Litigation for Sale: Private Firms and WTO Dispute Escalation

This paper presents a theory of lobbying by firms for trade liberalization, not through political contributions, but instead through contributions to the litigation process at the World Trade Organization. In this “litigation for sale” model, firms signal information about the strength and value of potential cases, and the government selects cases based on firms’ signals. Firms play a key role in monitoring and seeking enforcement of international trade law by signaling information and providing a bureaucratic subsidy, which increases a state’s ability to pursue the removal of trade barriers and helps explain the high success rate for WTO complainants. The theory’s implications are consistent with in-depth interviews with 38 trade experts and are tested through an analysis of WTO dispute initiation. Acknowledgements: I have benefited from generous feedback from friends and colleagues throughout my work on this project. Special thanks goes to Christina Davis for sharing her data and many suggestions. I also benefited from feedback from Timm Betz, Marc Busch, Stephen Chaudoin, Jason Davis, Sean Gailmard, Joanne Gowa, Alexandra Guisinger, Raymond Hicks, Leslie Johns, Amanda Kennard, Andrew Little, Helen Milner, Julia Morse, Megumi Naoi, Amy Pond, Tyler Pratt, Kristopher Ramsay and those who provided feedback where earlier versions of this work were presented. I am also grateful to Marcus Albino for his work as a research assistant on this project. ⇤Assistant Professor, University of California, Berkley, Travers Department of Political Science, Email: brutger@berkeley.edu. Web: https://sites.google.com/berkeley.edu/brutger/. Given the consensus among economists that free trade is welfare enhancing, domestic interest groups are often blamed for the persistence of trade barriers. Yet even though “protection for sale” arguments have significant support,2 domestic firms also play a prominent role in maintaining the liberal trading system, monitoring states’ international trade policies, and increasing access to foreign markets. In contrast to a significant body of work that examines when and why trade barriers arise (Betz, 2017; Hosek and Peritz, 2022), this paper studies how firms and governments monitor trade barriers and select which barriers to contest. In doing so, this paper contributes to a growing body of scholarship that examines how private firms shape the development and enforcement of international law. When it comes to understanding international law, the role of private firms is critical. Firms play an influential role from the creation of law itself to the monitoring and enforcement of international law. For example, Sell (2003, 8) emphasizes that state-centric “accounts of the Uruguay Round are at best incomplete, and at worst misleading” since they obscure the role of the private sector in establishing the agenda that led to the WTO agreements. Similarly, Perlman (2018) demonstrates that private firms can use their informational advantage to shape international standards.3 Given that private firms are influential in the creation of international law (Ginsburg and Sha↵er, 2010), it should not be surprising that they also play an important role in monitoring and enforcing those laws. For example, private actors can contribute to what Morse (2019) refers to as “market-enforcement,” whereby private actors alter their behavior, e↵ectively punishing states that don’t comply with international standards (Morse, 2022). When private firms engage in the monitoring and enforcement of international law, they also shape how the law is interpreted, which is especially influential at the World Trade Organization (WTO), where the need for consensus makes it challenging to alter the rules through negotiations (Sha↵er, 2004). At a time when the WTO is in crisis over disagreements about its dispute settlement system, it is critical to understand how firms and states engage with the WTO, since “proposals for amending the WTO system are of little value if they are not grounded in a clear understanding of how the system now operates” (Sha↵er, 2003, 6). While no agreement or institution has done more to liberalize the rules of the trading system than the General Agreement on Tari↵s and Trade (GATT) and subsequently the WTO, states regularly impose barriers that are in conflict with their WTO obligations. In the presence of a multitude 2See Goldberg and Maggi (2001); Grossman and Helpman (1994). 3Kennard (2020, 199) shows that firm contributions for environmental standards can influence international cooperation, noting that her model is consistent with lobbying as information provision.

Given the consensus among economists that free trade is welfare enhancing, domestic interest groups are often blamed for the persistence of trade barriers. Yet even though "protection for sale" arguments have significant support, 2 domestic firms also play a prominent role in maintaining the liberal trading system, monitoring states' international trade policies, and increasing access to foreign markets. In contrast to a significant body of work that examines when and why trade barriers arise (Betz, 2017;Hosek and Peritz, 2022), this paper studies how firms and governments monitor trade barriers and select which barriers to contest. In doing so, this paper contributes to a growing body of scholarship that examines how private firms shape the development and enforcement of international law.
When it comes to understanding international law, the role of private firms is critical. Firms play an influential role from the creation of law itself to the monitoring and enforcement of international law. For example, Sell (2003, 8) emphasizes that state-centric "accounts of the Uruguay Round are at best incomplete, and at worst misleading" since they obscure the role of the private sector in establishing the agenda that led to the WTO agreements. Similarly, Perlman (2018) demonstrates that private firms can use their informational advantage to shape international standards. 3 Given that private firms are influential in the creation of international law (Ginsburg and Sha↵er, 2010), it should not be surprising that they also play an important role in monitoring and enforcing those laws.
For example, private actors can contribute to what Morse (2019) refers to as "market-enforcement," whereby private actors alter their behavior, e↵ectively punishing states that don't comply with international standards (Morse, 2022). When private firms engage in the monitoring and enforcement of international law, they also shape how the law is interpreted, which is especially influential at the World Trade Organization (WTO), where the need for consensus makes it challenging to alter the rules through negotiations (Sha↵er, 2004). At a time when the WTO is in crisis over disagreements about its dispute settlement system, it is critical to understand how firms and states engage with the WTO, since "proposals for amending the WTO system are of little value if they are not grounded in a clear understanding of how the system now operates" (Sha↵er, 2003, 6).
While no agreement or institution has done more to liberalize the rules of the trading system than the General Agreement on Tari↵s and Trade (GATT) and subsequently the WTO, states regularly impose barriers that are in conflict with their WTO obligations. In the presence of a multitude of potentially noncompliant trade barriers, states must decide how best to allocate their resources to monitor and enforce trade agreements. Building from theories of informational lobbying and bureaucratic subsidies, this paper analyzes the interaction between firms and their government and finds that a type of "litigation for sale" occurs. Unlike traditional models of lobbying, where interest groups make campaign contributions, this paper identifies an alternative form of lobbying through litigation contributions -contributions to the fact-finding e↵orts, research costs, and litigation tasks -which play important roles by signaling the strength and value of potential trade disputes and mitigating bureaucracies' resource constraints.
Although the WTO restricts dispute initiation to national governments, I show that private firms play a critical role in the dispute settlement process. The theory presented expands our understanding of firms' liberalizing influence (Kim 2017;Osgood 2017), and also contributes to burgeoning literatures on transnational versus interstate dispute settlement. I argue that the formal rules of the WTO allow its members to benefit from increased monitoring and enforcement provided by informal private firm participation, without governments taking on the additional risk associated with transnational dispute settlement (Allee and Peinhardt 2010). Unlike their role in transnational dispute settlement mechanisms, where firms' access to international arbitration is often viewed as eroding the sovereignty of the state (Brutger and Strezhnev, 2022), I show that the WTO rules allow governments to garner increased information and resources from firms, while preserving governments role as legal gatekeepers.
I show that private firms monitor WTO compliance and motivate states to seek enforcement of treaty obligations in two complementary ways. From a purely economic perspective, firms can contribute resources to support the litigation of WTO disputes, which reduces the cost of filing a complaint for the state and potentially increases the strength of the case. Firms are also positioned to signal information regarding the legal strength and value of potential cases, which allows the government to more accurately predict the probability of success. As the gatekeepers, governments select cases based on potential strength and value, which helps explain the nearly 90 percent success rate of WTO complainants (Davis 2012). I also examine firms' incentives to monitor and seek enforcement of international legal obligations when firms within an industry have divergent valuations for initiating a WTO complaint. The implications of the theory are consistent with qualitative evidence from 38 author-interviews and statistical evidence of dispute initiations. This paper makes a number of contributions, including demonstrating that private firms alter the WTO dispute escalation process in at least four important ways. First, private firms' influence leads to cases being brought to the WTO that diverge from governments' priorities -product specific barriers are more likely to escalate than trade barriers with broader e↵ects. Second, the number of cases initiated is higher with firms participating, as opposed to government-only models of dispute escalation, because firms mitigate the governments budget constraint, which helps both developing and developed countries challenge more trade barriers at the WTO. Third, the probability of the complainant winning increases, since firms signal information and the government screens cases based on the strength and value of the case. Finally, the quality and clarity of argumentation is improved with private firm participation. These mechanisms provide new insights into trade dispute escalation and whose voices are represented at the WTO.

Framing Dispute Settlement Participation
When a trade barrier is enacted, governments and firms have a multitude of potential responses.
Many barriers are left uncontested, but those that cause significant distortion are likely to catch the attention of firms and governments. In many countries, when firms realize they are facing a trade barrier they can report it to their government. In the US this information is compiled in the National Trade Estimate (NTE) annual reports and for the EU in the Market Access Database.
As an initial strategy, firms and governments will typically seek to have the trade barrier removed through bilateral negotiations with the country imposing the barrier. However, when the parties do not make progress through negotiations, they may escalate the dispute at the WTO, or through alternative pathways, as discussed in §15 of the appendix. Since only governments are allowed to file WTO complaints, they have the final say on whether to bring a case to the WTO, though firms have the ability to request government action through Section 301 petitions in the US or with the Trade Barrier Regulation (TBR) in the EU. Though there is notable variation across countries in how governments and firms engage with each other, there are also a common set of strategic incentives that are commonly considered when firms and governments evaluate whether to escalate a trade dispute.
Much of the existing discussion over dispute escalation at the WTO examines determinants of participation, which can be divided into research regarding which states choose to participate and which cases those states choose to bring to the WTO. It is generally agreed that countries engage in strategic decision making when considering whether to participate in WTO disputes (Betz and Kerner 2016;Johns and Pelc 2016), and that they choose to initiate disputes when their expected benefits outweigh the expected costs (Bown 2005).
Significant research has focused on the costs of initiating a dispute. According to one trade o cial interviewed for this project, the average cost of litigation in most WTO cases is around one million dollars per year for the duration of the dispute (Trade O cial 2014). In addition to the direct costs of disputes, Horn and Mavroidis (2011) note that indirect factors can play an important role, such as the threat of retaliation (Bown, 2005) and concerns about domestic political pressure (Betz and Kerner, 2016;Davis, 2012). Davis (2012, 2) argues that adjudication is a tool used to manage domestic political pressure and that domestic constraints make it more likely that executives will turn to the WTO to resolve disputes. Furthermore she finds that industries that exert significant pressure, measured by political contributions, are more likely to have barriers against them challenged at the WTO (Davis, 2012, 134). Davis rightly emphasizes the role of domestic influences on the dispute escalation process, but overlooks complementary mechanisms that firms use to seek enforcement of states' trade obligations, specifically litigation contributions and informational lobbying.
While much of the literature on WTO disputes focuses on the costs of disputes, it is also critical that governments have the necessary information to e↵ectively advance their claim. However, most governments do not have the resources, or access, to gather the facts for a case, which is why litigation contributions from firms are so important. When firms provide information and assist in preparing the arguments for the case, they help alleviate capacity constraints of governments and provide information about the trade interest at stake and strength of the case. Sha↵er (2006) argues that two important capacity constraints on WTO participation are a lack of legal expertise in WTO law and financial constraints to organizing e↵ective representation in the WTO legal system. Yet even among the WTO members least constrained by legal knowledge and resources, such as the US and the EU, 4 the private sector often plays a role in mitigating these constraints, while also signaling the strength of the potential case. For example, one expert involved in numerous disputes noted: "There are two main reasons the government can't manage the facts [of a case]. First, they just dont have the facts... typically the data of what type of violation has taken place is proprietary. You need to have access to proprietary data, so you rely on private businesses to bring the data forward... Second, the costs and resources to put the facts together and process the dispute... Take the EC, they cannot a↵ord to put someone on fact finding for a case full time. They don't have those positions and can't assign someone to do it, because theres no place in the budget for it" (International Trade Attorney 2021e).
Firms thus play a striking role in the dispute settlement process by mitigating resource constraints and providing information about the strength and value of potential complaints. In turn, this a↵ects the types of disputes and arguments brought to the WTO, which Sha↵er, Elsig, and Puig (2017, 292) argue a↵ects the development of international law and "shape[s] the system, both substantively and procedurally." An example of this type of public-private relationship occurred in a WTO dispute 291 over genetically modified foods between the EC and the US. Prior to the initiation of consultations, Monsanto, a producer of genetically modified foods, which had 15 products that had allegedly been adversely a↵ected by the European Community's actions (World Trade Organization 2012), engaged the US government in an e↵ort to ensure the case was brought. Although domestic pressure had been rising for years for the USTR to initiate a WTO dispute, the tipping point occurred when private firms signaled their beliefs about the case and contributed to the litigation process. According to a USTR o cial, the CEOs from the companies met with USTR o cials and agreed to support the litigation e↵ort (USTR O cial 2009). To convince the government to bring the case, the firms funded and completed a "laundry list" of fact-finding and litigation assignments (USTR O cial 2009). In response to the firms' contributions, the USTR moved forward with the case with greater confidence in the strength of the case and at a drastically reduced cost.

The Argument
Existing arguments regarding private firms' influence on dispute settlement participation are generally limited to firms' ability to define the trade agenda of states through traditional lobbying or government established mechanisms, such as Section 301 petitions in the United States (Bown and Hoekman 2005;Davis 2012). While firms also pursue alternative means of influence, such as bilateral consultations or domestic litigation, some firms contribute to the litigation process in an e↵ort to increase the likelihood a case is brought to the WTO. I argue that firms protect their interests through the dispute settlement process by contributing to the litigation costs of a WTO dispute, while governments use firm contributions to screen potential WTO complaints. Firms' contributions can take many forms, including conducting research, preparing legal briefs, and even litigating the case on behalf of the state. When a government is unwilling to pursue a case due to high litigation costs or its belief that the case is weak, firms can step in to fill the gap between expected costs and expected profits and to signal the strength and value of the case. Importantly, governments retain control over the gatekeeping process, so if the diplomatic externalities of the case are too high, the government may choose not to bring the case, which is a key distinction between the legal procedures of the WTO and transnational dispute settlement mechanisms, such as Investor-State Dispute Settlement.
As the following sections discuss, firms' contributions must do at least one of the following to alter the case selection process of states. Litigation contributions can lead the government to update its beliefs about the strength or value of the case. This can occur due to the information signaled or by strengthening the case by providing improved argumentation, additional evidence, and expanding the total litigation budget. The firm's contributions may also lower the costs to the government, acting as a bureaucratic subsidy and mitigating the government's resource constraint. If litigation contributions either reduce the cost to the state or alter the state's beliefs about the strength or value, then firms can play a significant role in monitoring and seeking enforcement of international trade law at the WTO.
While firms have an incentive to signal the strength and value of cases to their government, the government's and firms' preferences are not necessarily aligned. Firms tend to have a relatively narrow focus on increasing market access by removing the trade barrier in question, which often comes with a desire to use an aggressive legal strategy (Sha↵er, Elsig, and Puig, 2017, 295). 5 For example, one expert noted that "Some firms push the envelope and try to bring more legal claims... Many governments are careful to avoid this, such as the USTR. They dont want to bring superfluous legal arguments." (International Trade Attorney, 2021b). By contrast, Bello (1996, 358) notes that governments tend to be "institutionally risk-averse." For example, in the case of the US, Sha↵er (2003,60) notes that there is a fundamental tension between firms and the government, which is caused by the fact that "the USTR represents the national interest, not the firm's interest. In particular, the USTR must consider that the United States may subsequently be on the defensive in a similar case." Similarly, the European Commission emphasizes their goal is to serve the "community interest" or "public interest" (Sha↵er, 2003, 108), which manifests in a preference for challenging systemic trade barriers, as opposed to narrow barriers that may only a↵ect a single product or firm.

Firms' Informational Advantage
I argue that firms have an informational advantage throughout the litigation process, given their position in perceiving and analyzing trade barriers. The unique position of firms can best be illustrated by considering their role in three phases of litigation known as "naming, blaming and claiming" (Felstiner, Abel, and Sarat 1981). The naming phase involves identifying an injury to one's trading prospects (Sha↵er 2006). The private industry has the greatest incentive and ability to identify an economic injury. In author-interviews, government o cials repeatedly emphasized "we don't find out about trade problems until the industry tells us, and we have to rely on market intelligence to tell us about the impact in the market and what they think is the problem" (Counsel for WTO Disputes 2021).
The "blaming" phase of a dispute determines who is responsible for the injury identified (Sha↵er 2006). Once the injury is perceived, blaming can be relatively straightforward. If the lost profits are due to a trade disruption with a specific trading partner or to a flood of imports from a specific country, minimal costs should be associated with identifying who is to blame.
The most complex phase of dispute settlement is "claiming," which consists of pursuing a claim through the WTO (Sha↵er 2006), although much of the e↵ort of claiming is done before the case is formally initiated. The information required for WTO disputes can be vast, and generally relies on the private information of firms a↵ected by the trade barrier. A USTR o cial interviewed for this project estimates that half to three-quarters of the litigation expenses are devoted to the fact finding portion of claiming (USTR O cial 2009). During this phase, firms quantify the value of lost revenue from trade, build the case connecting their losses to the barrier, and work with the government to formalize the complaint through the dispute settlement process.
As Sha↵er (2003, 35) notes, private firms are often the "eyes" for government and the importance of firms' information provision is increasing as WTO cases become more fact intensive. This leads to the expectation that firms should play an active role in providing information about the strength and value of potential cases in an e↵ort to convince the government to challenge trade barriers at the WTO. 6 My argument deliberately focuses on the strategic interaction between the firm(s) advocating to bring a case to the WTO and the government. A potential complication of the role of firms in the dispute escalation process would be the involvement of firms lobbying against dispute escalation.
However, counter-lobbying is incredibly rare for a number of reasons, which I discuss in detail in §10 of the appendix. For example, one of the challenges is that firms that could be negatively a↵ected by a dispute are often unaware that a dispute is escalating until after it's initiated. Trade lawyers emphasized this point, noting that they don't see counter-lobbying because the "process is so confidential that other firms may not know much prior to a request for consultations" (International Trade Lawyer, 2021b). The rarity of counter-lobbying was confirmed during the 38 expert interviews, with only one confirmed case of counter-lobbying identified, which is discussed in the appendix, §10.

Mechanisms of Influence
From the perspective of the government, private contributions are important for relaxing the government's budget constraint, since the contributions act as a bureaucratic subsidy. The budget constraint varies across countries, but even among the wealthiest members of the WTO, there are significant capacity constraints. For example, the USTR is responsible for initiating WTO complaints, but their total budget is only about $47.5 million annually (Cook 2013). Within their budget, the executive's top priorities are negotiating trade agreements -not litigating existing agreements (USTR 2014). This creates a situation where, as the USTR's top litigator noted, budget concerns limit the ability to initiate legal complaints and seek enforcement of trade agreements (World Trade Online 2013). One attorney involved in numerous WTO disputes noted that there have been situations where governments were willing to file WTO disputes, but without litigation contributions from the firms, the government lacked the resources to move forward with the complaint (Associate Trade Attorney 2009). 7 The importance of resource constraints was emphasized by a multitude of o cials from a variety of countries, as the interview quotes in Figure 1 show, which is also discussed further in §11 and §13 of the appendix.
Firm contributions also play an informational role as a signal of the strength and value of the case, which is a key factor in determining whether the government challenges potential WTO violations. The importance of changing beliefs about the legal strength of cases was emphasized by trade attorneys, who noted that there have been numerous cases where the government did 7 This situation illustrates that information provision and resource contributions are not necessarily substitutes. Even with su cient information to believe a case is strong, some governments still require resource contributions to bring the case. Conversely, a government may have financial resources for a case, but not have su cient information to believe they have a strong case.

Figure 1: Evidence of Resource Constraints Across Countries
The USTR is the most resourced, and they can't staff cases, they don't have the resources. There is so much going on that they can't think about starting cases on their own... The US don't admit they don't have the resources, but they don't (International Trade Attorney, European Union, 2021).
There's a significant budget and resource constraint on governments. One of the important roles the trade agencies play is as a filter... Ironically, the USTR's budget is tiny relative to other countries...USTR views themselves as the marines, "the few the proud" (International Trade Attorney, China, 2021).
The ministry is always struggling how to allocate within the budget. I think some ministries try to get budget from industry, and request from industry to pay the fees. Basically, the budget of the ministry is very limited (Legal Advisor to Ministry of Finance, Japan, 2021).
Sometimes it's a resource constraint. Governments have to be putting out lots of fires, so it's hard to dedicate resources on a full time or focused basis to prepare for written submissions and complete that type of analysis (International Trade Lawyer, Colombia, 2021. The cost of cases has been increasing year over year. In the early years of the WTO the reports were relatively short, but recently the cases are 100s of pages long. So I think the cost has been increasing. Government budget has increased, but industry has had to play a larger role (Ministry of Economy, Trade, and Industry Official, Japan, 2021).
There are times when the government says I can't do it myself because they don't have funds or don't have legal capacity… We can scale countries roughly based on GDP with larger countries being more sophisticated, and they will have more/stronger views about what is important and what is systemic. With smaller countries, the government tells industry it's fine to bring the case as long as they pay for it, and government just signs their name to it (International Trade Lawyer, United States, 2021).
With regard to budget constraints and legal knowledge, if we go back to Bananas and think of Ecuador. They weren't a rich country... Ecuador was not particularly well resourced... and they must have had assistance to mount that challenge... I think the gambling case in Antigua would be another such case (International Trade Lawyer, Hong Kong, 2021) The budget constraint is very real. … The steel industry for example. Government tells them to just pay for the case and lawyers (General Counsel, Ministry of Mexico, 2021).
not believe there was a viable case, and only through private firms preparation of arguments and pitch to the government was the government convinced to bring the case (Associate Trade Attorney 2009). For example, in dispute DS524 concerning the importation of fresh avocados, the Mexican government did not initially believe there was a strong legal claim to bring the case and so they didn't want to initiate a dispute, but the firm gathered information and convinced the government there was a strong case, which was then brought to the WTO (General Counsel 2021).
Beliefs about the strength of the case are particularly important given governments' concern about loosing WTO disputes. Two factors contribute to governments' heightened concerns, compared to firms. First, government o cials responsible for selecting cases must choose from a broad set of potential cases and only initiate a select few. One o cial noted that in many countries "The people may lose their job if they lose, so the chance of success is very important" (International Trade Attorney 2021e). A European Commission o cial emphasized that the "strength of the legal issue" is of primary importance (European Commission O cial, 2009), while a USTR o cial noted that they seek "slam dunk" cases (USTR O cial, 2009). While firms also face resource constraints, each firm has a smaller set of potential disputes to choose from, and pursuing the firms' strongest case may still be somewhat of a gamble, whereas government o cials have the opportunity to select a pool of strong cases, and are best o↵ choosing only the strongest. Additionally, when a government pursues and loses a WTO complaint, they not only face the losses from the dispute in question, but also a precedent where the issue in question is given a green light by the WTO. 8 For example, if the US were to file and lose a complaint against China regarding currency manipulation, not only would China be able to continue their policies, but other countries would then be able to adopt similar policies without fear of legal challenges (Davis 2012, 165-168). Due to these risks, governments place significant weight on the strength of cases when evaluating whether to challenge potential WTO violations, and much of the information about the legal quality of the case comes from private firms.
Hypothesis-1: If firms have an information advantage compared to governments, then firms will provide trade barrier information to governments, increasing the probability a trade barrier will be challenged in a WTO dispute.
By theorizing the strategic incentives of the government and firms, we can generate additional empirical implications, which are formalized in §1 of the appendix. For example, we know that a case will not be initiated if the litigation cost for a case is greater than the combined expected payo↵ to the government and firm, which is formally proven in the appendix, §2. 9 Such cases, by definition, are not profitable to pursue and so neither the firm nor the government would contribute to them. A further general result of the theory is that whenever the total litigation cost is less than the expected profit to the government, the case will be initiated. This means that the litigation cost of the case is low enough relative to the expected payo↵ that it is beneficial for the government to unilaterally initiate the case. Although rare, these types of cases would likely be brought when the precedent value of a case is high, which occurred in some of the early intellectual property rights The most interesting results of the theory are from the set of cases where the government would be unwilling to initiate the case without a litigation contribution from the firm. The first set of such 8 Divergent firm and government preferences are discussed further in appendix, §16. 9 The expected payo↵s are based on the probability of winning the case times the value of winning the case and are formally defined in §1 of the appendix.
10 It is widely accepted that the de f acto importance of precedent can be quite high in WTO disputes (Pelc 2014), and that case law matters at the WTO (Kucik, 2019).
cases are those where the expected profit to the government is less than the total litigation cost.
In a unitary actor model, these cases would be viewed as unprofitable, however a firm's litigation contribution can alter the expected payo↵s to the state by mitigating the resource constraint, making such cases profitable to the government. The logic leading to this implication is formalized in the appendix, §2.
Hypothesis-2: Ceteris paribus, firms litigation contributions mitigate the resource constraint, increasing the probability a trade barrier will be challenged in a WTO dispute.
A second, and potentially overlapping, group of cases are those where the government's prior belief regarding the strength of a case is su ciently low that the government does not believe case initiation is profitable. In this group of cases, if the firm knows that the case is strong, it can credibly signal the strength of case to the government, thus altering the expected payo↵s of the government and motivating the government to initiate the case. When a firm contributes more than it would expect to gain from a weak case, it signals that the firm believes the case is su ciently strong and valuable -otherwise the contribution would have negative expected utility. This is formally proven in the appendix, §2.
For simplicity, I refer to contributing more than the firm would expect to gain from a weak case, as the firm meeting the "contribution threshold," since this threshold provides a credible signal about the firms beliefs about the case. The existence of a contribution threshold helps explain the extremely high success rate of WTO complainants, given that governments are able to screen out cases that are not strong enough when working with private firms during the litigation process. 11 Although it is theoretically parsimonious to think of the existence of an easily observable threshold, in practice this threshold may be challenging to observe, in which case firms may have to go above and beyond to convince the government of the strength and value of their case. Sha↵er (2003, 47) also recognizes that such a contribution threshold exists, noting that governments often require "industry to submit convincing factual and legal memoranda as a prerequisite to its filing of a WTO complaint," which is consistent with the theory's implications under a broad set of beliefs, as discussed in §3 of the appendix.
Hypothesis-3: Ceteris paribus, when firms meet the contribution threshold, the government will update its beliefs, increasing the probability a trade barrier will be challenged in a WTO dispute.
A further implication of the theory is that a case will be more likely to be initiated when the trade distortion caused by a particular trade barrier is greater. A higher level of distortion means that a country will be forgoing relatively more trade, which increases the value of the case. Distortion also acts as a proxy for legal strength, given that proving economic harm can be an important facet of achieving compensation and securing a legal victory, and is indeed required for Article XXIII nullification or impairment complaints. 12 Distortion impacts the expected value and strength of the case, which means trade barriers with high levels of distortion should be contested in the WTO with a higher probability than similar barriers with lower levels of distortion.
Hypothesis-4: Ceteris paribus, trade barriers that cause high levels of distortion have a higher probability of being challenged in a WTO dispute.

Industry and Firm-Level Implications
Analyzing the interaction between a firm and the government provides a useful starting point for understanding WTO case initiation, but I now consider the incentives for an industry with multiple firms. I begin by considering the incentives of firms to contribute to the litigation process when multiple firms within an industry may be a↵ected by a trade barrier and have heterogeneous preferences with regard to the potential dispute.
While firms still have better knowledge about the strength of a case, I now examine how uncertainty over the heterogeneous valuations of the firms a↵ect the likelihood they contribute to the litigation process. 13 If we assume that firms within an industry can coordinate their litigation contributions, then this interaction perfectly resembles a contribution game where private actors with incomplete information engage in a game to provide a discrete public good -in this case the public 12 It has also been noted that high levels of distortion increase the likelihood of a violation ruling (Davis, 2012, 129). 13 Since the expected payo↵s to firms are a function of the strength and the valuation, all else equal, firms are still more likely to contribute when the case is strong. good is the initiation of the case, where the benefit from the case accrues to the firms within a given industry.
In such a contribution game, not all firms within an industry will benefit equally from a trade dispute, which is why firm-level valuations can be heterogeneous. A more complete discussion of such a game, which has been analyzed in di↵erent contexts by , is provided in the appendix, §4. In the most simplistic version of the game, I consider firms strategies when the cost of contributing to the good is low enough such that a single firm can initiate the case.
In this situation, a symmetric equilibrium always exists where a single firm will contribute enough to reach the contribution threshold and the good is provided (Menezes, Monteiro, and Temimi, 2001, 499), which means the government initiates the case.
The first implication to emerge from the game with incomplete information and heterogeneous firms and contributions is that industries with dominant firms will be more likely to initiate cases, since it is more likely that a dominant firm will be able to a↵ord to pay the contribution threshold.
This finding hinges on the fact that for an industry where a single firm has a relatively high expected payo↵ from a WTO case, there is a strictly greater probability of contributing to the litigation cost of a dispute than an industry where no single firm has an incentive to pay the contribution threshold, in which case the probability that a case is initiated is strictly less than one (unless the case is initiated unilaterally by the Government). Dominant firms will also be most likely to have the capacity to pay the contribution threshold.
Hypothesis-5a: Ceteris paribus, in industries where dominant firms have relatively high value and capacity to pay the litigation contribution threshold, it is more likely that trade barriers will be challenged through a WTO dispute.
Next, I consider the contribution game when no single firm can a↵ord to pay the contribution threshold, and find that a coordination problem exists that eventually becomes great enough that a symmetric equilibrium resulting in case initiation is no longer possible. For a wide range of costs of a public good, the coordination problem prohibits provision of the good (Menezes, Monteiro, and Temimi 2001, 496). Of particular importance is the finding that if the cost of the public good is slightly above the aggregate mean of the valuations then the unique equilibrium of the contribution game is for each player to contribute zero no matter what its value is (Menezes, Monteiro, and Temimi 2001, 502). This implies that even when an industry as a whole may stand to benefit from the initiation of a WTO dispute, if no single firm can a↵ord to pay the necessary litigation cost to motivate the government to file and the average valuation by all firms within the industry is low enough, the case will not be initiated. From this, a second implication emerges -as the mean value and capacity for the industry increases, case initiation becomes more likely, since there is a greater chance that the mean value and capacity for the industry will exceed the cost of litigation, making it more likely firms will contribute to the litigation process. 14 Hypothesis-5b: Ceteris paribus, as the mean value and capacity for an industry increases, it becomes more likely that trade barriers will be challenged through a WTO dispute.
The previous two hypotheses are derived from predictions regarding how firms within an industry overcome collective action problems when facing a trade barrier; however, other factors can also mitigate or remove collective action problems. Most importantly for an analysis of trade disputes is the specificity of the trade barrier in question -how many products within an industry are a↵ected by the trade barrier -determines the extent of the coordination problem firms face. For example, a barrier that distorts trade for all firms within an industry will create a significant collective action problem, especially if the stakeholders are smaller (Sha↵er, Elsig, and Puig, 2017, 294), whereas a barrier that only a↵ects a specific product will have a more concentrated impact, thus reducing or eliminating the collective action problem. An expert interviewed for this project confirmed "The collective action problem is an important one. We see that right now in Europe with respect to half a dozen di↵erent sectors" where they are unable to come together to challenge trade barriers (Counsel for WTO Disputes 2021). In some cases, when a trade barrier has a large e↵ect on a particular industry, the collective action problem may be overcome with the help of an association. For example, firms coordinated their e↵orts through the Coalition against Australian Leather Subsidies, pressuring the USTR to file a WTO complaint (Sha↵er, 2003, 33). While industry associations can alleviate collection action problems, they are most likely to do so when a trade barrier has a specific-targeted e↵ect on the industry, as opposed to a more di↵use trade barrier. However, when there is a product specific barrier "Normally there is one company that cares a lot and takes the lead" (International Trade Lawyer 2021c). In fact, numerous o cials emphasized that for many industries firms and association don't cooperate to initiate disputes. For example, one expert noted that "Firms work independently. They do not cooperate when asking for requests for consultations.
Sharing information may result in conflicts of interest so they don't work together" (METI O cial 2021). This suggests that product-specific trade barriers should be more likely to be challenged, since they are least likely to generate collective action problems, as discussed further in §18 of the appendix.
Hypothesis-6: Ceteris paribus, product-specific trade barriers should have a higher probability of being challenged at the WTO than more di↵use trade barriers.
Hypothesis-6 also provides a useful comparison against alternative theories of dispute initiation.
If governments independently evaluate whether to initiate a dispute at the WTO, then collective action problems at the industry level should not influence case selection. In fact, trade barriers that harm entire industries or multiple industries would be more likely to be challenged, since the government could help more firms with a single case. One o cial noted that the government prefers to pursue issues with horizontal e↵ects, "the motivation is to go after structural and systemic issues. Typically these would be issues that a↵ect multiple industries" (International Trade Attorney 2021a). Furthermore, an o cial familiar with USTR priorities noted that "An individual industry is almost always only concerned with the very narrow particular dispute or industry... The government wants to invest their resources in cases with broader impact" (International Trade Lawyer 2021c).
Thus, if product-specific cases are more likely to be initiated, then the government is bringing cases that impact fewer firms, which is consistent with firms influencing the types of cases initiated and having to overcome collective action problems to do so, as opposed to the government initiating their preferred cases that a↵ect broad issue areas.

Qualitative Evaluation of the Mechanisms and Theory
To examine whether the hypotheses and mechanisms put forth are consistent with dispute escalation patterns at the WTO, I conducted 38 in-depth interviews with trade experts from around the world. The interviews are especially helpful when evaluating hypotheses 1 through 3, which focus on the private actions of firms and governments. The selection of interviewees was guided by a number of goals. First, interviewees must have had significant experience with international trade policy and disputes. Second, to capture variation in WTO participation across contexts, I sought interviewees from a range of countries from all levels of development and frequency of WTO participation. Third, the interviewees were selected to ensure that the perspectives of government, private industry firms, and law firms were represented. These goals led me to contact potential interviewees who worked for firms that were a↵ected by trade barriers, government o cials involved in trade policy and disputes, and lawyers who practice international trade law. The response rate was near 50 percent, and the resulting sample included individuals representing countries across Africa, Asia, Australia, Europe, South America, and North America. 15  Interviewees agreed to be interviewed anonymously, given that many are still involved in trade disputes. Participants agreed to be cited by either their previous or current professional position.
The sample includes those who served as members of the WTO appellate body, United States Trade Representative, USTR General Counsel, ambassadors, in-house counsel for private firms, etc.
Although many of the o cials held very high-ranking positions, most opted to be cited by more generic titles, such as "International Trade Lawyer" to preserve their anonymity. A complete list of 15 More on the interview process is included in §20 of the appendix.
interviewees is presented in Figure 2, which displays the title -as the subject chose to be citedand primary country associated with each interviewee's WTO experience, though most interviewees have experience working with and representing firms or governments from multiple countries.
I begin by considering Hypothesis-1, which argues that, if firms have an informational advantage, we should observe firms providing information to increase the likelihood a case is brought. An empirical challenge of examining litigation contributions, whether they be informational or financial, is that they are private activities that are not publicly known across a broad range of disputes.
However, the interviews consistently showed that private firms play a critical informational role in the dispute escalation process. The qualitative evidence in Figure 3 further highlights the informational advantage of firms and the reliance of governments on firms' information, as predicted by Hypothesis-1.
The information asymmetry was also emphasized by interviewees, who noted governments don't have direct access to the market data needed to evaluate the case. A USTR o cial summarized this point saying "we need the firm to bring data to show the problem really exists, the magnitude of the problem... what we need is confidential and proprietary information" (Assistant for WTO and Multicultural A↵airs 2021). The point was echoed by others in Figure 3, with numerous government o cials noting that firms have better knowledge of the market. One o cial noted that the government has to "filter cases" since they can't bring them all, so the firm must "say what's the argument, what's the damages, and what's the prospect of winning the case" (International Trade Attorney 2021d).
Firm contributions also alter the dispute settlement process by mitigating the resource constraint, as predicted in Hypothesis-2. As noted in Figure 1, the interviewees emphasized the impor-

Figure 3: Evidence of Private Firms Identifying Trade Barriers
Market operators are always a reality check because they face the day to day business... Governments generally don't systematically monitor what other governments are doing. Maybe USTR and maybe the EC does so to some extent, but that radar screen still has problems... Sometimes the actions aren't detectable, except by those actors directly facing the measure (International Trade Attorney, Egypt, 2021).
We need to hear from industry to know there's a problem. We have the National Trade Estimates Report, which is a mix of things we've heard from industry and also things we've been monitoring... USTR will sometimes have companies come to them, and we need the firm to bring data to show the problem really exists, the magnitude of the problem (Assistant for WTO and Multicultural Affairs, United States Trade Representative, 2021).
The government would rely on the industry or commercial entity to complain to it, so I think the private sector involvement is absolutely basic to the whole system (International Trade Attorney, Hong Kong, 2021).
Private companies are involved because they know the market. The government doesn't know what happens. For TRIMS and TRIPS measures, the private firms are always involved and generally pay part or all of the lawyers fees (International Trade Attorney, Belgium, 2021).
If you're a poor and understaffed country, you don't even know if you're facing barriers hurting your firms (US International Trade Commission Attorney, 2021).
What happens in the majority of cases, maybe not all but certainly in the great majority, the commercial entity feels it's not getting a fair deal and presents its complaints to its own government (International Trade Attorney, Egypt, 2021).
Outside of the US and Europe everyone relies on the private sector to bring information about the case… But generally, even in Europe, the law firm is supposed to bring the facts to the European Commission (International Trade Attorney, Russia, 2021) It's not uncommon for a company, especially very large companies, to approach the law firm and say "were having this issue in this market, can we do something about it" and then approach USTR to address the trade barrier (International Trade Attorney, Brazil, 2021).
tance of resource constraints and the critical role of private firm contributions in mitigating those constraints. One o cial familiar with the USTR's cases noted, "The AB has encouraged everybody to drill down and write 400-500 pages, and its very possible that USTR is literally swamped. They

Figure 4: Evidence of Litigation Contributions Mitigating the Resource Constraint
The first thing is they [the government] will ask the firm to come up with evidence, the facts, and all those sorts of things. And that is a very normal thing for governments to do even before they decide whether to file a request for consultations (International Trade Lawyer, Switzerland, 2021).
First question is always "who is going to pay for this litigation?" In every case I know of, Industry pays (Senior Official familiar with WTO and Airbus-Boeing Dispute, 2021).
The government always tells the industry they have to take care of experts. The industry has more flexibility for arranging contracts, so for government it just takes too long with their procurement processes. The government can't get approval to hire the experts (International Trade Lawyer, United States, 2021).
The government often says yes this is fine, but I don't have the money to pursue it, so they need industry to pay. The government may also say they don't care about it commercially, so the government doesn't want to put money into it, and thus industry has to pay... There are times when the government says I can't do it myself because they don't have funds or don't have legal capacity (International Trade Lawyer, United States, 2021).
Usually we [the government] pay 30% and the rest, 70% is usually paid by industry. This 30/70 is a basic formula, but it depends on the case (Ministry of Economy, Trade, and Industry Official, Japan, 2021).
Private lawyers were paid by private companies, and in the Banana case everything was privately funded by Naboa… The complaint from developing countries, they always say we don't have the money.
In the [redacted] case, there was one major company... It was expected that they hired expert counsel throughout the case, and came to a special arrangement, and essentially got them to pay for the lawyers (Counsel for WTO Disputes, Canada, 2021 The ministry is always struggling how to allocate within the budget. I think some ministries try to get budget from industry, and request from industry to pay the fees. Basically, the budget of the ministry is very limited (Legal Advisor to Ministry of Finance, Japan, 2021).
Then they [the private firm] take it to government and say this is an intergovernmental treaty and were the beneficiaries. Since the government holds the legal right, we ask them to bring the case. In many instances the company would say, you litigate this on our behalf because we don't have legal standing, but we will hire the law firm and pay the fees (International Trade Lawyer, Egypt, 2021).
Certainly when we need more, we're not shy about asking for the info, which is mostly technical information and market information. We can get high level information from industry associations, but we really need to talk with individual companies because what we need is confidential and proprietary information (Assistant for WTO and Multicultural Affairs, United States Trade Representative, 2021).
There is also evidence of change over time across countries. Some countries, such as China, have proactively sought to overcome their capacity constraints and empower firms to engage in WTO support. 16 For example, firms are playing a larger role in alleviating the governments resource constraints in Brazil (Sha↵er, Sanchez, and Rosenberg, 2008) and the EU. Sha↵er (2003) found that in the early years of the WTO, the US had closer ties to private firms, but overtime other countries have been catching up. An o cial with the European Commission emphasized that in the GATT and early years of the WTO, European firms believed it was the government's responsibility to manage trade disputes (European Commission O cial, 2009). However, over time expectations shifted, with the same o cial noting that European firms learned to play a more active role in the fact finding and legal preparation, since the government was unable to handle the increasingly fact-intensive and complex cases. In aggregate, the role of firms in mitigating the resource constraints across countries strongly supports Hypothesis-2.
If the theory is correct, we should also find evidence that firms' information provision and signaling leads the government to update its beliefs about the strength or value of the case, as 16 For more on how the litigation process varies across contexts, see §17 of the appendix. predicted in Hypothesis-3. For many governments with fewer resources, they are simply ill-equipped to evaluate the strength of the case on their own. As noted above, Mexico did not believe it had a strong case to challenge the rules a↵ecting fresh avocados, but after firms provided information the government was convinced to initiate the dispute (DS524). Even in better resourced countries, such as Japan, government o cials expect the firm to show the case is strong; "METI asks a lot of industry. There is a burden of proof and industry has to prove it's a serious issue, the damage is quite high, and if we request a consultation, we are probably going to win. They have to convince METI or else they wont move" (METI O cial 2021). Similarly, a Mexican o cial a rmed that it's up to the firm to make the government "believe the industry really has a good case. We need to have some certainty of winning the case" (General Counsel 2021). In order to convince the government the case is strong, firms generally engage in extensive fact finding, drafting of legal arguments, and information provision to the government, as is shown in Figure 5. It's very difficult for a government lawyer to become educated and it takes a long learning curve and it would be a waste of resources to have the government lawyers dealing with it. Government lawyers may be better on the legal theories and institutions, but not the facts of the case... When it comes to market data, then the private companies and their associations are the ones who the government has to rely on. The EC was calling more than once to the private firms, to get information about market share and consumption information... Anything that has to do with the market and/or micro indicators, the private sector is better.
Industry helps with the fact finding and resources for the case. Sometimes, the governments only job is to be present at the meetings, and the attorneys paid for by the industry do all the speaking (Ambassador, Brazil, 2021).
When fact finding is needed and experts are needed, then it is much more complicated. Industry will be involved in assisting in picking the consultants, working with the consultants, etcetera, who then submit the expert reports (International Trade Attorney, China, 2021).
One of Private industry's main contributions is financial resources and product specific knowledge. Say you have a relatively simple case on national treatment, you still need lots of specific knowledge and private industry is best placed to have that info, and can be very helpful in developing the factual record (International Trade Attorney, Brazil, 2021).
For USTR, more often providing technical information is the most important… USTR can often handle the legal case, but they rely on the technical information about how the market works, and support and partnership [from private firms] in developing arguments (Assistant General Counsel, United States Trade Representative, 2021).
Governments tend to look for expertise from private firms. We interact very early with our clients, and government may request early memos on market access issues. They might ask for help collecting information about the measure itself, the legislation or regulation. It may involve working with local counsel to understand the domestic regime (International Trade Lawyer, Colombia, 2021).
Certainly when we need more, we're not shy about asking for the info, which is mostly technical information and market information. We can get high level information from industry associations, but we really need to talk with individual companies because what we need is confidential and proprietary information (Assistant for WTO and Multicultural Affairs, United States Trade Representative, 2021).
Although the consensus amongst those interviewed is that private firms play a critical role providing information to governments, it was also noted that di↵erent cases and countries yield di↵erent styles of government-firm interactions. This is consistent with Sandholtz and Whytock (2017), who argue that di↵erent governance systems will yield di↵erent interactions between the law and politics.
For example, the United States and European Commission are sometimes better positioned than other countries to identify trade barriers, with one expert noting that "Generally outside of the US and Europe everyone relies on the private sector to bring information about the case" (International Trade Attorney 2021e). However, even for the US and EC it was emphasized that "the radar screen still has problems... Sometimes the actions aren't detectable, except by the actors directly facing the problem" (International Trade Attorney 2021c). For cases addressing systemic issues, the fact finding is often at a higher level and less intensive, which somewhat reduces the information asymmetry; "We can get high level information from industry associations, but we really need to talk with individual companies because what we need is confidential and proprietary information" (Assistant for WTO and Multicultural A↵airs 2021). Another o cial confirmed, "When it comes to market data, then the private companies and their associations are the ones who the government has to rely on... Anything that has to do with the market and/or micro indicators, the private sector is better" (International Trade Lawyer 2021a). By contrast, when a trade barrier harms a State Owned Enterprise, which has occurred most prominently in China and Russia, there is less of an information asymmetry (International Trade Attorney 2021e).
The interviews illuminate the mechanisms of WTO case selection, especially those components not readily measurable across a broad set of cases and is consistent with hypotheses 1, 2, and 3.
. Governments don't have the capacity to comprehensively monitor trade barriers, and thus firms are better positioned to identify trade barriers and know how significant they are. Second, due to resource limitations and regular sta↵ turnover in many countries, government o cials don't have the expertise or time to gather and process the necessary information for WTO cases (International Trade Attorney 2021a, International Trade Lawyer 2021e). 17 Finally, much of the information needed to build a WTO case involves proprietary firm-level data and market data, and thus the government is reliant on firms to provide this information, which is essential for assessing the strength and value of cases. Taken together, the interviews point to prominent informational and resource roles for private firms in the dispute escalation process.
17 See appendix, §12 for more on the constraints caused by sta↵ turnover.

Dispute Escalation Analysis
To further test the implications of the theory, I use firm-level data gathered from Compustat in conjunction with the Foreign Trade Barrier Dataset (FTBD), which allows me to test the e↵ect of trade barrier-specificity, firms' litigation capacity, the level of trade barrier distortion, and competing theories on the probability of dispute initiation from a set of potential WTO cases. The FTBD is comprised of a set of potential disputes, which are defined as harmful trade barriers to US exports identified in the National Trade Estimate (NTE) annual reports (Davis 2012 that the FTBD provides a much more comprehensive set of potential disputes than previous studies, which can be used to analyze dispute escalation patterns. The data allow me to test hypotheses 4 through 6 within a subset of potential trade barriers that have met a minimum threshold to be recognized by the government. Although the selection process may result in some barriers being omitted from the dataset if they have not been raised in the NTE, any such omission would bias against my findings, since the most likely cases to be omitted would be those with low levels of distortion and a low chance of escalation, as discussed in appendix, §14. Furthermore, the data are restricted to trade barriers against the US, which has the advantage of holding the initiating country constant, which controls for a multitude of potential covariates. Focusing on US dispute initiation during the first ten years of the WTO has advantages and limitations. On one hand the US is one of the most well-resourced countries, with an experienced set of personnel at the USTR, which would make the US less likely to be reliant on private firms than other countries. Furthermore, the complexity of cases at the WTO -and therefore governments reliance on private firms -has increased over time, making the first ten years of the WTO a relatively hard test of the theory. 18 On the other hand, the US has a history of firms having direct contacts with government o cials (Sha↵er, 2003), making the US a more likely case to observe firms influencing dispute escalation. Although focusing on the US has some limitations, using the FTBD 18 This poins is developed in appendic, §13.

22
complements the cross-national interviews, by providing a rigorous empirical analysis for the most frequent user of the WTO's dispute settlement system.
The unit of analysis is the trade barrier, with an observation included for every year the NTE mentions the barrier in their report. 19 Focusing on the trade barrier allows me to directly test Hypothesis-4, testing the e↵ect of distortion caused by a trade barrier on the probability that the trade barrier is challenged in the WTO. While each barrier in the dataset is assumed to cause some level of distortion, the hypothesis focuses on the relative di↵erence between low and high distortion barriers. The Distortion variable for each trade barrier is coded as an indicator variable that identifies cases with significant market closure that are highly distorting. Significant market closure is defined as resulting from a ban, quota, or increase of tari↵/duty of more than 10 percent, standards or rules of origin that create a de facto ban on imports, violation of intellectual property rights, or subsidies to competitors (Davis 2012). The expectation for distortion is positive, as the variable directly increases the payo↵ from the case and the expected legal strength.
To test Hypotheses 5a and 5b, which focus on the connection between firms' capacity to contribute to the litigation process and dispute initiation, I compiled firm-level data using the Compustat database. I test Hypothesis-5a, using Dominant Firm Capacity, measured as the log of the earnings in a given year for the top earning firm in an industry. 20 This measure acts as a proxy for the firm's ability to pay the contribution threshold necessary to signal information and the firm's ability to mitigate the bureaucracy's resource constraint. I also test Hypothesis 5b using the Average Firm Capacity for each industry, which measures the average earnings of firms for each industry in a given year.
Next I examine Hypothesis-6, which says that product-specific trade barriers should have a higher probability of being challenged than more di↵use trade barriers. The FTBD codes the specificity of each trade barrier by identifying the industry and product a↵ected by the particular barrier.
The industry a↵ected by the trade barrier is coded at the level of the ISIC3 4 digit classification. 21 Of the 1635 trade-barrier-years analyzed in the data, 23 percent are product-specific. Productspecific barriers are coded as those where the policy a↵ects a single product within the industry.
An example of a product specific barrier was Canada's import restrictions placed on periodicals.
Canada implemented Tari↵ Code 9958, which prohibited imports of "special edition" periodicals (World Trade Organization 2010). Such a specific barrier did not impact the media industry as a whole, or even the entire print-media, and thus its specificity reduced the collective action challenge faced by the a↵ected firms. In response to the trade barrier, the United States escalated the dispute in 1997, which became DS31.
To account for other trade barrier-specific factors, I include a range of controls. First, I examine whether progress has been made in negotiating the removal of the trade barrier. Progress is coded on a four point scale indicating the level of progress toward resolving the disputed trade barrier (Davis 2012 the key variables of interest to be analyzed across the dataset. 23 The results are also robust to alternative fixed e↵ects and ordinary least squares models, which are shown and discussed in the appendix, §6 and 7. The results of the baseline model are reported in Table 1, Model 1. Hypothesis-4 receives strong support, shown by the positive relationship between the trade barrier's level of distortion and the likelihood a dispute is initiated. There is also support for Hypothesis-5a, which states that industries with a dominant firm with high capacity will be more likely to have their cases brought to 22 Using the Hausman test, I compared the random e↵ects model to a fixed e↵ects model , with both at the ISIC3 4-digit level, and found that the null hypothesis -that the random e↵ects model is consistent -cannot be rejected (prob> 2 = 0.29). 23 The results are robust to grouping on trade barrier as Davis (2012)  Perhaps most interesting, Hypothesis-6 receives strong support, with the results showing that product-specific barriers are much more likely to be challenged than their di↵use counterparts. This result is in stark contrast to theories where the government independently evaluates the value and strength of cases, since the government alone would prefer to challenge broader cases that benefit more firms. This is consistent with the qualitative evidence emphasizing that firms' collective action problems inhibit dispute escalation (Counsel for WTO Disputes, 2021), and that product specific barriers help resolve them. It also also supports the argument by Sha↵er, Elsig, and Puig (2017) that private actors influence and shape how  Random e↵ect models calculated using xtmelogit with STATA14. Random intercepts calculated for groups at the industry level, defined as the ISIC3 4 digit industry. Canada is the omitted comparison. P-values are calculated using a two-tailed test and standard errors are displayed in parenthesis.
found neither to be significant, and the main results all retained significance. 24 These tests show that the significance of product-specific barriers and dominant firm capacity are robust to measures of size and employment of the industry and firm and suggest that governments are not selecting cases to benefit the largest producers or the biggest employers. To evaluate the substantive significance of the findings, I estimate the predicted probabilities of filing a WTO complaint given varying levels of product-specific barrier, dominant firm capacity, trade barrier distortion, and negotiation progress. I evaluate the change in the probability of dispute initiation for a shift from one standard deviation below the mean to one standard deviation above the mean for significant variables, or a shift from zero to one for indicator variables, which are reported in Table 2. The remaining variables are set to their mean, or a value of zero for indicator variables, except for the defendant country (Mexico) and distortion, which are each set to a value of one. 26 24 Results not shown here. 25 Due to data availability the number of observations fluctuates across models. In appendix, §5, all results are replicated using the same sample of 1407 observations. 26 Similar results are obtained when using other countries or a value of zero for distortion.

27
The predicted probability of filing a complaint with dominant firm capacity one standard deviation below the mean, when the hypothetical defendant is Mexico, is 0.06. The same probability with the dominant firm's capacity one standard deviation above the mean is 0.17. Similarly, the predicted probability of case initiation for a trade barrier that is di↵use is only 0.10, but the probability of a WTO dispute jumps to 0.34 when it is a product-specific trade barrier. These examples highlight the importance of product-specific barriers and dominant firm capacity for overcoming the collective action problems faced by firms that are considering making litigation contributions, in addition to the significant e↵ects of trade barrier distortion and negotiation progress, which are also displayed in Table 2.
To test Hypothesis-5b, which states that increases in the average value and capacity of firms within an industry will make dispute initiation more likely, I progress through the same model specifications as Table 1, but now include the variable for average firm capacity, as shown in appendix, §9. Average firm capacity has a strong positive e↵ect on dispute initiation that is robust to the full range of controls for competing theories and country specific e↵ects. The substantive influence of average firm capacity on the predicted probability of dispute initiation is about three-quarters of the e↵ect of dominant firm capacity. This strong, but smaller e↵ect than dominant firm capacity is consistent with the implications from the contribution game. Change in predicted probability is calculated from Model 5 of Table 1. Estimates and 95 percent confidence intervals are calculated using a quasi-bayesian simulation that samples 2000 times from a distribution based on the coe cients and variance. Changes in predicted probabilties represents a shift from one standard deviation below the mean to one standard deviation above the mean of the variable, or a shift from 0 to 1 for distortion and product-specific barrier. All other variables are set to their mean, or a value of zero, except for the defendant country (Mexico) and distortion, which are each set to a value of one.
Taken together, the interviews and regression analysis are consistent with the theory of firms using litigation contributions to influence the WTO dispute escalation process. While the statistical analysis alone cannot test the micro-level mechanisms of the theory, the results are remarkably consistent with the qualitative evidence where firms signal the strength and value of cases, where product-specific trade barriers that do not present collective action problems are most likely to be challenged, and where industries with high capacity dominant firms and high average capacity are most likely to make litigation contributions and seek dispute initiation. Although empirical analysis of the largely confidential trade dispute escalation process is inherently challenging, the consistent accounts of leading trade experts and government o cials, provide strong support for the theory.

Conclusion
The theory and evidence presented in this paper has direct implications for our understanding of To examine the strategic interaction of firms and the government, I formalize the argument in a simplified game that demonstrates when and why firms' signals are credible. I begin with a basic form of the model with just two players, Firm and Government. The subscripts F and G are used to identify the actions of each respective player. The model shows that a contribution threshold exists, such that the signal is su ciently costly so the government can infer the credibility of the message.
The model begins when the players are presented with a potential WTO dispute and nature determines whether the particular case is strong or weak, ✓ S or ✓ W . 1 The potential case is exogenously given, as is the total cost of litigation, the probability the case is won, and the value of winning the case. The players' priors over the strength of the case are that with probability P the case is strong and with probability 1 P the case is weak, and the total litigation cost for the case is L. The trade value of winning a case is defined as the benefits from trade with the trade barrier removed minus the benefits from trade with the trade barrier in place, which is written as ⌧ j (0) ⌧ j (1), where j 2 {F, G}. The value of a case will depend on the level of distortion caused by the trade barrier and magnitude of the a↵ected trade flow, but for simplicity the payo↵s can be normalized such that ⌧ j (1) = 0 and ⌧ j (0) = 1, so the trade gains for both players are 1 if the case is won. Additionally, the model allows for the possibility of externalities to the government, which can take many forms, such as restricted foreign aid or greater domestic political support for the government from appearing to stand strong with domestic industry, which can be incorporated in the externality term, E G .
The model captures the information asymmetry between the firm and the government, with the firm receiving a message about the strength of the case, m 2 {s, w}, but the government does not. The firm's private information means it has more accurate knowledge of the probability of winning a case than the government. 2 Once the firm knows whether the case is strong or weak, the firm decides to contribute or not. If the firm contributes it pays a cost, L F > 0. The firm selects the exact cost it pays, which is deducted from the total cost of litigation.
The firms contribution functions as a bureaucratic subsidy that reduces the litigation cost remaining to be paid by the government. This is captured in the model by the litigation cost function, where L L F = L G , which states that as the firm pays more the government's share of the litigation cost is reduced. 3 The sequencing of the model is shown in Figure 1, which proceeds with the following steps.
After the firm decides how much to contribute, the government observes the firm's action and then the government is faced with the decision whether to initiate a WTO complaint or not, I G or ¬I G . If the government initiates it pays L G = L L F , and has an expected payo↵ If the government does not initiate it has an expected payo↵ of EU G (¬I G ) = ⌧ G (1). The payo↵s capture two of the most important elements of the case selection process, the probability the case is won and the value of the case.
• I G = Government initiates a WTO complaint.
• L = Total litigation cost for a case.
• L j = Litigation cost paid by j 2 {F, G}.
• ⌧ j (1) = Value of trade for j with the barrier in place.
• ⌧ j (0) = Value of trade for j with the barrier removed.

Weak& m&=&w""
As is typical in signaling games, there are a broad range of potential equilibria if one allows the actors to have any o↵-the-path-beliefs, but the core insights of the model can be maintained if the beliefs are restricted to those that are intuitively reasonable. 4 The most interesting results of the theory are from the set of cases where the government would be unwilling to initiate the case without a litigation contribution from the firm. The first set of such cases are those where the expected profit to the government is less than the total litigation cost. In a unitary actor model, these cases would be viewed as unprofitable, however the equilibrium result shows that the firm's litigation contribution can alter the expected payo↵s to the state by mitigating the resource constraint, making such cases profitable to the government. 5 A second, and potentially overlapping, group of cases are those where the government's prior belief regarding the strength of a case is su ciently low that the government does not believe case initiation is profitable. In this group of cases, if the firm knows that the case is strong, it can credibly signal the strength of case to the government, thus altering the expected payo↵s of the government and motivating the government to initiate the case. 0)) the Firm has contributed a su cient amount, such that the government now believes its expected payo↵ from case initiation is greater than or equal to zero and the government initiates the case. For simplicity, I will refer to this contribution threshold for the firm as L ⇤ F . In order for the firm's signal to be credible, the equilibrium condition requires that the litigation contribution of the firm, L ⇤ F , must be greater than the firm's expected profit from a weak case. The litigation contribution threshold means that, on the path, the government does not believe the case is strong when the firm contributes less than L ⇤ F . This means there exists a separating equilibrium where firms will only contribute L ⇤ F when they know a case is strong. 6 The equilibrium contribution levels for both the firm and government (for P = .5 and E G = 0, and a given set of beliefs discussed in section 2 of the appendix) are shown in Figure   2. The figure shows that if the litigation cost is low enough (L  .5), then the firm pools on contributing nothing and the government pays the full amount and initiates on its own. 7 In the next portion of the parameter space (.5 < L < .75), the firm pools on contributing L .5, which is just enough to make the government initiate the case, but does not convey a credible signal and thus the government does not update its beliefs about the strength of the case. In the next portion of the parameter space (.75  L  1.5), the firm strategies fully separate, with contributions equal to zero when the case is weak (right panel) and contributions equal to L ⇤ F if the case is strong (left panel). In this range of potential disputes, the firm's signal is informative and allows the government to only pursue cases that are strong. Lastly, once cases become prohibitively costly (L > 1.5), the firm again pools on contributing nothing and the government does not initiate.   Figure 2 plots the equilibrium contributions for the firm and government over a range of total litigation costs (L), set values for P and E G , and reasonable set of beliefs discussed in the following section. Comparing the left and right panels of the figure shows that for certain ranges in the parameter space firm strategies pool, but in the middle range of the parameter space (.75  L  1.5) the strategies fully separate based on the strength of the case. The litigation costs include all potential positive values, but for the selected parameters the firms and government pool on contributing nothing when L > 1.5. about the strength of the case. The boundary shifts to the right as the government's prior belief that the case is strong increases.

Equilibrium Solution
Overview of equilibrium solution: The main intuition of the model can be captured by considering equilibrium outcomes across the regions of the parameter space for a reasonable set of beliefs. In the next section of the appendix, I discuss the equilibria that exist with a broader set of beliefs and refinements, which further illustrate that the primary implications of the model hold when the Intuitive Criterion is applied.
The equilibrium consists of three regions of interest of the parameter space, which I define as the low cost, middle cost, and high cost cases: , the firm plays the same strategy regardless of whether the case is strong or weak and the case is initiated by the government.
In the middle parameter space, where L max{ c ✓ W (⌧ G (0)) + E G + c ✓ W (⌧ F (0)), .5p + .5} 0)) the firm contributes if the case is strong, which leads to the government initiating the case, but will not contribute if the case is weak, in which case the government does not initiate.
Lastly, for high cost cases, where L > b ✓ S (⌧ G (0))+E G + b ✓ S (⌧ F (0)) the firm and government do not contribute and the case is not initiated, regardless of whether the case is strong or weak.
In the low and middle cost cases, if the firm chooses to contribute, it selects a litigation contribution from L F > 0 that is the minimum amount to convince the government to initiate the case. The equilibrium contribution levels for both the firm and government for strong and weak cases (for P = .5 and E G = 0), with the governments beliefs defined as shown below, are presented in Figure 1 of the paper and are proven in the following pages.
Low Cost Parameter Space: 1. Firm's Strategy: 2. Government's Beliefs: Assume the government does not update its prior, since the firm is pooling and thus no information is conveyed. (Deviations from this set of beliefs are discussed in the next section.) 3. Government's Strategy: If L G  .5p + .25 then: If L G > .5p + .25 then:

Sequential Rationality
We now consider whether the beliefs and strategies of the actors are sequentially rational.
If L  .5p+.25, the cost to initiate the case is low enough that the government initiates the case on its own and the firm has no incentive to deviate and contribute, since any additional contribution will be an unnecessary cost to the firm.
No incentive for the firm to deviate by contributing.) If L > .5p + .25 (but still in the "low cost" space), the government will not initiate on its own, so the firm has an incentive to pay just enough to convince the government to initiate the case, up to the expected profit of winning a weak case.
1. Firm's Strategy: The specification of L F for the strong case is the threshold requirement that the firm contributes at least as much as it could expect to gain from pursuing a weak case ( c ✓ W (⌧ F (0))), and thus the firm does not have an incentive to blu↵ with weak cases.
The firm may contribute more than c ✓ W (⌧ F (0)) when the total cost of initiating the case is higher, in which case the firm contributes L b ✓ S (⌧ G (0)).

Government's Beliefs:
The government's beliefs about the case are that a case is strong if the firm contributes at least c ✓ W (⌧ F (0)) and otherwise the government believes the case is weak. (Only this set of beliefs survives the Intuitive Criterion, as discussed in the next section.) 3. Government Strategy: First we consider the government's strategy when the firm has contributed enough to meet the litigation threshold, L F c ✓ W (⌧ F (0)).
The government will initiate if the expected utility of initiating is greater than or equal to not initiating.
Substitute and rearrange using: Next we consider the government's strategy when the firm has contributed, but not enough to meet the litigation threshold, Substitute and rearrange using: Next we consider the government's strategy when the firm has not contributed (L F = 0).

Sequential Rationality
We now consider whether the beliefs and strategies of the actors are sequentially rational.
If the case is strong and the litigation cost is in the middle range, then the firm will contribute and the government will initiate in equilibrium:

Payo↵ from deviation:
If the case is strong and the firm deviated and played L F 0 < max{ c ✓ W (⌧ F (0)), If the firm deviated and chose to contribute less than the contribution threshold (including L F = 0), then the government would not initiate and the firm's expected utility would be L F 0 , which is weakly less than the expected utility of initiating (0  .75 L F ), so the firm will not deviate.
Furthermore, the firm would never pay more than the contribution threshold, since any additional expenditure cannot influence the government's decision to initiate (since the government already chooses to initiate once the threshold is met), and thus any additional contribution only reduces the firm's expected payo↵.
Next we consider whether there is an incentive to deviate from the equilibrium if the case is weak in the middle cost range.
If the case is weak and the litigation cost is in the middle range, then the firm will play L F = 0 and the government will play ¬I G : Payo↵ from deviation: If the case is weak and the firm deviated and played L F 0 > 0 and 0)), then the government believes the case is weak and plays ¬I G .
If the firm deviated and contributed some amount that was less than the threshold, the government would not initiate the case, and the firm's expected utility would be L F 0 , which is less than 0, so the firm will not deviate.
If the case is weak and the firm deviated and played L F 0 c ✓ W (⌧ F (0)), then the government will believe the case is strong and play I G . ) If the firm deviated and contributed at least the contribution threshold, the government would then initiate the case, and the firm's expected utility would be c which is no better than the payo↵ for not contributing (0), so the firm will not deviate.
This demonstrates that the firm does not have an incentive to blu↵ and try to convince the government to initiate weak cases.
High Cost Parameter Space: 1. Firm's Strategy: Assume that on the path the government does not update its prior, since the firm is pooling and thus no information is conveyed. (I assume the government is agnostic o↵ the equilibrium path, and explore other beliefs in the following section.) 3. Government's Strategy:

Sequential Rationality
We now consider whether the beliefs and strategies of the actors are sequentially rational.
If the case is strong and the litigation cost is in the high cost range, then the firm will not contribute and the government will not initiate in equilibrium: Payo↵ from deviation: If the case is strong and the firm deviated and played L F 0 > 0, let us consider the maximum amount the firm would ever contribute, (the most the firm would ever contribute, since it is the most it could ever expect to win from the case). Since we are o↵ the path let's assume the most favorable beliefs of the government for the case, and thus the government believes the case is strong.
Given that the firm contribution, even when it is at the maximum where the firm could break-even, cannot convince the government to initiate the case in the high cost range, there is no incentive for the firm to deviate and contribute in any amount if the case is in the high cost range.

Alternative Beliefs and Refinements
The preceding section considered an equilibrium when the government's beliefs were restricted to reasonably assume the firm, as the first mover in the game, acts in an e cient manner and that the o↵-the-path beliefs were unsurprising. However, consistent with a general class of signaling models (Cho and Kreps, 1987), a broad range of equilibria exist given the range of potential beliefs. However, it can be shown that the substantive intuition of the model is maintained when the Intuitive Criterion is applied.
In this section I consider additional equilibria that may exist under di↵erent belief structures, and demonstrate that the key substantive implications of the model hold under a broad range of beliefs. I progress by considering each portion of the parameter space, using E G = 0 and p = 0.5 as was done in the previous section. With regard to the Intuitive Criterion as applied to the game's setup, it is clear that a firm that knows the case is weak can never profitably deviate by paying L F > c ✓ W (⌧ G (0)). Therefore we can restrict the inferences drawn by the government when L F > c ✓ W (⌧ G (0)), such that the government will always believe the case is strong in such circumstances.
Within the lowest cost parameter space, when L < c ✓ W (⌧ G (0)) + E G the Firm pools on L F = 0 given that for any government beliefs, the government knows it is profitable for the government to unilaterally initiate the case.
However, in the upper portion of the low cost parameter space, where: the government will always initiate the case, but there may be equilibrium where the Firm contributes L F > 0. To demonstrate, we may consider the conjecture that L = 0.45 and L ⇤ F = 0.1. If the o↵-the-path beliefs of the government are such that if L F < 0.1 the government believes the case is weak, then neither type has an incentive to deviate. If the firm deviated and contributed L F < 0.1 the government would believe the case was weak and choose not to initiate, and the firm would end up losing their contribution. Thus neither side has an incentive to deviate. That said, the general intuition in this portion of the parameter space remains unchanged -the government will unilaterally initiate very inexpensive cases, and those cases that are slightly more expensive will also be initiated, but potentially with some level of contribution by the firm. The only substantive di↵erence is exactly when, and how much, the firm contributes, but the general process remains unchanged.
When considering the middle cost parameter space, numerous equilibria exist. First, we can consider a situation where the government only believes the case is strong for some , it is profitable for the firm to contribute L ⇤⇤ F when the case is strong and to contribute L F = 0 when the case is weak. This equilibrium follows the same logic as that described in the low cost parameter space. Importantly, the substantive implication remains largely unchanged. In this middle cost parameter space, firms will contribute some threshold amount when they know the case is strong, though the exact threshold depends on the beliefs of the government. Just as described in the previous section of the appendix, the firm only contributes the threshold amount when they know the case is strong, so the government learns from the signal, while also benefiting from the litigation contribution that functions as both a signal and a bureaucratic subsidy.
Lastly, we can consider the high cost parameter space, where: 0)). Given the cost of initiating the case, there are no beliefs that make it profitable for the firm to contribute, since the cost of litigation is greater than the total potential gain for the firm and government. Even if both the firm and government believed the case was strong, it would not be profitable to contribute.

Multiple Firms with Incomplete Information
This section discusses the role of private firm contributions when there are heterogenous firms within an industry and uncertainty among the firms about how much they each value initiating the WTO dispute. Rather than focusing on the government's uncertainty about the strength of the case, this extension holds the strength of the case constant and examines how uncertainty over heterogeneous valuations by firms within an industry a↵ect the likelihood that firms contribute a su cient amount for the case to be brought.
Building from the previously discussed model, I examine the set of cases where the litigation cost is su ciently high such that the government will not initiate the case on its own (L > .5p + .25). It is helpful to consider the government's decision to initiate the dispute as a provision of a public good, where each firm in the industry values initiating the The values for individual firms are independently drawn from a continuous distribution F , and each firm only knows its own value, although each knows the distribution from which the values were drawn. 9 In the two player game, it was shown that for given parameters, the firm could contribute L ⇤ F which was the necessary threshold for the government to initiate the case. In this extension, L ⇤ F is the cost of the "public good," or the necessary contribution threshold that the firms must reach for the government to bring the case. I allow L F to be the sum of litigation contributions (L i ) from all firms within the industry. Because firms' litigation contributions involve sinking costs into the litigation process through fact finding and preparation of materials, I consider firms' contributions to be non-refundable in the model. This means that if firms contribute and fail to reach the necessary threshold (L ⇤ F ), the costs are sunk. Given this setup, which is based on a set of realistic assumptions drawing upon how the WTO litigation process functions, this game is best described as a contribution game with uncertainty and heterogenous preferences. The game is formalized through the existence of N 2 firms, where each firm i, i, ..., N , only knows his own value (V i ) for initiating the case, which will be brought if the the firms contribute a combined L ⇤ F . This type of game has been analyzed in the generic form (for the provision of any discrete public good) by  in "Private Provision of Discrete Public Goods with Incomplete Information." Since the game has been thoroughly analyzed elsewhere, I draw from the earlier insights and discuss the key implications for dispute initiation in the WTO. To illustrate the connection between between the specifics of the game here and the work of , the following lines specify how the games are linked.
1. In each game there are N 2, where i, i, ..., N , knows its own value (V i ), but only knows the distribution of others' values (F ).
2. The cost of providing the public good for Menezes, Monteiro, and Temimi (2001, 496) is c, which is equivalent to L ⇤ F . This holds, given that for any set of constant parameters there exists an L ⇤ F such that the government will initiate the case.
3. In each game, the individual players make a simultaneous decisions to contribute, where the contribution is any amount greater than or equal to zero (L i 0). 4. The "public good" is dichotomous, as shown by the government's decisions to either initiate (I G ) or not initiate (¬I G ) the case.
In this set up,  prove the following theorem.
Theorem 1. Suppose F: [0, 1) ! R is a continuous distribution. Suppose there are N 2 players for a project with cost c > 0 and that F (c) < 1. Then there is an equilibrium strategy for the contribution game.
Theorem 1 implies that when the cost of the provision is not prohibitively high as to prevent a single player from providing the good, there always exists an equilibrium where a player with a su ciently large valuation provides the good himself.(emphasis in original) From this, we can consider the situation when there are multiple industries, each with multiple firms. All else equal, in expectation the industry that has the firm with the highest valuation (and ability to contribute) will be the most likely to have at least one firm where v > ↵, and is thus the most likely industry in which a firm would contribute and a dispute would be initiated. Based on this, in the empirical section I examine how the size of dominant firms within industries a↵ects the likelihood of WTO dispute initiation. Menezes, Monteiro, and Temimi (2001, 503) also show that, when no single firm can a↵ord to pay the cost of providing the public good, and the cost of the good is high enough ("slightly above the aggregate mean of the valuations") then the unique equilibrium of the game is to contribute nothing.
Theorem 2. If the public project cost is higher than C N , then the unique equilibrium of the contribution game is the strong free riding equilibrium, i.e.
This theorem implies that as the average value for firms within an industry declines, it is increasingly likely that they contribute nothing and the dispute will not be initiated.
Conversely, as the average value for firms within an industry increases, it is increasingly likely that they contribute and the dispute will be initiated. Based on this, in the empirical section I examine how the average size of firms within industries a↵ects the likelihood of WTO dispute initiation.

Sample Variation in Empirical Models
Due to data limitations, the number of observations varies in the empirical analysis of the paper. In the following table, the models from Table 1 are replicated, but use the same constrained sample across all models. The results show that the main results are not an artifact of the changing samples across models. This table reports results using the smallest subset of data with results reported in Table 1 of the main paper. Random e↵ect models calculated using xtmelogit with STATA14. Random intercepts calculated for groups at the industry level, defined as the ISIC3 4 digit industry. Canada is the omitted comparison. P-values are calucalted using a two-tailed test and standard errors are displayed in parenthesis.

Industry Level Fixed E↵ects
It is possible that certain industries are more or less likely to engage in trade disputes, regardless of dominant firm capacity. In the main analysis this concern is addressed by using a multilevel random e↵ects model, which allows each industry to have its own intercept, while allowing for the e↵ects of the key variables of interest to be analyzed across the dataset.
However, to isolate the e↵ect of within industry variation the following table replicates the models from Table 1, but uses fixed e↵ects models, with fixed e↵ects for each industry. In the main paper, the choice to use the random e↵ects model was evaluated using a Hausman test, comparing the random e↵ects model to a fixed e↵ects model , with both the random e↵ects and fixed e↵ects at the ISIC3 4-digit level. The finding showed the null hypothesis -that the random e↵ects model is consistent -cannot be rejected (prob> Fixed e↵ect models calculated using xtlogit with STATA14. Fixed e↵ects are at the industry level, defined as the ISIC3 2 digit industry. Canada is the omitted comparison.
P-values are calucalted using a two-tailed test and standard errors are displayed in parenthesis.

OLS Regression Analysis
To further probe the robustness of the results, I also replicate the analysis from Table 1 of the main paper, but now do so with ordinary least squares (OLS) regression. The OLS results show that the results are consistent regardless of model choice. Barrier A potential concern with the main analysis is that the trade barrier-year observation could bias the results given the structure of the data. The concern would be that trade barriers that do not escalate to the WTO remain in the dataset, whereas those that are brought to the WTO exit. This would lead to an overrepresentation of barriers that don't escalate, which could alter the findings. I address this issue by conducting an analysis where the data is collapsed to a single observation for each trade barrier. This abandons the trade barrier-year setup, and so I also drop the duration variable that was originally included to address the fact that barriers remain in the dataset over time. For the variables in the main models that change over time, such as the dominant firm's capacity, I take the average value of the variable across the years from original dataset. The results are included below as Table 4 and are consistent with the main analysis, showing that the paper's findings are not sensitive to whether the trade barrier or the trade barrier-year is the unit of observation. 10  Random e↵ect models calculated using xtmelogit with STATA14. Random intercepts calculated for groups at the industry level, defined as the ISIC3 4 digit industry. Canada is the omitted comparison. P-values are calculated using a two-tailed test and standard errors are displayed in parenthesis.
To assess the substantive e↵ect of average firm capacity, I replicate the analysis from Table 2 of the paper using average firm capacity instead of dominant firm capacity, which is shown in Figure 6. The results are consistent with expectations, with average firm having a substantively significant e↵ect, though it is substantially smaller than dominant firm capacity, as expected. Change in predicted probability is calculated from Model 5 of Table 1. Estimates and 95 percent confidence intervals are calculated using a quasi-bayesian simulation that samples 2000 times from a distribution based on the coe cients and variance. Changes in predicted probabilties represents a shift from one standard deviation below the mean to one standard deviation above the mean of the variable, or a shift from 0 to 1 for distortion and product-specific barrier. All other variables are set to their mean, or a value of zero, except for the defendant country (Mexico) and distortion, which are each set to a value of one.

Potential Firm Counter-Lobbying
One potential complication to the model would be incorporating firm counter-lobbying. However, this is omitted from the model since such counter-lobbying does not appear to factor into the dispute escalation process, except on very rare occasions. To understand why, I consider counter-lobbying from both a theoretical and empirical approach.
From a theoretical standpoint, domestic firms with the potential to counter-lobby would be most likely to do so when initiating a dispute is expected to have a negative e↵ect on the firm's economic situation. This could occur if they believed the dispute would hurt their exports or raise the cost of imports on their intermediate goods. In either case, the most likely mechanism through which counter-lobbying would shape the decision process is by providing information about the economic e↵ects of the dispute. Counter-lobbying could thus lower the expected value of the case, which would reduce the likelihood the case would be initiated. However, there are limited cases where domestic firms with the ability to counter-lobby the trade bureaucracy would be beneficiaries from a trade barrier violating WTO law imposed by foreign government. For example, when a new trade barrier harms US exporters, it is typically because their access to export markets has been curtailed. The most likely firms to benefit from such a policy are import-competing firms from the country imposing the trade barrier, or exporters from other countries that are not a↵ected by the barrier. In either case, most firms benefiting from the trade barrier, who would have an incentive to lobby against initiating a dispute, would be foreign companies whose interests would not give them significant standing to lobby domestic bureaucracies.
This situation is somewhat complicated by multinational corporations, who may seek to take advantage of di↵erences in trade law across countries; however, empirically, the one quantitative study that examines firm-level lobbying and WTO disputes, found that total lobbying expenditures toward the US government by Fortune 500 companies was nearly seven times higher by firms supporting the complaint than those opposed to it (Ryu and Stone, 2017). While existing analysis only measures aggregate lobbying once a dispute has already been initiated, it suggests that any lobbying against complaints is relatively minor when compared to the e↵orts of firms who advocate in favor of WTO disputes.
Finally, it is worth noting that counter-lobbying before a WTO complaint is initiated is even rarer than counter-lobbying once a complaint is initiated. In the 38 author-interviews conducted, counter-lobbying prior to dispute initiation was only identified by a single interviewee, whereas trade experts and government o cials almost universally agreed that counter-lobbying was an exceptionally rare or non-existent practice during the WTO dispute initiation process, as shown in the quotes in Figure 3. The single case of counter-lobbying identified in the interviews involved a potential challenge against sanctions imposed by the US and EU where the government was "lobbied by private law firms that were concerned that WTO dispute settlement would be overtaken by sanctions disputes" if a dispute was initiated (International Trade Attorney 2021f). This was not a case of firms from industry counter-lobbying to protect their economic interest, but was instead a case of lawyers counter-lobbying because they were concerned about the future state of WTO dispute settlement. In hindsight, the o cial noted that choosing not to initiate the dispute "was a major mistake" (International Trade Attorney 2021f). Given the existing evidence and limited domestic standing of most firms who could potentially oppose filing a dispute against a foreign country's trade barrier, I focus my analysis on the role of firms pursuing dispute escalation.

Figure 3: Expert Quotes Regarding Counter-Lobbying
I did not personally ever see cases saying "don't bring this case" (Assistant General Counsel, United States Trade Representative, 2021).
No firms don't counter lobby. Never heard of firms counter lobbying (Deptartment of Commerce Official, United States, 2021) No I haven't seen that a firm ever comes forward and asks the government not to bring a case, that someone else wants to bring (International Trade Lawyer, Egypt, 2021) So no, there has been no cases I'm familiar with where firms lobby against the case (International Trade Lawyer, United States, 2021).
I have not seen any case of counter lobbying by an industry or association or firm, but I guess it could happen (International Trade Attorney, China, 2021).
Counter-lobbying doesn't happen when bringing cases, but the government will consider the different stakeholders they have to deal with (Senior Official familiar with WTO and Airbus-Boeing Dispute, 2021).
It's the government who is reluctant. So no, industries don't lobby against (International Trade Lawyer, United States, 2021).
Can't think of any instances of counter lobbying (Assistant General Counsel, United States Trade Representative, 2021).

Resource Constraints
The budget constraint is a real challenge for all governments when it comes to WTO litigation, though it is a greater challenge for some than others. Throughout the author-interviews it was often noted that the US and EC have more capacity than other countries to manage trade disputes, but it was also explicitly noted that both have insu cient resources to independently manage their high case load. One USTR o cial summed it up, saying "At the USTR we have very limited resources" (Assistant General Counsel 2021a Budget constraints are also a significant challenge for other countries considering challenging trade barriers through the WTO. For example, in Mexico it was noted that "The budget constraint is very real... Government tells them [the private firm] to just pay for the case and lawyers" (General Counsel 2021). The interviews show that such budget concerns were a persistent challenge, as shown in Figure 1 of the paper, which shows additional references to budget constraints in the United States, Japan, and other countries.

Sta↵ Turnover
An additional challenge faced by many governments is that they cannot retain trade experts who are able to identify the strength and quality of cases, and thus the governments are reliant on private firms to signal the strength and provide litigation support. Though not a problem for all countries, "The way the diplomatic career is setup in many countries actively discourages specialization, which is what you need for WTO dispute settlement" (International Trade Attorney 2021a). Furthermore, the "Rotation of sta↵, especially for countries that don't frequently use the WTO system, will have people move on and so the current government o cials won't have the expertise" (International Trade Attorney 2021b), which makes the government more reliant on private industry and private lawyers to build the case to bring to the WTO. This challenge was reiterated by numerous interviewees, with another noting that the problem exists "Not only for developing countries. Many countries face this problem, because they hire someone, but they move on... The problem is in poor countries, but also others is the frequent change of sta↵ (International Trade Lawyer 2021c).

Increasing Complexity of WTO Litigation
The WTO dispute settlement process has become increasingly complex and governments have become more reliant on firms over time. Although the process of WTO disputes may have become more regularized, the fact finding burden and costs have increased dramatically as well. The author-interviews found that respondents were unanimous in their opinion that the the dispute settlement process at the WTO has become more complex over time.
Speaking in an interview with the author, a USTR Assistance General Counsel confirmed this, saying "cases have become more complex over time and taken on more of a legal character, with procedural things that we didn't see 15 years ago" (Assistant General Counsel 2021a). Similarly, an o cial from Japan's METI noted the costs have been increasing over time and that "industry has had to play a larger role" (Assistant General Counsel 2021a).
The increasing complexity and costs of WTO disputes is recognized by WTO panelists, government o cials, and private lawyers, as is shown in Figure 4. This means that the period from 1995-2004 analyzed in the paper represents a conservative test for the importance of private firms, given that governments have become more reliant on private firms for the increasingly fact intensive and expensive cases at the WTO. There are certain trends in WTO. The first is the increasing complexity of the cases, which has implications for the time it takes to resolve the issue and the expertise you need to mobilize (International Trade Lawyer, Egypt, 2021).
The WTO process has become more litigious as the WTO has been basically unable to write and interpret new rules. The panel processes have become more fact intensive undoubtably (WTO Panelist, 2021).
The cases have absolutely become more complex. It's not just the "low hanging fruit" was picked first, but the case process and the AB have become much more complex over time (WTO Adjudicator, 2021).
Things have generally become more complex and technical over the years. Trade barriers used to be the tariffs, but that's not usually the biggest issues anymore (Assistant for WTO and Multicultural Affairs, United States Trade Representative, 2021).
The cost of cases has been increasing year over year. In the early years of the WTO the reports were relatively short, but recently the cases are 100s of pages long. So I think the cost has been increasing. Government budget has increased, but industry has had to play a larger role (Ministry of Economy, Trade, and Industry Official, Japan, 2021).
The increasing complexity also means that it is often harder for the government to assess the strength and value of cases on their own. In practice this means that most potential cases that would be considered for WTO disputes do not fall in the low parameter space, discussed in §2 of this appendix. Instead, governments are increasingly reliant on firms to help them assess the strength and value of the potential dispute. Indeed, the interviewees emphasized their reliance on private firms, noting "Most of the cases brought to the WTO come from a demand from the private sector" (Ambassador, 2021) and that "Most WTO litigation involved governments that do not have that expertise, so the private law firm advice has historically been really important" (International Trade Attorney, 2021e). Even in the US, where the USTR has some of the most experienced government lawyers and trade experts, it was noted the "USTR can often handle the legal case, but they rely on the technical information about how the market works, and support and partnership in developing arguments" (Assistant General Counsel, 2021a). The reliance of the government on firms to provide the information about the legal issues, arguments, and values at stake emphasize the importance of information provision to the government in the dispute escalation process.

The US Case and NTE Selection E↵ects
As discussed on page 22 of the manuscript, there are advantages and limitations of using the US case for the quantitative analysis. From a pragmatic perspective, the data from the National Trade Estimate annual report provides a useful set of trade barriers, given us a set of potential claims that could be initiated at the WTO. This data can also be matched with firm-level data using Compustat, providing us important variation on one of the key independent variables. From a case selection standpoint, the US case also has aspects that correspond with being an influential and typical case (Seawright and Gerring, 2008), both of which provide advantages. The case provides valuable variation on the independent variables, which is due to both the variation in firm-level attributes, but also the number of potential disputes that can be coded, allowing us to compare product-specific to more di↵use barriers.
Second, since the US is the most frequent user of the WTO dispute settlement system, it represents a typical case in that the modal case at the WTO is initiated by the US. Of course, this also means that the US is unique in that it has more experience with dispute initiation, meaning that the USTR sta↵ generally has more expertise than sta↵ from many other countries, especially those with more limited WTO dispute experience. However, as discussed in the manuscript, this should make the US relatively less reliant on firms' expertise, which would potentially bias against finding a significant e↵ect on the role of firms influence in the dispute escalation process. As shown in Figure 5 of the paper, firms also play a role in bringing trade barriers to the attention of the government, including some of the barriers in the NTE. This is potentially concerning if a systematic bias in reporting of barriers that make their way to the NTE would cause us to exaggerate the influence of firms. However, since it is larger firms that are more likely to have the capacity to identify trade barriers in the first place, they are the most likely firms to report barriers that make their way into the NTE. This means that trade barriers that primarily a↵ect industries with smaller firms are the most likely to be underrepresented, but these firms are also the most likely not to be brought to the WTO.
Given this selection process, any bias caused by firms role in generating the NTE would lead the paper's analysis to underreport the significance of dominant firms, and thus not a major concern for this paper.
Although a complete analysis of alternative paths of trade dispute resolution is beyond the scope of this paper, it is worth considering what those options are and how they fit into the strategy of firms and governments. 12 When firms seek to have a trade barrier removed, their preference is to first pursue a low cost strategy to remove the barrier. One USTR Assistant General Counsel referred to this as the desire to keep things at "the lowest temperature possible" (Assistant General Counsel 2021a). This means that firms almost always engage in direct consultations with the government imposing the trade barrier or have consultations with the foreign government in coordination with a domestic agency prior to pushing for a WTO complaint, which is only one of the many tools firms use to shape trade policy.
One international trade attorney noted that "Industry would rather deal with it themselves, and so if there's a problem they would try to resolve it with the other government directly" (International Trade Lawyer 2021a). In the United States, one USTR o cial noted that firms sometimes "come to USTR in the regional o ce that focuses on that country" and the desk o cer for the country will evaluate bilateral options to resolve the issue (Assistant for WTO and Multicultural A↵airs 2021). The preference of the government at this early stage is generally aligned with the firm, since both would prefer to see the issue resolved without a WTO dispute or other form of escalation. For this reason, the empirics in the paper control for whether progress is being made in resolving the dispute through other mechanisms, since making progress through bilateral consultations is preferred to initiating a claim at the WTO. However, the same individuals noted that even when a trade barrier is brought to the attention of the government through an agency process, potentially coordinated through the ITA or USDA, "we need the firm to bring data to show the problem really exists, the magnitude of the problem... We can get high level information from industry associations, but we really need to talk with individual companies because what we need is confidential and proprietary information" . This is also true when the case has been litigated through a domestic court, but does not yield a 12 For an excellent analysis of dispute settlement forum shopping, see Busch (2007).
satisfactory result for the firm, and so the firm may then push to bring a WTO dispute.
If alternative venues of litigation do not resolve the dispute, the firm still plays a critical role in alleviating the resource constraint and providing information when deciding to escalate the complaint to the WTO, as was specified by one expert; "The only lawyer who has the knowledge on trade remedies cases is the domestic lawyer in the underlying case, but that lawyer is never from the government, and so the inexperienced government lawyer can't just step in" (International Trade Attorney 2021d). Even though domestic agency processes have been established in various forms across countries, a consistent refrain of government o cials is that they first try to resolve the issue bilaterally, but if the case moves forward to the WTO, the agencies are reliant on firms to provide the information and support to build the case.
When bilateral consultations are unable to resolve the dispute, then the firm must consider further options, including initiating a claim at the WTO. While the WTO option is one of the most costly, it is also viewed as less of an escalation than some other domestic options. For example, a USTR o cial said that "trade remedies or some other form of domestic dispute settlement are more provocative and hard edge than the WTO... The WTO is quasi judicial and quasi diplomatic, making it less provocative" (Assistant General Counsel 2021a). When it comes to other domestic options, at least in the United States, "Section 301 is pretty middle of the road. It sends a signal, but the actual substance is not punitive and can't dictate what happens" though sometimes "301 can be used to do the investigatory work to build a case that would go to the WTO" (Assistant General Counsel 2021a), which is also why the paper includes a control for Section 301 proceedings. Firms thus make a calculation about how much the case is worth and the likelihood of winning a dispute at the WTO, and whether a complaint at WTO is an e cient strategy relative to the other options. The path to dispute escalation almost always involves bilateral consultations, often coordinated with a domestic agency, but when these options do not resolve the situation, then the firm must decide whether it is worth contributing to the litigation process and signaling to the government the need to initiate a WTO complain.
If a dispute escalates to the WTO, then the government begins by filing a request for consultations. The parties then engage in consultations in an attempt to resolve the dispute.
If consultations are unsuccessful, then a panel is established to hear the case. The parties engage in adjudication and the panel may then issue a ruling on the case. The parties to the dispute have the option to appeal the ruling to the WTO appellate body, which then issues its own report. Once the dispute settlement body adopts the panel/appellate report(s), the process proceeds to the implementation phase. If a party was found to be in violation of their WTO obligations, they must revise their policies to implement the ruling, or in cases of non-implementation they may negotiate compensation or retaliation may be authorized. Although firms' contributions to the litigation process can improve the argumentation and quality of submissions to the WTO, not all firm contributions are helpful to the government and there can be potential downsides. Sometimes the firm and government are in conflict over which arguments to present, and the government then must exercise its gatekeeping role and make the final decision about which arguments to submit. The most common point of tension is over how many claims to raise, with private firms generally preferring to bring more claims than the government. The firm is typically only concerned about the specific dispute, but the government must also be concerned about how the arguments in today's dispute could be used against them in the future. For example, a Japanese o cial a rmed that "sometimes the government has to tell the outside counsel, we don't want to file a case based on that argument. The ministry needs to be consistent with the interpretation of the treaty, so if industry or counsel is inconsistent, even if it's a strong argument, the government has to be aware of those issues" (Legal Advisor to Ministry of Finance 2021). A former Assistant General Counsel to the USTR echoed this sentiment, noting that "Industry often wanted to take more strident or stringent steps than USTR wanted or needed to take," and so the USTR would have to be selective about what arguments and strategies proposed by the firm to integrate into its legal strategy (Assistant General Counsel 2021b).
In one of the most egregious cases of the private firm proposing an argument at odds with the government, there was "one instance where the outside firm was pressing for [the government] to make an argument that would be completely inconsistent with Canada's image. This was in the Brazil case, and they were asking [the government] to argue that Brazil was not a developing country for the purposes of the aircraft industry. There was a famine and there were literally children dying in Brazil, and Canada had provided aid" (Counsel for WTO Disputes 2021). In this case the government said that the argument was immoral and chose not to proceed with the claim.
Although the private firms sometimes propose strategies and arguments that are in tension with the long-term interests of the government, the government has the ability to override the proposals of the private firms. This is a notable di↵erence between WTO dispute settlement and transnational dispute settlement, such as ISDS. Since the government is able to use their gatekeeper status at the WTO to have the final say on which claims are raised in the submissions, this generally results in stronger arguments being presented and clearer submissions to the WTO than would occur in the absence of private firm contributions. It is worth noting that some countries with very limited capacity do not necessarily exercise significant gatekeeping status, and essentially "rubber stamp" the arguments prepared by private firms (General Counsel 2021). In such cases, there would not be any dispute without the private firms, but the arguments being presented are not necessarily thoroughly vetted by the government.

Variation across Contexts
The cross-national implications are clear when it comes to resource constraints, with governments being more reliant on the information provision and financial resources of private firms when the government has the least capacity to pursue WTO disputes. For example, throughout the interviews experts noted that countries such as Mexico, Ecuador, and Antigua were all reliant on private firms to finance the cases (General Counsel 2021; Ambassador 2021; International Trade Lawyer 2021b). 13 Although countries such as the US, EU, and Japan are also reliant on private firms, they generally have more resources than other countries and are thus able to share the financial burden to some extent. For example, Mexico may require the industry to sometimes pay the entirety of the litigation cost (General Counsel 2021, whereas the Japanese government is more likely to cover about a third of the litigation costs (METI O cial 2021). For governments with more in-house expertise and attorneys, they also have greater ability to screen the arguments of the cases, as opposed to acting as a rubber stamp for the private firms' case. This is why the USTR can always vet, and generally prepare, the final submissions to the WTO, whereas many less-resourced bureaucracies are heavily reliant on the private firms to prepare the case and argumentation, as discussed in greater detail in section 16 of this appendix.
Countries resource constraints are not static, and some countries proactively seek to increase their WTO expertise and litigation capacity. For example, China made substantial investments in developing both their government's capacity and also domestic firms' knowledge and capacity to pursue WTO complaints. Sha↵er and Gao (2018) detail the learning curve that China faced, noting that the government participated as a third-party in a multitude of complaints where they hired private law firms to help them build capacity and develop the necessary expertise to initiate WTO complaints. Interestingly, private firms and SOEs were taught about WTO law through an extensive series of seminars and outreach efforts so that they were better positioned to support WTO litigation. Sha↵er and Gao (2018, 163) found that "Larger Chinese companies independently saw the need to develop WTO knowledge... and built in-house expertise." For example, one large telecommunications company hired James Lockett, who previously worked for the U.S. Department of Commerce, to be their Vice President and Head of Trade Facilitation and Market Access (Sha↵er and Gao, 2018, 164). However, it was also noted that "Building in-house trade law expertise takes time and resources that most Chinese small-and medium-sized enterprises cannot a↵ord" (Sha↵er and Gao, 2018, 164). This is consistent with the theory's expectations that larger companies are better positioned to contribute to the litigation process. Furthermore, the increase in government and private capacity in China is consistent with the rise of China's role as a claimant at the WTO.
When it comes to information asymmetries, the role of private firms is greatest when there is a larger information asymmetry between the government and private market actors. As was noted in one of the interviews, when trade barriers a↵ect state owned enterprises (SOEs) the government has greater access to information than if the trade barrier a↵ects a private firm (International Trade Attorney 2021f). This means that in countries with a higher proportion of SOEs, such as China, the information asymmetry is less likely to be a critical component of the dispute selection process. That said, the importance of SOEs to the Chinese economy has declined; "private companies now represent around 54% of the country's GDP" (Sha↵er and Gao, 2018). The role of these private companies has increased since China's WTO accession. According to the research of Sha↵er and Gao (2018), once Chinese law firms had developed su cient expertise in WTO law, they increasingly represented private companies who seek to proactively fight foreign trade barriers.

Product Specific Barriers and Collective Action
The importance of collective action and coordination problems when firms address trade barriers was brought up by multiple experts in the interviews. One noted that industry associations are often unable to overcome the collective actions problem since "The association doesn't bring the case, because at the end of the day it depends on whose going to pay for it" and so having a trade barrier with a more concentrated e↵ect reduces the potential for free riding (International Trade Attorney 2021c). Additionally, when there is a product specific barrier "Normally there is one company that cares a lot and takes the lead" (International Trade Lawyer 2021e). A government o cial from Japan noted that "If the issue is product specific, or the barrier is limited to a↵ecting a single industry then there are not so many conflicting views. Firms work independently. They do not cooperate when asking for requests for consultations. Sharing information may result in conflicts of interest so they don't work together" (METI O cial 2021). From the perspective of private firms and government o cials, there is evidence that collective action problems are significant in the dispute settlement process, and that product specific barriers help reduce these challenges.
The evidence presented in the paper clearly shows that product specific barriers are more likely to escalate than barriers a↵ecting a broader range of products. What makes this so interesting is that the interviews repeatedly noted that the government's preference (on its own) would be "to go after structural and systemic issues. Typically these would be issues that a↵ect multiple industries;" however, consistent with the paper's theory, it was also noted that private firms "can get their specific cases brought" (International Trade Attorney 2021f). Similarly, an o cial familiar with the USTR priorities found that "An individual industry is almost always only concerned with the very narrow particular dispute or industry... The government wants to invest their resources in cases with broader impact" (International Trade Lawyer 2021e). Furthermore, a number of experts also noted that it was easier for the government to bring systemic cases than product specific cases to the WTO. The rational provided was that many governments (and the Advisory Centre on WTO Law) have su cient expertise for the broad legal theories needed for systemic cases, but they lack the resources and expertise for the fact finding needed for more specific cases (General Counsel 2021). Given that governments would prefer to pursue broad claims and find it easier to pursue systemic claims, the fact that product specific barriers are challenged regularly is especially surprising and consistent with the theory advanced in the paper.

Discussion of Ethical and Human Subjects Principles
The human subjects research included in this paper complies the with Principles and Guidance for Human Subjects Research outlined by the APSA. The qualitative interviews consisted of individuals who were o cials who had represented governments or international organizations. Each interviewee was provided a description of the research agenda, research protocols, and citation protocols in their invitation to participate. All interviewees volunteered to participate and did not receive any compensation. Details about the selection of interviewees and citation protocols that preserve the anonymity of the o cials is provided in the main text of the paper.
With regard to Principal 10 on the impact of the research on the political processes, we do not believe there is any reason to believe that our studies would have had an impact on political processes such as elections or policy creation. Respondents were asked to share their experience and knowledge of dispute escalation, and were generally providing information about cases that they had already worked on, and thus the interviewees should not a↵ect the dispute escalation process. Therefore the research is not seeing as presenting any information to interviewees that would alter their political behavior or political processes.

Qualitative Methods
As discussed in the paper, I conducted 38 in-depth interviews with trade experts from around the world. The selection of interviewees was guided by a number of goals. To expand upon the selection criteria discussed in the paper, I provide additional details on the qualitative methodology. To identify potential interviewees, a research assistant compiled a list of individuals who worked for government o ces responsible for trade, individuals who worked in government relations or in-house counsel for firms a↵ected by trade barriers, and attorneys who worked for firms with practice groups focused on international trade. This generates a diverse list of individuals who represented people with a variety of potential perspectives on trade barriers and disputes.
The research assistant then contacted those on the list via email and/or phone. If we did not receive a response from the first email, I followed up with an additional email or phone call. After multiple contacts, we had about a 50 percent response rate. Though the sample is not randomly generated, it was generated based upon purposive selection to ensure respondents represented a range of perspectives in the dispute escalation process, which can be especially useful for identifying recurrent relationships and themes (Lynch, 2013). For each interview, I followed an outline of questions, though I also allowed the interviewees to elaborate based on their expertise. Each respondent was first asked to confirm their professional experience as it related to trade policy and disputes. Depending on their position, they were asked to provide the perspective of the government and/or firm when assessing trade barriers. Interviewees were asked how they learned about trade barriers, how they evaluated which barriers were worth contesting, how firms and governments interacted (if at all) when considering challenging a trade barrier, whether they faced resource constraints, etc. Most interviewees had worked in numerous positions that were involved with trade disputes, so they often provided multiple perspectives.
Given the open-ended nature of the questions, there was ample opportunity for evidence to be gathered that would support or falsify the theory. For example, if o cials with the government had noted that they are generally able to identify trade barriers through their embassies and other o ces, that would have discredited the idea that firms are critical to identifying the presence of trade barriers. Similarly, respondents could have spoken about their government's ability to prepare and fund cases without significant private assistance, but that was not the case. Instead, the interviews repeatedly emphasized similarly aspects of the dispute escalation, which painted a fairly consistent picture of firms and governments interactions, though they also highlighted interesting variation across countries (discussed in §17 of this appendix).