Intraparty divisions also flourished within the Democratic party in result of Clinton's "broken promises" after NAFTA. The difficulties of the administration were reinforced by the hiring of lobbyists to promote fast-track before the existence of a solid proposal. Clinton even made statements requesting that the congressional representatives did not to oppose fast-track before the final proposal. These statements, however, should not be seen as an all-out support, as Clinton was engaged in other matters. Barshefsky, in turn, had constant meetings with business representatives, who clearly pointed out that they would only support the administration after the launch of a final proposal (USTR, 1997).
The hesitation was a direct result of a lacking sense of purpose in getting fast-track approved. Although the negotiations involving the Free Trade Agreement of the Americas (FTAA) were advancing, they did not seem to suffice to convince the policy actors to support a fast-track authority. While the administration pointed out that the authority would be necessary to show the engagement of the United States in the international economy, some congressional representatives insisted that the US was important enough to attract commercial partners, even in the absence of fast-track (Destler 2005;Devereaux et al 2006). The fact that the fast-track debate did not explicitly mention one or more agreements to be negotiated under the authority diminished the chances of creating a solid sense of purpose in getting fast-track approved. As such, the context was not appropriate for the emergence of business groups as pivotal actors given the perception of lack of urgency.
If, on the one hand, the pro-trade coalition faced a number of obstacles, on the other, the deadlock coalition was reinforced by the union between labor and environmental groups and by the intraparty division within the Republican and Democratic parties. The initial AFL-CIO's wait-and-see position toward fast-track quickly escalated to a solid lobby against the authority, even before the administration had a final proposal. The lobbying effort began in March 1997 and was extended until the last possible moments before the expected date of voting. The engagement of AFL-CIO, in coalition with environmental groups such as the Sierra Club, made it difficult for the administration to pursue a "divide and conquer" tactic, which had divided environmental groups during the NAFTA battle (Mayer 1997). The anti-fast-track groups also made very good use of the framing tactics, associating fast-track with NAFTA and with other issues related to human rights abuse (Dolan 2002).
The opposition from labor unions and environmentalists was also a reverberation of the apparent resolution among some administration officials to leave aspects related to labor and environment behind. In March 1997, a memorandum written by the Subsecretary of Commerce, Stuart Eizenstat, exposed the administration's strategy for promoting fast-track in the Congress. The memorandum, which leaked, showed that the Executive was taking into consideration the promotion of a broader piece of legislation, without labor and environment provisions. The National Economic Council had a similar position (Union, 1997).
The convergence of preferences of the anti-fast-track group in association with the awful perceptions inherited by the negotiation, approval, and implementation of NAFTA made it extremely difficult for the administration to reach a compromise (Barshefsky 1997a). Regarding the opposition that came from intraparty divisions within the Republican and Democratic parties, it is possible to note an abnormal coalition comprising of the extreme right-wing faction of the Republican party -which included extreme right-wing Christian groups -and the extreme-left wing of the Democratic party (Lott 1997). What these groups had in common was that they did not support trade liberalization. However, they did so for different reasons, either because of possible impacts on the US sovereignty, for instance, or on the US worker. That awkward coalition of convenience helped to consolidate the coalition against the approval of fast-track. The deadlock coalition was thus reinforced by the fact that actors were interested in uniting with each other for a common purpose.
After the presentation of a final proposal in September 21, 1997, the bureaucratic divisions that existed were suppressed as the entire administration became engaged in lobbying the Congress. The apparent union did not hide, however, a late and baldly-coordinated proposal. Given the late proposal, the Clinton administration had only a few weeks to lobby the Congress. Furthermore, given that there was a consensus that the authority had to be voted in 1997, as the legislative agenda -including the mid-term elections -would make it unfeasible to vote fast-track in 1998, the scenario was a grim one for the administration. Neither did Clinton manage to actually take up the role of a pivotal actor because of lack of time, nor did the business groups assume such a position, as they pledged that they would only provide a more solid support after the final proposal, and did not see urgency in approving the instrument at that time. In the end, according to the administration, fast-track would fall 22 votes short of the 218 required for the approval, at best. Because of that, Clinton called Newt Gingrich (R-GA) to ask him to cancel the voting that would take place in November 11, 1997, granting a major victory to fast-track opponents.
In sum, the lack of willingness to bargain and of a pivotal actor, along with the corresponding ability of the deadlock coalition to mobilize, illustrate that hypotheses 1 and 2 are plausible in the case of fast-track. On the one hand, political and civil society polarization was high, and the president did not have enough resources -time -to actually take a pivotal role in the fast-track fight, and costs of non-participation did not seem to have been high enough for the business groups. On the other hand, the convergence of preferences in the deadlock coalition was motivated mainly by the association of fast-track with NAFTA's broken promises and Clinton's reelection. Finally, the lack of efficiency of the administration during the formulation stage contributed to delay the lobbying efforts of the president, negatively influencing the possibilities of a minimum winning coalition.

The approval of the Permanent Normal Trade Relations (PNTR) with China (2000)
For the United States to be able to engage in free trade with a foreign nation, the status of permanent normal trade relations (PNTR) -or most-favored-nation -has to be granted. Usually it is granted automatically, but until 1999 China required an annual waiver according to the Jackson-Vanik amendment, which applies for a set of countries accused of human rights abuses and restriction of freedom of emigration, among others. For the PNTR with China to be approved, the Congress would have to approve changes on the Jackson-Vanik amendment. The debate surrounding the China PNTR gained relevance as, according to the supporters of US-China trade relations, for the US to benefit from the accession of China to the WTO, PNTR would have to be approved by Congress.
With the tensions involving North Korea increasing and the anti-Americanism after the 1997 Asian financial crisis rapidly growing in the East, China consolidated itself as a key actor in US foreign policy. However, there were sensitive matters in US-China relations, particularly the trade deficit between China and the US, accusations of human rights abuse, and accusations that China influenced the 1996 US presidential election (Dolan 2002;Destler 2005). In other words, there were mixed feelings in relation to the Asian country, which could easily present problems for granting the PNTR status. Although the legislative was not directly involved in the decision of the executive to back the accession of China to the WTO, the Congress approval was required in order to modify the Jackson-Vanik amendment.
In order to assure the PNTR issue would receive the necessary attention and survive the reluctance of Congress, US companies took the initiative of "going beyond the MFN". It occurred after the creation of the "China Normalization Initiative", a set of actions at the district level organized by big companies in a decentralized fashion, with the aim of improving the trade relations with China trough grassroots tactics (Chinese official 1999). Some of the enterprises involved included Boeing, United Technologies, Nike, and Motorola. Within the government, the position of the USTR and the Department of Commerce was that there should necessarily be a careful negotiation of the terms of the agreement to grant the PNTR to China, in order to obtain concessions from that country in the occasion of its accession to the WTO. In spite of an initial division within the business sector (Chinese official 1999), the relevance of the subject for the business sector was clear, opening opportunities for their participation as pivotal actors in the process of approving PNTR.
Within the administration, Gene Sperling (Director of the National Economic Council) and John Podesta (Chief of Staff ) were also favorable of a stronger stance towards China. However, this position worried Robert Rubin, Secretary of the Treasury, and former Director of the NEC, to whom the more assertive position could potentially cause more stress to the international financial markets after the 1997 crisis (Dolan 2002). Samuel Berger, from the National Security Council, was also against a more assertive position, along with Madeleine Albright (Harris 1997), who believed that such an action could isolate China and contribute to the escalation of the tensions involving North Korea and Taiwan. These divisions within the Executive could potentially result in a badly coordinated position towards China.
Throughout 1998, an array of meetings between US and China were held in order to negotiate the accession of China to the WTO in, "commercially viable terms", but they did not result in significant breakthroughs. The absence of solid results made the division within the Executive even more pronounced, with William Daley, from the National Security Council, defending sanctions against China (Daley attacks China 1998). Other officials and agencies, as a result of the deadlock, considered the possibility defended by companies such as Boeing: to seek concessions after the accession of China to the WTO and to grant the PNTR without "asking too much" from this country. Charlene Barshefsky, however, openly stated that, "we will not hesitate to certify that we are going to be treated fairly in China" (Barshefksy says 1999). Within Congress, there were also divisions between the liberal Democrats, who were against a possible agreement with China without due bargain, and the "new" Democrats -moderates -, relatively more favorable of the agreement. Until 1999, however, the administration did not have a solid strategy to get concessions from China and to get the PNTR approved.
After the constant meetings without results, Zhu Ronghi, the prime minister of China made relevant concessions to the United States. Many officials from the USTR, the Department of State and National Security Council believed that if an agreement was not reached at that point, China would step away from many of the promises it had made. Others, such as John Podesta and Gene Sperling, affirmed that if a deal was made, a "storm of resentment" against China would resurface, particularly within Congress, and opponents would fiercely criticize Clinton's inability to push for more concessions. For them, it was necessary to have a spectacular agreement, otherwise no agreement would be approved in Congress (Bluestein 1999;Clinton 2004). In spite of the expectations, the US and China did not reach an agreement in April 1999, partially because of the division of the Executive itself and the resulting lack of an effective coordination during the formulation of the position towards China. Such an outcome throws light on the division among bureaucracies in the formulation phase of a policy and the potential effects of such conflicts in the prospect of a coherent trade policymaking.
A US negotiator, however, commented that the final result was also a reflection of the lack of pro-activity by Clinton, who did not take the position of pivotal actor at that moment.
"Some official familiarized with the negotiations affirmed that, distracted by Kosovo, Clinton did not give clear instructions to the negotiators as to how desired was the agreement with China by the end of the negotiations" (Kaiser and Mufson 1999). Others, such as Dolan (2002, 311) considered that, "Sperling and Podesta advised the President to ask for concessions that were beyond what China was willing to consider". After a reconciliation period, an agreement was reached in 15 November 1999. At that moment, the position held by Sperling and Podesta was weakened. The Democrats, however, kept on associating the accession of China to nonproliferation and human rights issues, following the well-known framing tactics. Opponents in Congress were worried about the loss of room for maneuver by approving the permanent MFN: "I believe that never again will the Congress members apply sanctions to China after the approval of the permanent MFN", said Nancy Pelosi (D-CA) (Gephardt 2000).
After the agreement was reached, the preparation for voting the changes in the Jackson-Vanik amendment took place, marking the passage to the voting stages of a China PNTR. At that moment, no overwhelming opposition towards the changes to be voted were raised within the administration. The division between Sperling and Rubin was left behind. Even some liberal Democrats were hesitant to vote against China in view of the economic benefits of its accession to WTO and thus seemed to be willing to bargain their support. In addition to the potential economic benefits that could result from the participation of China in the WTO, another element that played a crucial role in helping the supporters of the PNTR mobilize was the high costs of non-participation and the possible retaliation from China, in case the US stepped away from its commitments. This contributed to increase the solidity of the supporters, led by some big US companies such as Boeing and Motorola, that took the role of pivotal actors in lobbying the Congress and the executive.
While the business sector already had a campaign in course, AFL-CIO was still planning its own campaign to counter the business lobby. Differently from what happened during the NAFTA and the fast-track battles, US companies started lobbying before their opponents in order to assure the effectiveness of the pro-fast-track effort (Babington and Vita 2000). This is related to the fact that the coalition of PNTR opponents was not homogeneous. While labor unions were interested in entering into coalitions with other civil society organizations, the same did not seem to be the case for environmental and human rights groups. While the groups Friends of Earth and Sierra Club were against PNTR, most of the environmental groups were not interested in taking part of a possible coalition, mainly because this was not their priority at the moment, but also because there was the perception that the participation of China at WTO could actually contribute to the improvement of its environmental standards (Sierra Club 2000). In other words, the deadlock coalition was not nearly as solid as it was during the fast-track battle: lobbying efforts took longer to surface as groups were divided.
Clinton, in turn, managed to devote a larger share of his time to the approval of the PNTR, and there was a consensus that the earlier the voting took place in the year 2000, the less difficult it would be to approve PNTR. William Daley, former "NAFTA-czar" served as liaison with the US companies, particularly those within the high-tech sector. Madeleine Albright, Larry Summers, Charlene Barshefsky, Samuel Berger, and Gene Sperling met on a weekly basis to discuss the best approach to approve the PNTR (Daley pushes 2000). Clinton created a "China room" in the White house to coordinate the lobbying effort and kept himself engaged in the process. According to Samuel Berger from the National Security Council, "The effort in course is more substantial than anyone else we have seen during the seven years we have been in the White House" (Babington and Vita 2000). In other words, such an effort makes it clear that the administration was fully engaged in playing a pivotal role in getting China PNTR approved and enough resources were available for such a role to be taken. And indeed, after a much more emphatic lobbying effort than the one seen during the fast-track case, the administration brought together 164 Republicans and 73 Democrats to vote "yes" for the PNTR, assuring 237 congressional representatives in favor of PNTR (197 against).
In sum, hypotheses 1 and 2 seem plausible for analyzing the China PNTR. Although divisions in the bureaucracy resulted in delays, such divisions receded and gave space for a legislative victory. In spite of problems of coordination, an agreement has been reached and PNTR profited from a context of high willingness to bargain among supporters and opponents, given expected economic benefits and possible retaliations if the US stepped away from its promises. Such a willingness to bargain, in turn, points out that Sperling and Podesta may have misinterpreted the level of polarization surrounding the issue. In addition, while business groups were apparently willing to go as far as necessary to get PNTR approved, opponents did not want to make it a litmus test issue (Sierra 2000). Actually, the deadlock coalition was divided and preferences were not as converging as in the fast-track fight. Once an agreement had been reached, the administration had more time to lobby Congress and assumed a pivotal role. One could argue that the costs of non-participation for Clinton, in that context, had been leveraged by the previous fast-track fiasco and by the need to assure a solid trade policy legacy.

Conclusion
In a fragmented decision-making process, policy actors cannot make decisions without the support or acquiescence of other relevant players. Given that, we assume that a coalition is the decision unit that will define whether or not a given trade policy is approved (Hagan et al 2001). Having that in mind, the main objective of this work was to set out a coalition-based framework for the study of US trade policy. This approach goes in the opposite direction of the recent US trade literature which tends to focus on very disaggregated variables, seldom giving attention to relation among multilevel actors. The recent foreign policy analysis literature with focus on coalitions (Hagan et al 2001;Hermann 2001), in turn, has not paid attention to the trade policymaking process, creating the opportunity for an adaptation to that specific context. The cases presented here indicate that a framework based on the variables "willingness to bargain" and the "pivotal actor" offers an interesting alternative for analyzing the US trade policy-making process. We also consider of good use the analysis of both the voting and formulation stages of a policy, given feedback effects that influence the formation of minimum winning coalitions. It is necessary to underscore, however, that the cases presented by this work are plausibility probes rather than a full test of the hypotheses generated.
While we believe that the model offers some promising insights, we also acknowledge that it has limitations. Primarily, as it also is the case for other works using the concepts of "willingness to bargain" and "pivotal actor" (Hagan et al. 2001;Çuhadar-Gürkaynak and Özkeçeci-Taner 2004), the framework requires further evidence as to the occasions in which certain actors are more likely to assume the role of a pivotal actor. The same applies for the concept of willingness to bargain, which demands further research on the elements that make distinct actors willing to offer/accept concessions. In addition, while our approach offers an alternative to the view that a divided government is an obstacle for trade liberalization, it is necessary to define the occasions in which constituency-based preferences are more or less relevant than party opposition, for instance. In other words, our attempt to advance a trade policy approach based on coalitions can only be fully attained by further studies.