Content uploaded by Arlo Poletti
Author content
All content in this area was uploaded by Arlo Poletti on Feb 18, 2021
Content may be subject to copyright.
1
The European Union and the political economy of enforcing international
trade rules
Forthcoming in European Union Politics
Emile van Ommeren, Department of Political Science, University of Antwerp
Arlo Poletti, Department of Sociology and Social Research, University of Trento
Dirk De Bièvre, Department of Political Science, University of Antwerp
Abstract
The European Commission keeps track of foreign trade barriers through its Market Access
Strategy. In this study, we examine some of the key political-economic conditions under
which the European Union decides whether and how to address these trade issues. Drawing
on an original dataset of (allegedly) illegal foreign trade barriers faced by European Union
businesses, we show that industries dominated by a few large companies are more successful
in gaining the support of the Commission to challenge these foreign trade barriers. Moreover,
we find that the European Commission’s strategy depends on the economic power
relationship with the trading partner: the European Union privileges negotiations when
seeking to enforce international trade rules against economically weaker states, while it
prefers to use litigation against stronger trading partners.
Keywords
Dispute settlement, enforcement, European trade policy, lobbying, World Trade Organization
2
Introduction
The European Union’s (EU) pivotal role in international trade relations is undisputed. The EU
is the world’s largest trading bloc, ranking first both as a trader of manufactured goods and
services and as source of, and destination for, international investments. Two factors have
crucially contributed to making the EU such a pivotal international trade player. First, placing
trade policy under supranational competence provided the European Commission (EC) the
authority to elaborate, negotiate and enforce trade relations with the rest of the world. Second,
the rise of the EU as a ‘great trade power’ was facilitated by the existence of a set of multilateral
trade rules – the General Agreement on Tariffs and Trade (GATT) - which proved very
successful in channelling and furthering the interests of powerful European exporting
constituencies by providing them predictable and non-discriminatory access to nearly all
foreign markets (De Bièvre and Poletti, 2013).
As tariff barriers declined worldwide, continuing to effectively promote the interests of
EU exporters required expanding the functional scope of the multilateral trade regime to a wide
array of new regulatory issues. In turn, this required effective enforcement mechanisms that
could be eventually activated to restore compliance with these new multilateral regulatory
commitments. The EU confronted this challenge in two ways. For one, it sponsored, together
with the United States (US), the creation of a quasi-judicial mechanism of enforcement of rules,
which materialized with the creation of the World Trade Organization (WTO) in 1995. In
parallel, in 1996 the EU adopted a new Market Access Strategy (MAS), strengthening its
administrative capacity to keep track of foreign trade barriers and enabling it to take full
advantage of such an institutional innovation at the WTO level (Poletti et al., 2016).
Despite the fact that the enforcement of WTO rules plays such a central role in both the
rhetoric and practice of EU trade policy, we know surprisingly little about the political and
economic factors underlying it. In particular, the EU trade policy literature has so far remained
largely silent on two key issues. First, under what conditions can European exporters count
upon EU policymakers’ support with respect to their demands to seek the removal of foreign
trade barriers? Existing works explaining cross-national variations in observed propensity to
enforce WTO rules as a function of the legal capacity and resources of WTO members
(Brutger, 2017; Guzman and Simmons, 2005) suggest that the EU may be better equipped than
less resource-rich trade actors in this regard. Over the years European firms have reported many
instances in which they were confronted with (allegedly) illegal barriers to trade in foreign
countries, and yet only a small subset of these cases have prompted action by EU policymakers.
This means that, as in other political systems, EU policymakers act as gatekeepers for the
demands of domestic exporters seeking the enforcement of international trade rules, which
begs the obvious question why they do respond to some of these demands, while in other
instances they refrain from taking action against its trade partners’ protectionist measures
(Hoekman et al., 2017). Uncovering the logic that underlies how policymakers select among
the various exporters’ demands is key to understand whether, and eventually in what ways,
access to enforcement of international trade rules is systematically biased in favour of
particular subsets of domestic constituencies in the EU.
Second, when EU policymakers do respond to exporters’ demands, what factors affect
the precise course of action that they then choose? One obvious way through which the EU can
seek the enforcement of multilateral trade rules is by initiating formal WTO disputes as
complainant. Yet, while judicial dispute settlement is used in some of these cases, in many
other instances EU policymakers seek the removal of foreign trade barriers using instruments
such as diplomatic missions, sending letters and raising matters in bilateral negotiations.
Shedding light on the determinants of how EU policymakers choose among these alternatives
is crucial to understand the EU’s role in international trade politics as it has the potential to
3
illuminate the conditions under which we can expect the EU to continue upholding a rules-
based global trading system.
Drawing on an original dataset on foreign trade barriers reported by European
exporters, this study addresses both questions. More specifically, we rely on the set of trade
barriers as reported by European businesses and recorded by the EC services in the EC’s
Market Access Database (MADB) after a preliminary screening of their actionability under
international trade law. This database includes information on the type of measures, the country
in which they were encountered, the product category and thus the sector which alerted public
authorities about the foreign trade barrier, as well as a description of what the EU is doing to
remove the barrier. We use these data to subject to empirical scrutiny a number of factors
explaining the political power of European businesses and the strength of the EU’s bargaining
position vis-à-vis the relevant trading partner. We thus provide the first systematic analysis of
potential cases out of which the EU selects and prioritises the issues that lead to actual trade
policy decisions.
Our findings support two main hypotheses. First, we show that EU policymakers are
more likely to cater to demands to enforce international trade rules when these come from
industries dominated by a few large companies. Second, we provide evidence in support of the
hypothesis that EU policymakers prefer to seek the enforcement of international trade rules via
legal means, i.e. acting as complainant in WTO dispute settlement, when the relevant trading
partner is economically strong, whereas they privilege more informal negotiations when
dealing with relatively weak states.
Our article makes several contributions. First, we contribute to filling an important gap
in the EU trade policy literature as we offer the first attempt to explain whether and how the
EU acts as an enforcer of international trade rules. We thus contribute to the literature on the
politics underlying the enforcement of international trade rules within other political systems
and from a cross-national perspective (Davis, 2012; Kim and Spilker, 2019; Sattler and
Bernauer, 2011; Yildirim et al., 2018). Second, rather than focusing on peculiar or unique
characteristics of the EU’s trade policymaking process, we embed the study of EU trade policy
in the broader literature on the political economy of trade policy (Dür et al., 2020). We show
that the political-economic logics underpinning the enforcement of multilateral trade rules in
the EU are not dissimilar to those found at play in the US and other political systems. Third,
our findings inform real-world relevant debates about the future of the global trading system,
and the EU’s role within it, in the face of rising protectionist sentiments and policies across the
globe. Since the financial crisis of 2008, there has been a rise in the number of trade restricting
measures around the globe (Hoekman et al., 2017). The world-wide coronavirus crisis and the
ensuing trade restrictive measures only add to the urgency of the question under which
conditions the EU is likely to consistently act as a fervent advocate of the rules-based trading
system.
The EU and the enforcement of international trade rules: Process and hypotheses
The multilateral trade regime evolved from a typical case of intergovernmental international
cooperation where states retain near-full control over decisions to an institution where
enforcement powers are partially delegated to third party bodies (Poletti and De Bièvre, 2016).
To reap the benefits of the increased judicialization of the global trade regime, the Directorate-
General (DG) for Trade of the EC engaged in a process of institutional reform through the
adoption of the MAS in 1996. This strategy was aimed at easing the path to enforcing
international trade rules by creating more performant information gathering processes and
market access investigation procedures. The reform culminated in the creation of the Market
Access Unit, the Trade Barriers Regulation Unit, and the WTO Division. In this new
4
institutional constellation, the Market Access Unit is entrusted with the task of processing
information stemming from EU businesses about foreign trade barriers, which in turn supports
the Trade Barriers Regulation Unit and the WTO Division in defending the rights of EU
exporters in litigation processes or in negotiations. The most important responsibility of the
Market Access Unit is the management of the MADB, an online-available computer database
on export formalities, WTO bound tariff levels, and existing barriers to trade. Since its creation,
the MADB represents the informational backbone of the EU’s strategy of enforcement of
international trade rules, establishing a systematic and centralized source of information on
market access barriers encountered by EU industry, actionable under multilateral and bilateral
trade law – a capacity until then dispersed in the different services of the Commission.
Disposing of this institutional armour, EU policymakers face two key political choices
when it comes to enforcing international trade rules. First, they have to decide whether to act
upon the many complaints that are brought to their attention by European exporters. As in other
WTO members disposing of formal or informal mechanisms of market access investigations,
EU policymakers act upon information provided by organized domestic exporters, who
represent the key enforcement constituency of international trade rules. However, only a small
fraction of the foreign trade barriers officially reported by European exporters to be (allegedly)
in violation of international trade rules in the MADB prompt action by the EC (Figure 1), which
makes the politics of multilateral trade enforcement akin to that of the US and other political
systems where similar patterns have been observed (Brutger, 2017; Davis, 2012; Yildirim et
al., 2018) and begs the question of what determines this political choice.
[Figure 1]
Second, once EU policymakers have selected among the multiple demands they confront, they
also have to make a decision on the most effective strategy to enforce international trade rules.
The EU, in particular its Market Access Advisory Committee, basically has two options from
which it can choose: negotiations or litigation. While the literature on the enforcement of
international trade rules has so far focused on the drivers of WTO dispute onset (Kim and
Spilker, 2019; Ryu and Stone, 2018; Yildirim et al., 2018), this exclusive focus on legal
enforcement obscures that policymakers can also use other political tools, such as engaging in
various kinds of formal and informal negotiations with its relevant trading partners.
Policymakers do not solely rely on litigation as an enforcement strategy but often do opt for
less formal political negotiations to convince trade partners to remove their trade barriers
(Figure 2), begging the question as to why.
[Figure 2]
We therefore advance a number of hypotheses on the determinants of states’ strategies
regarding the enforcement of international trade rules and distinguish between the decision
whether to take action (stage 1) and the choice of strategy – i.e. the type of action (stage 2).
Stage 1: Choosing to take action
Existing research on the US has illuminated the role of domestic exporting constituencies
demanding that policymakers act to remove foreign trade barriers (Brutger, 2017; Davis, 2012;
Kim and Spilker, 2019; Ryu and Stone, 2018; Yildirim et al., 2018). These political-economy
approaches of trade policy suggest that policymakers’ choice to act to remove foreign trade
barriers can be viewed as a bottom-up process largely determined by the political influence of
domestic exporting constituencies. According to this perspective, policymakers’ decision to
5
act on behalf of these exporters’ demands, either via legal or political means, can be viewed as
a signalling device to demonstrate their resolve to defend the interests of domestic exporting
constituencies that can provide a number of crucial resources that are instrumental to increasing
their chances of being re-elected or re-appointed.
Drawing on this body of work focusing on the US, we conceive of EU policymakers’
choice to act to remove foreign trade barriers as a function of the political influence of the
relevant domestic constituencies demanding the enforcement of international trade rules. This
choice does pose some empirical challenges. For one, grasping the political influence of trade-
related constituencies in the US context may have proven relatively straightforward due to the
public availability of data on their financial contributions (Davis, 2012; Ryu and Stone, 2018;
Yildirim et al., 2018). However, such data are not available in the EU and, moreover, the
connection between interest group resources and political influence is much less
straightforward in the EU system of interest representation (Stevens and De Bruycker, 2020).
We therefore rely on alternative empirical strategies drawing on a number of arguments that
allow us to derive expectations about the likely sources of exporters’ influence on EU trade
policy making.
First, the scope for collective action has been approximated by the extent to which an
industry is dominated by a few large companies. Scholars working with trade models that
account for firm heterogeneity show that there are large intra-industry differences in export
ability (Melitz, 2003). In some industries, only a few large and competitive firms are capable
of selling their products abroad. Echoing Olson (1965), producers in such ‘oligopolistic
industries’ need to coordinate their actions with fewer firms, which suggests that they can more
easily overcome the collective action problem. For instance, the multinational aerospace
corporation Airbus – by far the largest manufacturer and exporter of aircraft in Europe – has
no chance to free ride on the lobbying efforts of other companies and can be expected to
autonomously develop a strategy to maximise pressure on relevant political actors. In contrast,
when a sector consists of many small and medium-sized enterprises (SMEs), such as the textile
and clothing sector in the EU, the impact of one company will be negligible, so companies will
be less inclined to contribute to their collective good. In other words, when sales are skewed
towards a small number of successful companies, or ‘superstar exporters’, the potential for
cooperation is relatively high (Osgood et al., 2017).
H1: EU policymakers are more likely to respond to demands to enforce international
trade rules when these come from sectors characterized by industrial concentration.
Second, industrial clusters may well reap the economic and political benefits of
agglomeration (Busch and Reinhardt, 1999; Krugman, 1991). Examples include clusters, such
as Silicon Valley as the world’s high-tech centre in the US and the ARRRA cluster (Antwerp-
Rotterdam-Rhine-Ruhr Area) as the largest petrochemical hub in Europe. Such concentration
of industrial activity lowers communication and transportation costs, which may make it easier
to establish close links between companies within the region, while physical proximity may
enable face-to-face interaction among businesspeople and create opportunities for information
exchange, political knowledge-sharing and the development of a group identity - all related to
increases in the level of cooperation (Ostrom, 1998). Moreover, denser social networks may
allow industry members to observe each individual’s contribution toward the collective effort
and it is generally understood that low monitoring costs and repeated interaction significantly
reduce the chances of defection (Axelrod, 1981). As argued by Busch and Reinhardt (1999),
geographically concentrated sectors should be better able to articulate common preferences, to
lobby government officials and to influence trade policy decisions than dispersed sectors.
Evidence from the US confirms that close-packed industries, such as manufactures of oilfield
6
equipment, are much more politically mobilised and are relatively successful in gaining trade
protection (Busch and Reinhardt, 1999; McGillivray, 1997). In Europe, a key collective actor
from the ARRRA cluster, the German federation of the chemical industry VCI, has repeatedly
stated that the WTO ‘dispute settlement system is a successful instrument and should be used
more intensively by the EU to tackle non-WTO compliant practices’ and that the ‘Market
Access Strategy is a good framework for tackling non-tariff trade barriers in countries outside
Europe’ (VCI, 2010).
H2: EU policymakers are more likely to respond to demands to enforce international
trade rules when these come from geographically concentrated sectors.
A third line of research suggests that a broad representation of sectoral interests across
member states is needed to engender sufficient political support (Pincus, 1975), which would
run counter to the logic of geographical concentration. However, the key finding of Busch and
Reinhardt (1999) for the US is that geographically concentrated but politically dispersed
sectors are most likely to see their demands fulfilled. When sectoral constituents that share a
common economic interest are spread out over multiple member states, their issues can become
salient and relevant in the minds of several political leaders. In other words, politically
dispersed sectors enjoy greater ‘political clout’ (Rogowski et al., 1999). In the EU, all member
states are represented in the main trade policy advisory committees of the EC where they
submit opinions on the measures to be taken.
1
If foreign trade restrictions cause strain on
businesses that are active in many member states, such as manufacturers of food products, then
this might put strong pressure on the EC to take action.
H3: EU policymakers are more likely to respond to demands to enforce international
trade rules when these come from politically dispersed sectors.
Stage 2: Choosing a strategy
Once the decision is made to follow up on a business complaint, the EC needs to determine the
most effective strategy for addressing the trade barrier. This process involves a further
investigation into the nature of the trade barrier and broadly entails a choice between the legal
route, i.e. initiating formal dispute settlement procedures, and the negotiation route, i.e.
beginning formal or informal negotiations with the relevant trading partners. We purposefully
dichotomize this choice into one between negotiations and litigation although the latter is a
process that unfolds in several discrete steps. When one or more WTO members file a formal
complaint on specific trade policy measures taken by another member, the process begins with
consultations taking place as confidential negotiations between the parties, and only when the
two parties fail to reach a political compromise at this stage then the proper legal process begins
with the establishment of a panel of experts, the issuing of a ruling and, potentially (and until
recently), the appeal phase where the standing Appellate Body (AB) further reviews the
dispute. Hence, WTO litigation itself does include a phase, i.e. consultations, in which parties
can negotiate diplomatically. Understanding why certain WTO disputes escalate from
consultations to panel stage is important and has received due consideration in the literature
(see Guzman and Simmons 2002). Yet, for our purposes here, we treat litigation as one
category for two reasons. First, the ‘shadow of WTO law’ generates systematic incentives for
disputants to negotiate differently during consultations than in purely inter-governmental
negotiations (Busch and Reinhardt 2001; Poletti et al. 2015). Second, the decision to initiate a
dispute brings about significant financial and administrative burdens, and potential negative
diplomatic externalities irrespective of whether it escalates or not (Davis 2012). In short, the
7
strategic incentives that actors face during the consultation phase of litigation are more similar
to those of subsequent phases of litigation than to those characterizing pure inter-state political
negotiations outside it.
Existing studies suggest the choice of strategy can depend on the power relationship
between the complainant and the defendant. For instance, Sattler and Bernauer (2011) argue
that countries are more likely to engage in bilateral negotiations when there are large economic
power asymmetries. The implicit assumption underpinning this argument is that, whenever
possible, countries prefer obtaining the removal of foreign trade barriers via negotiations rather
than via WTO litigation. We deem such an assumption plausible for three reasons. For one, as
argued above, initiating a WTO dispute entails administrative and financial burdens that are
significant even for WTO members least constrained by legal knowledge and resources
(Brutger, 2017) and potential negative diplomatic externalities (Odell 2000). In addition,
litigation is also characterized by a high level of uncertainty, both because of its slow timeline
and because of the intrinsic difficulty of anticipating the likelihood of a successful ruling (Davis
2012). Moreover, potential complainants can anticipate that, by increasing visibility of the
contested issue, initiating a WTO dispute can increase trade partners’ audience costs and
therefore make it more difficult to obtain the desired concessions (Poletti et al., 2015).
The probability of obtaining a favourable outcome in negotiations is higher when power
asymmetries are high. On the one hand, powerful states will try to obtain concessions through
negotiations by effectively threatening to impose great costs on less powerful states. Small
economies are generally more vulnerable to market closure of the EU than vice versa, which
gives these countries a relatively weak bargaining position. The EU might exploit its position
as the world’s largest trading bloc by threatening with countermeasures – like tariffs or
regulatory restrictions – if the foreign trade barriers remain in place. In turn, small economies
facing (potential) retaliation or reprisals may succumb to European pressure by opting for an
early ‘out-of-court’ settlement. On the other hand, economic coercion will be less effective
when power differentials are small. The leverage of the EU to induce a concession from the
US or the Chinese government will be substantially lower than vis-à-vis small countries such
as Armenia or Ecuador. When the EU is not able to exert enough influence to achieve a
favourable outcome in bilateral negotiations, the EC might be more willing to take the costly
hurdle of filing a dispute at the WTO. This leads us to formulate our fourth hypothesis:
H4: EU policymakers are more likely to choose litigation rather than negotiations
when economic power asymmetry is small.
The model
To test our hypotheses, we rely on an original EU dataset of foreign trade barriers encountered
by EU exporters. We examine foreign trade barriers reported by EU businesses in the period
between 1995 and 2018. We do not consider barriers that are identified more recently because
the EC needs a reasonable amount of time to determine an appropriate strategy. Our dataset
includes 520 observations of trade issues with 61 partner countries reported by 13 European
sectors. This section sets out the model we use to explain the likelihood of the EU action and,
subsequently, the choice of strategy in removing foreign trade barriers. In addition to our
variables of interest, we include a set of control variables typically incorporated in models of
trade disputes.
Dependent variable
8
For the construction of our dependent variable, we draw upon the recorded business complaints
in the EC’s MADB. While we cannot exclude that some businesses refrain from filing
complaints in the first place due to limited organizational resources or expectations that their
lobbying effort will not trigger any reaction by EU policymakers, the MADB offers the most
comprehensive existing source of information to get a grasp of the universe of potential cases
EU policymakers can select upon when deciding whether and how to enforce international
trade rules. The MADB gives information on the type of restrictive measures, the country in
which they were encountered and the product category. These restrictive trade policies range
from simple tariff changes, administrative burdens, protectionist measures through non-tariff
policies such as standards and technical regulations to the suspension of imports. What is
crucial for our analysis, however, is that the MADB also provides a description of what the EU
is doing to remove each barrier. We use these descriptions, in combination with other sources,
to code our dependent variables.
First of all, the EU has several instruments to address market access barriers. Hoekman
et al. (2017) provide a list of possible actions the EU could take to resolve a trade issue: (1)
send letters of request to third party authorities, (2) organise discussions with foreign officials,
(3) table the matter in a special committee, (4) send diplomatic missions, (5) offer technical or
financial assistance, (6) address the issue in ongoing trade negotiations, (7) raise the matter in
a WTO committee or (8) in a WTO accession process, (9) resort to dispute settlement within
the context of a bilateral trade agreement or (10) initiate a WTO case. For each trade barrier in
our dataset, we scrutinize whether the EU has used (at least) one of these instruments in an
attempt to resolve the issue. When the Commission states, for example, that it has ‘frequently
raised its concerns in bilateral meetings’,
2
our dependent variable takes the value of 1 in the
first stage of our analysis. If the MADB makes no reference to any of the actions identified,
the dummy is assigned a 0 value.
In a next step, we distinguish between two broad strategies the EU uses to remove trade
barriers. First, the category ‘negotiation’ consists of actions in which the EU addresses
demands and makes proposals to the trading partner with the purpose of informally reaching
an agreement. From the above list of actions, (1) to (8) can all be considered as attempts to find
a negotiable solution, although in case of (5), ‘technical or financial assistance’, the problem
to be solved might also be practical rather than political. This instrument, however, is not used
in our estimation sample. Second, the category ‘litigation’ consists of instances in which the
EU decides to follow the judicial route to remove a trade barrier. Instruments (9) and (10),
initiations of legal disputes, fall under this category.
3
If the EU initiates litigation, our
dependent variable takes the value of 1 in the second stage of our analysis, and 0 otherwise.
We primarily draw on the MADB to observe whether the EU has filed a legal case against
another country, but we always cross-check this information with data from the EC’s and the
WTO’s websites (see the Online appendix).
Independent variables
In the first stage of our analysis, we focus on the link between several types of concentration
and the probability of EU action. To explain the choice of strategy, we direct our attention to
economic power asymmetry. We describe the measurement of our variables of interest in this
section.
The measure for industrial concentration (INDCON) represents the proportion of export
sales (r) in a sector (i ∈{1,2…n}) accounted for by the largest producers:
𝐼𝑁𝐷𝐶𝑂𝑁!=
∑
𝑟!"##
∑
𝑟!##
*
1−
∑
𝑐!"##
∑
𝑐!##
.
,00000000000000000000000000000000000000000000000000000000000000000000000000000000000000(1)
0
9
where the largest producers (p) are the ones with more than 250 employees. INDCON will be
relatively high when the large producers constitute only a small fraction of the total producer
population in a given EU industry (ci). Because there are no EU-level data on export sales by
enterprise size, we aggregate the available member state (h) level data to construct INDCON.
4
Our hypothesis holds that the probability of EU action increases when exports sales are skewed
towards a few successful producers. In other words, the coefficient of INDCON should have a
positive sign.
We quantify the geographical concentration (GEOCON) of a sector with a decreasing
function of distance in kilometres (djk) between EU regions (j, k ∈{1,2…m}) in which
companies (cij) are active. As recommended by White (1983), we use the inverse of distance
to make sure that extreme distances do not receive undue weight, which makes it a measure of
concentration rather than dispersion (see also Busch and Reinhardt, 1999). We multiply the
inverse distance between regional pairs with the share of companies (of the total sector) that
are active in the respective regions, such that:
𝑋!$% =𝑐!$ +𝑐!%
𝑐!⋅𝑑$% 0000𝑤ℎ𝑒𝑟𝑒00𝑐!=0
:
𝑐!$
&
$'( 00000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000(2)0
To measure GEOCON, we take the sum of all the regional pairs Xijk and divide by a degree of
inequality in the distribution of firms across regions. More specifically, we calculate the ratio
of firms within each region (cij / ci) and then sort the regions based on the resulting values in
ascending order. We use a continuous cumulation of this ratio (qij) – i.e. the cumulative
proportion of firms in the respective regions – to derive the ‘concentration area’ (ai) in the
following way:
5
𝑎!=1
2−1
2∙
:>
𝑞!$ +𝑞!$)(
@A
1
𝑚
C
&
$'( 0000𝑤ℎ𝑒𝑟𝑒00𝑞!$ =𝑐𝑢𝑚𝑢𝑙𝑎𝑡𝑖𝑣𝑒0𝑝𝑟𝑜𝑝𝑜𝑟𝑡𝑖𝑜𝑛0𝑜𝑓0𝑓𝑖𝑟𝑚𝑠0000(3)0
𝐺𝐸𝑂𝐶𝑂𝑁!=
∑ ∑
𝑋!$%
&
$'%*(
&)(
%'( 1−𝑎!000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000(4)
0
By means of the denominator, we give additional weight to industries that are concentrated in
fewer EU areas. Our measure of GEOCON has a number of advantages compared to other
estimates of proximity. First, this measure explicitly captures the spatial relationship between
geographical units, in contrast to measures such as the (alternative) Gini coefficient (Krugman,
1991) and the Ellison-Glaeser index (Ellison and Glaeser, 1997), which is important to test our
theoretical argument about the possibility of face-to-face communication in industrial clusters.
Second, our measure attaches an especially high weight to industries that are geographically
concentrated (numerator) in relatively few regions (denominator), thereby eliminating the
problematic element in other models that cannot appropriately discriminate between dispersion
and concentration in a couple of distant places (e.g. Busch and Reinhardt, 1999). Third,
GEOCON is a relatively simple and intuitive measure to calculate geographical concentration
in the EU using available NUTS2-level data.
6
The data, retrieved from Eurostat, provide
information on the number of enterprises (or parts thereof) in 272 EU regions on a
disaggregated NACE industry-level.
7
We use the same source for regional data on agricultural
10
sectors. This allows us to analyse the patterns of geographical concentration in a more precise
manner than scholars that have used EU member state level data (e.g. Brülhart, 2001).
To measure political concentration (POLCON), we calculate a Herfindahl-Hirschman
index for each NACE industry, based on employment statistics from EU member states.
𝑃𝑂𝐿𝐶𝑂𝑁!=
:T
𝑙!#
:
𝑙!#
#
U V
+,000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000(5)
#
0
where lih is the number of employees in sector i in member state h. An increase in the
Herfindahl-Hirschman index indicates that a larger share of the sector’s workforce is
concentrated in fewer EU member states. In contrast to GEOCON, we expect this variable to
be negatively related to the probability of EU action.
All sector-level statistics come from Eurostat (see the Online appendix). The NACE
data covers the 2008-2016 period, whereas the agricultural data is only available for the years
2005, 2007, 2010, 2013 and 2016. Linear interpolation between two known points is used to
increase the temporal coverage of agricultural data. In other years, if data were not available,
we used the closest available data.
We measure economic power asymmetry (ECOASY) as the difference between the (log)
gross domestic product (GDP) of the EU and the (log) GDP of the partner country that has
introduced the trade barrier (Guzman and Simmons, 2005; Sattler and Bernauer, 2011). A
positive value shows that the EU has more economic power than its trading partner – a negative
value means the opposite. A value of zero indicates that the states are equally powerful. The
data are gathered from the World Development Indicators.
Control variables
We also add a set of control variables to the analysis. In stage 1, we account for the fact that
there has been a rise in potential disputes after the financial crisis (Hoekman et al., 2017), which
may have forced the Commission to be more selective in addressing trade barriers. For this
reason, we introduce a dummy variable (P-CRISIS) that takes the value of 1 for the trade
barriers that are reported since 2008. We also add a dummy for trade barriers in the agricultural
sector (AGRICU) as trade in agricultural products is often regarded as particularly sensitive in
Europe (see Davis, 2003). Moreover, we include a dummy variable for WTO membership of
the partner country, because the WTO offers several formal and informal structures
(committees, dispute mechanism) to address trade issues, which may increase the probability
of EU action.
In stage 2, we consider the trade balance (TBALANCE) between the EU and its trading
partner as an alternative dimension of trade power (calculated as the log of exports minus the
log of imports). A sizable bilateral trade deficit can increase the capacity to retaliate and offers
credible means to impose economic costs on the trading partner by raising import barriers.
When the potential for retaliation threats is high, a more favourable outcome might be achieved
in negotiations rather than via litigation. Second, we control for the political system of the
partner country (DEMOCR) using the Polity IV index. Allee (2004) finds that democratic
countries are more likely to resolve their disputes in a legal manner. Third, the legal capacity
of the trading partner, as proxied by GDP per capita (GDPCAP), is added to the model
(Guzman and Simmons, 2005). We include this variable because the EU might be more
reluctant to initiate formal disputes against countries that have strong institutional, financial
and human resources to (successfully) defend their case. Fourth, we add a dummy for the
existence of a preferential trade agreement (PTA) between the EU and the trading partner as
PTAs are often concluded between countries with strong diplomatic ties (Plouffe and Van der
11
Sterren, 2016). This may in turn increase the probability that issues are resolved politically
rather than legally.
In both stages, we also add the control variables of the other stage to the analysis. The
descriptive statistics on all of the variables are given in the regression tables. An overview of
the data sources is provided in the Online appendix.
Results
In this section, we present an analysis of our two-stage model of EU enforcement policy. The
first part considers different ‘concentration hypotheses’ to explain the probability of EU action,
and in the second part, we test whether the EU’s choice of strategy to remove trade barriers is
in line with our ‘trade power hypothesis’.
Action or no action
Table 1 sets out the results of the probit regression estimating the probability of EU action. The
pseudo R2 and the percentage of correctly predicted values are reported at the bottom of the
table. In Model I, we include our variables of interest as well as the controls that are likely to
be associated with EU Action. Models II and III examine the same relationships along with the
control variables that are considered to be particularly important in the second stage of our
analysis, but which, nonetheless, might (also) play a role in the Commission’s decision to
follow up on a business complaint in the first place. As some variables – DEMOCR and
GDPCAP – contain missing data on the level of the trading partner, we include these variables
only in Model III.
8
[Table 1]
Turning to the regression models, we note that the coefficients on many of the variables have
the signs we expected based on the existing literature. Our primary interest is in the values of
the parameters representing the different concentration ratios. First of all, we find support for
the hypothesis that lobbying power is positively related to industrial concentration. In Models
I and II, the coefficient of INDCON is positive and statistically significant at the 5% level. The
observed effect, however, is slightly lower when more controls are added to the specification.
Still, the finding is quite striking in substantive terms: on average, a 10 percentage points
increase in INDCON is associated with a 5.9 percentage points increase in the probability of
EU action (in Model III).
9
Second, the coefficients of GEOCON do not lend support for the
argument that geographically concentrated sectors are more successful in gaining the support
of the Commission to remove foreign trade barriers. Third, the table also reveals that POLCON
does not have a statistically significant effect on the probability of EU action. This means that
politically dispersed industries do not seem to have a stronger voice in European trade politics.
Looking at the control variables, we observe that exporter complaints filed after the
financial crisis have a lower probability of being acted upon by the Commission than cases that
are registered pre-2008. Moreover, we find that agricultural interests tend to be well-
represented in EU enforcement policy for reasons that are not fully captured by other
explanatory variables in the model. However, this could still be well understandable: although
the production of agricultural commodities is relatively dispersed among many farmers, the
next step in the supply chain is characterised by far higher degrees of concentration in the form
of concentrated agribusinesses. Furthermore, the probability of EU action is relatively high
when the trading partner is a member of the WTO or a signatory of a bilateral treaty with the
EU. This implies that the Commission uses the formal and informal structures (committees,
12
dialogues, dispute mechanisms) of these agreements to address trade issues. Lastly, the partner
country’s GDP per capita is positively related to the probability of EU action.
In sum, from our three variables of interest, only INDCON is consistently found to
increase the likelihood of EU action. This result suggests that industries dominated by a few
large companies (e.g. manufacturers of motor vehicles and chemicals, see the Online appendix)
are better able to form a powerful lobby that can influence EU trade policy.
10
Contrary to our
expectations, geographical and political concentration do not seem to have an impact on EU
action.
Litigation or negotiation
Table 2 shows regression results concerning the EU’s choice of strategy – i.e. litigation or
negotiation. The number of observations is lower at this stage, because all cases in which the
Commission undertook ‘no action’ are removed from the estimation sample. In addition, all
trading partners in the remaining sample are members of the WTO, so this variable (WTO) is
excluded from the estimation models. The dependent variable takes the value of 1 if the EU
has filed a legal case against the trading partner to resolve the trade barrier (L = 1), and 0
otherwise.
[Table 2]
To focus on our variable of interest, we find that economic power asymmetry (ECOASY)
between the EU and the trading partner is negatively related to the choice to start legal
proceedings. In other words, litigation becomes less likely when power differences between
the EU and the trading partner are large. This result is consistent with the idea that powerful
states will try to obtain concessions through negotiations by effectively threatening to impose
great costs on smaller economies (Sattler and Bernauer, 2011).
Next, we note that GDPCAP, our proxy for legal capacity, is indeed negatively related
to the prospect of litigation (Guzman and Simmons, 2005). The coefficient of DEMOCR
confirms Allee’s (2004) finding that democratic countries tend to resolve their disputes through
legal institutions. Moreover, the EU is more likely to negotiate rather than litigate about
agricultural trade barriers, which corresponds to earlier findings by Busch and Reinhardt
(2003). They observe that agricultural cases, relative to other disputes, are often settled
bilaterally without going through the adjudication process of the GATT/WTO. This suggests
that EU policymakers anticipate the greater escalation risk for agricultural disputes (e.g. the
defendant does not accept the ruling) and, therefore, decide to resolve these issues out of court
(see Sattler et al., 2014).
To sum up, our results lend support to the hypothesis that the EU is more likely to
negotiate rather than litigate when economic power asymmetry between the EU and the trading
partner is large. As anticipated, we show that equal powers, such as the EU and the US, will
sooner resort to a legal dispute settlement mechanism, probably because neither side has the
leverage to induce a concession from the other.
Robustness
As a robustness check, we also evaluate our hypotheses using alternative estimators that take
potential selection effects into account. If the decision to take action (stage 1) is influenced by
the choice of strategy (stage 2), or vice versa, the error terms in both stages may be correlated,
which in turn can lead to biased estimates. To control for this possibility, we employ a probit
13
version of the Heckman (1976) selection model as well as Leemann’s (2014) strategic selection
estimator.
Table 3 reports the results of the Heckman (1976) probit model with sample selection.
In this two-step model, the corrected version of the second stage equation includes an
additional regressor – the inverse Mills ratio – that is estimated from the initial selection
process. In the first column, we show the regression results based on a parsimonious set of
variables that maximises the number of observations. We find a negative relationship between
economic power asymmetry and the initiation of formal dispute settlement procedures, which
is consistent with our previous findings. The coefficient on ECOASY is statistically significant
at the 1% level. In columns 2 and 3, we add our sectoral variables and we also check whether
our results hold when using an alternative measure for the distribution of economic resources
among EU businesses (industry value added, INDVAD). Our main findings remain unchanged:
industrial concentration, in terms of size and resources, increases the probability of EU action
while economic power asymmetry affects the choice of strategy. Moreover, we note that the
null hypothesis of the Wald test of independent equations cannot be rejected (errors are not
correlated), which means that both equations could also be estimated using two separate probit
models.
[Table 3]
Table 4 presents the results of Leemann’s (2014) strategic selection estimator. In contrast to
the Heckman (1976) model, this estimator accounts for strategic behaviour (anticipation) in the
first stage, meaning that the Commission weighs the expected probability of using a particular
strategy to resolve a trade issue when deciding whether to take action. There are three possible
outcomes in this model: No Action (NA), Negotiation (N) or Litigation (L). The overlap
between the sets of variables for both stages should not be too large and, therefore, the first-
stage equation does not include all (second-stage) control variables. The coefficients on the
explanatory variables show the direct effects on the utility for the Commission from pursuing
a certain outcome. When looking at the second-stage outcomes, we see that, as expected, the
EU is more likely to litigate when economic power asymmetry is small. The utility of
negotiation, however, does not appear to increase when economic power differences become
larger. This suggests that the choice to start negotiations may actually be driven by an indirect
effect – i.e. it follows from the decision not to litigate. Turning to the first-stage results, the
negative coefficient on INDCON indicates that industrial concentration reduces the EU’s utility
of No Action, which is in line with our expectations. Yet, the coefficient is only statistically
significant at the 10% level. This diminished effect of INDCON – compared to previous
estimates – suggests that strategic anticipation plays a role in the decision to act upon exporters’
complaints. This adds an important nuance to the understanding of the representation of
concentrated interests in EU trade policy: the Commission is more likely to respond to demands
coming from large exporters, not only due to lobbying strength, but also in anticipation of the
next stage in which the EU may need the (financial) support of these rich and powerful firms
to resolve a trade issue.
[Table 4]
Conclusion
Our overarching aim was to uncover the political-economic factors that drive the EU’s
enforcement of international trade rules, investigating both when and how it acts upon
exporters’ demands to remove foreign trade barriers. Our analysis supports two broad sets of
14
arguments. First, as in other political systems, EU policymakers are very likely to act as
enforcers of international trade rules in response to demands emanating from sectors
characterized by industrial concentration. By contrast, our findings do not lend support to the
argument that the interests of geographically concentrated and politically dispersed industries
are well-represented in EU trade policy. These findings have important implications for our
understanding of the politics of trade in the EU. In other areas of trade policymaking such as
trade negotiations the autonomy of the EC is severely constrained by formal rules providing
veto power to a wide array of actors - the European Parliament, the member states, and even
national parliaments – which increase the odds that political preferences that are dispersed
across sectors and member states will weigh in the policy process. Our findings suggest that,
on the contrary, when the EC enjoys a high degree of autonomy, EU trade policymaking may
end up being more systematically biased in favour of concentrated business interests.
Second, we find support for the hypothesis that the EU chooses different enforcement
strategies depending on the economic power of its trading partners, privileging political
negotiations with economically weaker states, while preferring to use litigation with stronger
trading partners. The way in which EU policy makers respond to industry demands petitioning
the enforcement of international trade rules thus seems to crucially depend on the
characteristics of its trading partners.
While this study has a much broader empirical scope than studies that have focused
only on actual trade disputes, the analysis of recorded business complaints still has its
limitations. Even though there might be no obvious biases apparent, there can be instances in
which exporters choose not to report trade barriers to the publicly accessible MADB, either
because they are reluctant to reveal their problems to a wider audience or because they deem
it unlikely that they will get the attention of EU policymakers. We cannot rule out that our
dataset suffers from selection bias, yet we believe our empirical strategy allows us to obtain a
comprehensive view of the universe of potential cases EU policymakers can select from when
deciding whether and how to enforce international trade rules.
Especially now that the multilateral trading system is under heavy strain, it will remain
key to conduct further research on the conditions under which states do, or do not, choose to
mutually enforce and re-enforce their trade commitments. By showing that EU policymakers
tend to privilege concentrated interests when it comes to deciding whether to uphold
international trade rules, our work underscores that the political-economy underpinning
international trade enforcement in the EU displays important similarities with respect to that of
the US and other political systems. While there are little doubts that the EU is a peculiar
political system, this means that EU trade policy research had better not cast itself away from
broader political-economy debates as it will benefit from applying general theories of trade
policy.
At the same time, our article also underscores there is ample room for future research
to exploit the potential of comparative designs that can cast light on whether, and how,
differences across political systems produce systematic differences in enforcing and upholding
international trade rules. For one, as highlighted by our null finding concerning the political
concentration hypothesis, the electoral incentives for collective action on trade policy may well
be quite different. Single-member districts in the US may encourage cooperation across voting
districts and parallel lobbying, whereas the EU’s proportional representation system prevalent
at the national level may do so to a far lesser extent. Second, with regard to the actual use of
the WTO as an enforcement institution, the outgoing Trump administration has engaged in a
relentless campaign of obstruction aimed at paralyzing the WTO and its dispute settlement
body by dismembering its appeals instance. Also, the possible further rise of go-it-alone trade
policies in a post-coronavirus-crisis world economy might pose enormous challenges to the
active maintenance of that open world trade order. The EU, however, despite the rise of
15
populist leaders and parties advocating protectionist policy platforms in numerous member
states, has so far proven immune to this anti-WTO rhetoric. All in all, the EU remains firmly
committed to the multilateral trading system and its mechanisms of resolution of trade disputes,
suggesting that something in the EU’s political system has so far contributed to taming the anti-
globalization popular backlash’s potential to affect trade policymaking.
Acknowledgements
For many helpful comments and suggestions, we thank Gabriele Spilker, Peter Bursens, Lucas
Leemann, as well as three anonymous reviewers and the editors. A special thanks goes to Aydin
Yildirim for sharing his expertise during the early stages of this project.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship,
and/or publication of this article: This work was supported by the Research Foundation
Flanders (Fonds Wetenschappelijk Onderzoek Vlaanderen, FWO) [grant number G057018N]
as well as the Special Research Fund (BOF) of the University of Antwerp.
References
Allee, T. (2004). Legal Incentives and Domestic Rewards: The Selection of Trade Disputes for
GATT/WTO Dispute Resolution. Manuscript. University of Illinois.
Axelrod, R. (1981). The Emergence of Cooperation among Egoists. The American Political
Science Review 75(2): 306-318.
Brülhart, M. (2001). Evolving geographical concentration of European manufacturing
industries. Weltwirtschaftliches Archiv 137(2): 215-243.
Brutger, R. (2017) Litigation for sale: private firms and WTO dispute escalation’, Unpublished
manuscript. Available at:
https://ipespeakerseries.mit.edu/sites/default/files/images/Litigation_For_Sale_9-14-
17_ (accessed 03-07-2019).
Busch, M.L., Reinhardt, E. (1999). Industrial Location and Protection: The Political and
Economic Geography of U.S. Nontariff Barriers. American Journal of Political Science,
43(4): 1028-1050.
Busch M.L and Reinhardt E. (2001). Bargaining in the Shadow of the Law: Early Settlement
in GATT/WTO Disputes. Fordham International Law Journal 24: 58-172.
Busch, M.L., Reinhardt, E. (2003). Transatlantic Trade Conflict and GATT/WTO Dispute
Settlement. In: Transatlantic Economic Disputes: The EU, the US, and the WTO, edited
by Petersmann, E., Pollock, M.A. Oxford: Oxford University Press.
Davis, C.L. (2003). Food Fights Over Free Trade: How International Institutions Promote
Agricultural Trade Liberalization. Princeton, NJ: Princeton University Press.
Davis C.L. (2012), Why Adjudicate: Enforcing Trade Rules in the WTO. Princeton, NJ:
Princeton University Press.
De Bièvre, D., Poletti, A. (2013). The EU in EU trade policy. From regime shaper to status quo
power. In: G. Falkner and P. Müller (Ed.), EU Policies in a Global Perspective. London,
Routledge: 20-37.
16
Dür, A., Eckhardt, J. and Poletti, A. (2020). Global value chains, the anti-globalization
backlash, and EU trade policy: a research agenda. Journal of European Public
Policy 27(6): 944-956.
Ellison, G., Glaeser, E.L. (1997). Geographic Concentration in U.S. Manufacturing Industries:
A Dartboard Approach. Journal of Political Economy 105(5): 889-927.
Guzman, A., Simmons, B.A. (2002). To settle or empanel? An empirical analysis of litigation
and settlement at the World Trade Organization. The Journal of Legal Studies, 31(S1),
S205-S235.
Guzman A., Simmons B. (2005). Power plays and capacity constraints: the selection of
defendants in WTO disputes. Journal of Legal Studies 34(2): 557–598.
Heckman, J.J. (1976). The common structure of statistical models of truncation, sample
selection and limited dependent variables and a simple estimator for such models.
Annals of Economic and Social Measurement 5(4): 475-492.
Hoekman, B., Fiorini, M., Igliozzi, R., Ralaison, N., Yildirim, A. B. (2017). The EU’s Market
Access Stategy: does it reach its main goals? European Parliament Think Tank, 13-12-
2017.
Kim, S.Y., Spilker, G. (2019). Global value chains and the political economy of WTO disputes.
The Review of International Organizations 14(2): 239-260.
Krugman, P. (1991). Geography and Trade. Cambridge: MIT Press.
Leemann, L. (2014). Strategy and sample selection: A strategic selection estimator. Political
Analysis 22 (3): 374-397.
Melitz, M.J. (2003). The impact of trade on intra‐industry reallocations and aggregate industry
productivity. Econometrica, 71(6): 1695-1725.
McGillivray, F. (1997). Party discipline as a determinant of the endogenous formation of
tariffs. American Journal of Political Science 41(2): 584-607.
Odell, J.S. (2000). Negotiating The World Economy. Ithaca: Cornell University Press.
Olson, M. (1965). The Logic of Collective Action: Public Goods and the Theory of Groups.
Cambridge, MS: Harvard University Press.
Osgood, I., Tingley, D., Bernauer, T., Kim, I. S., Milner, H. V., Spilker, G. (2017). The
charmed life of superstar exporters: Survey evidence on firms and trade policy. The
Journal of Politics 79(1): 133-152.
Ostrom, E. (1998). A Behavioral Approach to the Rational Choice Theory of Collective Action.
American Political Science Review 92(1): 1-22.
Pincus, J.J. (1975). Pressure Groups and the Pattern of Tariffs. Journal of Political Economy
83(4): 757-778.
Plouffe, M., & van der Sterren, R. (2016). Trading representation: Diplomacy’s influence on
preferential trade agreements. The British Journal of Politics and International
Relations, 18(4): 889-911.
Poletti, A., De Bièvre, D., Chatagnier, T. (2015). Cooperation in the shadow of WTO law: why
litigate when you can negotiate. World Trade Review 14(S1): 33-58.
Poletti, A., De Bièvre, D. (2016). Judicial Politics and International Cooperation: From
Disputes to Deal-Making at the World Trade Organization. London: ECPR Press.
Poletti, A., De Bièvre, D., Hanegraaff, M. (2016). WTO judicial politics and EU trade policy:
business associations as vessels of special interest? The British Journal of Politics and
International Relations 18(1): 196-215.
Rogowski, R., Kayser, M., Kotin, D. (1999). How Geographical Concentration Affects
Industrial Influence: Evidence from U.S. Data. Paper prepared for delivery at the 1999
Annual Meeting of the American Political Science Association, 2-5 September 1999.
Ryu, J., Stone, R.W. (2018). Plaintiffs by proxy: A firm-level approach to WTO dispute
resolution. The Review of International Organizations 13(2): 273-308.
17
Sattler, T., Bernauer, T. (2011). Gravitation or discrimination? Determinants of litigation in the
World Trade Organization. European Journal of Political Research 50(2): 143-167.!
Sattler, T., Spilker, G., & Bernauer, T. (2014). Does WTO dispute settlement enforce or
inform? British Journal of Political Science, 44(4): 877-902.
Stevens, F. and I. De Bruycker (2020). "Influence, affluence and media salience: Economic
resources and lobbying influence in the European Union." European Union Politics
21(4): 728-750.
VCI (2010), Trade Policy Guidelines of the German Chemical Industry, Position Paper, Verein
der Chemischen Industrie, November 2010.
White, M.J. (1983). The Measurement of Spatial Segregation. American Journal of Sociology
88(5): 1008-1018.
Yildirim, A.B., Poletti, A., Chatagnier, J.T., & De Bièvre, D. (2018). The globalization of
production and the politics of dispute initiation at the World Trade Organization. Global
Policy 9: 38-48.
Figure 1. Trade barriers reported to the Commission, No Action or Action (1995-2018).
Figure 2. Trade barrier removal actions by the Commission, Negotiation or Litigation (1995-2018).
Table 1. Probit regression: Stage 1
Model I
Model II
Model III
Mean ± SD
Min–Max
Action or No Action, Prob (A=1)
0.329±0.471
0–1
INDCON
2.187**
(1.040)
2.104**
(1.063)
1.844*
(1.090)
0.524±0.207
0.220–0.793
GEOCON
3.248
(5.941)
3.783
(5.804)
3.631
(5.675)
0.353±0.030
0.256–0.498
18
Note: Robust standard errors in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1. INDCON:
Industrial concentration; GEOCON: Geographical concentration; POLCON: Political
concentration; P-CRISIS: Post-crisis. AGRICU: Agricultural sector; WTO: World Trade
Organization; PTA: Preferential trade agreement; TBALANCE: Trade balance; DEMOCR:
Democracy index; GDPCAP: GDP per capita.
Table 2. Probit regression: Stage 2
POLCON
-4.728
(5.488)
-4.038
(5.410)
-4.030
(5.344)
0.118±0.028
0.077–0.176
P-CRISIS
-1.012***
(0.267)
-0.966***
(0.271)
-1.103***
(0.285)
0.892±0.311
0–1
AGRICU
1.203**
(0.483)
1.236**
(0.489)
1.066**
(0.499)
0.350±0.478
0–1
WTO
1.072**
(0.503)
1.172**
(0.462)
0.950*
(0.485)
0.948±0.223
0–1
PTA
0.335**
(0.164)
0.398**
(0.172)
0.416±0.494
0–1
TBALANCE (log)
0.053
(0.295)
-0.167
(0.327)
-0.001 ±0.300
-0.738–1.213
DEMOCR
-0.020
(0.013)
4.143±6.369
-10–10
GDPCAP (log)
0.177**
(0.085)
9.081±1.076
6.090–11.291
_cons
-2.750
-3.274*
-4.248**
-
(1.994)
(1.952)
(2.057)
-
No. of observations
296
293
286
-
% correctly predicted
71.3
71.3
71.0
-
Pseudo-R2
0.085
0.097
0.107
-
Model I
Model II
Model III
Mean ± SD
Min–Max
Litigation or Negotiation, Prob (L=1)
0.189±0.394
0–1
ECOASY (log)
-0.621***
(0.194)
-0.622***
(0.194)
-0.834***
(0.222)
1.352±0.711
-0.061–3.196
TBALANCE (log)
0.650
(0.526)
0.640
(0.537)
1.054
(0.683)
-0.018±0.250
-0.493–1.201
DEMOCR
0.039*
(0.020)
0.040*
(0.021)
0.063**
(0.026)
4.284±6.599
-10–10
GDPCAP (log)
-0.331**
(0.129)
-0.333**
(0.131)
-0.337**
(0.159)
9.247±1.110
6.094–11.045
PTA
-0.569*
(0.298)
-0.573*
(0.299)
-0.129
(0.381)
0.495±0.503
0–1
P-CRISIS
0.044
(0.305)
-0.497
(0.427)
0.779±0.417
0–1
AGRICU
-0.983**
(0.391)
0.453±0.500
0–1
_cons
3.160**
3.149**
3.847**
-
19
Note: Robust standard errors in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1. ECOASY:
Economic power asymmetry; TBALANCE: Trade balance; DEMOCR: Democracy index;
GDPCAP: GDP per capita; PTA: Preferential trade agreement; P-CRISIS: Post-crisis.
AGRICU: Agricultural sector.
Table 3. Heckman probit selection model.
(1.298)
(1.294)
(1.511)
-
No. of observations
134
134
95
-
% correctly predicted
79.9
79.1
86.3
-
Pseudo-R2
0.134
0.135
0.196
-
Model I
Model II
Model III
Second stage: Litigation or Negotiation, Prob (L=1)
ECOASY (log)
-0.620***
(0.194)
-0.893***
(0.248)
-0.895***
(0.268)
TBALANCE (log)
0.659
(0.556)
1.382*
(0.731)
1.293
(0.859)
DEMOCR
0.038*
(0.021)
0.043
(0.027)
0.047*
(0.025)
GDPCAP (log)
-0.330**
(0.130)
-0.375**
(0.153)
-0.380**
(0.152)
PTA
-0.567*
(0.299)
0.035
(0.354)
0.023
(0.367)
_cons
3.203**
(1.406)
3.749**
(1.556)
3.558*
(1.850)
First stage: Action or No Action, Prob (A=1)
INDCON
2.321**
(1.181)
INDVAD (log)
0.678**
(0.297)
GEOCON
3.763
(6.308)
4.913
(6.217)
POLCON
-4.803
(5.972)
-3.499
(5.522)
P-CRISIS
-0.667***
(0.171)
-1.019***
(0.279)
-1.014***
(0.298)
AGRICU
1.234**
(0.491)
0.649**
(0.315)
WTO
0.830**
(0.355)
1.074**
(0.501)
1.092**
(0.499)
_cons
-0.869**
-3.016
-9.716**
(0.389)
(2.120)
(4.277)
No. of observations
517
293
293
Wald c2
21.35
[0.001]
16.13
[0.006]
15.93
[0.007]
Wald test of indep. eqns.
0.01
[0.942]
0.09
[0.764]
0.00
[0.992]
Pseudo-R2 1st stage
0.041
0.083
0.087
20
Note: Robust standard errors in parentheses, p-values in square brackets. *** p < 0.01, ** p <
0.05, * p < 0.1. ECOASY: Economic power asymmetry; TBALANCE: Trade balance;
DEMOCR: Democracy index; GDPCAP: GDP per capita; PTA: Preferential trade agreement;
INDCON: Industrial concentration; INDVAD: Industry value added; GEOCON: Geographical
concentration; POLCON: Political concentration; P-CRISIS: Post-crisis. AGRICU:
Agricultural sector; WTO: World Trade Organization.
Table 4. Strategic selection estimator.
Note: Standard errors in parentheses, p-values in square brackets. *** p < 0.01, ** p < 0.05, *
p < 0.1. ECOASY: Economic power asymmetry; TBALANCE: Trade balance; DEMOCR:
Democracy index; GDPCAP: GDP per capita; PTA: Preferential trade agreement; INDCON:
Industrial concentration; GEOCON: Geographical concentration; POLCON: Political
concentration; P-CRISIS: Post-crisis. AGRICU: Agricultural sector; WTO: World Trade
Organization.
Second stage: Litigation or
Negotiation
First stage: Action
or No Action
UL (Litigation)
UN (Negotiation)
UNA (No Action)
ECOASY (log)
-0.527**
(0.250)
-1.840
(1.406)
TBALANCE (log)
-0.068
(0.289)
-2.207
(2.590)
DEMOCR
0.050*
(0.027)
0.127
(0.097)
GDPCAP (log)
-0.325**
(0.156)
-0.376
(0.448)
PTA
-0.499**
(0.247)
-0.656
(1.939)
INDCON
-2.768*
(1.592)
GEOCON
-6.127
(6.997)
POLCON
7.691
(6.931)
P-CRISIS
0.001
(0.305)
0.455
(1.235)
1.368***
(0.509)
AGRICU
-1.068***
(0.323)
2.819
(1.930)
0.679
(1.612)
WTO
-1.212**
(0.606)
Constant
3.712**
-0.241
1.184
(1.669)
(4.199)
(2.795)
No. of observations
286
Log-Likelihood
-194.16
ρ (errors)
-0.485
[0.240]
21
Notes
1
Trade policy committee and Market access advisory committee.
2
Market Access Database (barrier id 10787).
3
In this category, by far the most frequently used instrument is WTO litigation, whereas the EU has only
sporadically initiated disputes under bilateral trade agreements.
4
In some cases, export sales by enterprise size are confidential, meaning that the measurement is based on a subset
of member states. Therefore, we also use the value added of each industry as an alternative measure (INDVAD)
for the distribution of economic resources among EU businesses.
5
Note that the ‘concentration area’ (ai) is equivalent to the area between the line of equality and the Lorenz curve.
6
Nomenclature of Territorial Unit for Statistics (NUTS).
7
Nomenclature of Economic Activities (NACE).
8
Most of the missing values are from Hong Kong, Iran and Taiwan.
9
These percentages are not reported in Table 1. The marginal effects are calculated for each observation in the
data and then averaged.
10
Our results hold when using an alternative measure for the distribution of economic resources among EU
businesses, INDVAD (see the Online appendix).
