Interest group success in the European
Union: When (and why) does business
lose?
Andreas Dür, University of Salzburg
Patrick Bernhagen, Zeppelin University
David Marshall, University of Salzburg
Paper prepared for presentation at the General Conference of the ECPR in
Bordeaux, September 4-7, 2013
Abstract
Lobbying is widespread in the European Union (EU). As of June 2013, no
fewer than 5,703 interest representatives have signed up to the EU’s transparency register. A large majority of these groups represent business interests.
But how effective is business lobbying in the EU? When does business win and
when does it lose in the context of legislative policy-making in the EU? We argue that business lobbying success in the EU is much more circumscribed than
often expected. In fact, with much legislative activity in the EU being about
greater market regulation and business frequently opposes such regulation,
business tends to be in a defensive position in much of the EU’s legislative
politics. A new dataset on the preferences of more than one thousand interest
groups with respect to 70 legislative acts proposed by the European Commission between 2008 and 2010 allows us to evaluate these expectations. We
find robust support for our argument about limited business influence. This
finding has implications for the literature on legislative decision-making in the
EU and for the debate on the state of democracy in the EU.
Key Words: Interest group influence, lobbying success, business and politics, lobbying, European Union, spatial model, EU legislation.
1
Introduction
Business interests’ attempts at exercising influence are omnipresent in the European
Union (EU).1 Illustratively, when the European Commission started work on antitobacco legislation in 2012, tobacco producers and retailers engaged in an intensive
lobbying campaign. At the height of the campaign, the EU’s health commissioner
John Dalli was forced to resign amidst allegations that he may have known of an
attempt at “selling influence” to a Swedish tobacco producer. Similarly, in late 2012
and early 2013, the EU witnessed a massive lobbying campaign by US companies,
including Facebook and Google, asking for changes to a proposed data protection
directive. More than 3,000 proposals for amendments were forwarded to the European Parliament. Witnessing these lobbying efforts by business in Brussels and
Strasbourg, many observers suggest that business power is rampant in the EU (Burley et al., 2010). But so far systematic studies of the effectiveness of lobbying are
rare. We thus ask: how much power does business really possess in EU legislative decision-making? When and why do policy-makers override business lobbying?
More concretely, when does business win and when does it lose?
We propose that business power in EU legislative decision-making is much more
circumscribed than suggested in both popular and scholarly accounts. In particular,
as most legislative proposals in the EU concern market regulation, and business
mostly opposes market regulation, business faces a defensive battle as soon as the
1
Without the many respondents in the European Commission that were willing to be interviewed
this research would not have been possible. We also acknowledge financial support from the
Austrian Science Fund (FWF), project number I 576-G16.
2
European Commission puts forward a proposal for new legislation. With most
Commission proposals eventually leading to legislation, in many cases business loses
compared to the status quo and at best manages to limit the extent of the losses.
Citizen groups, by contrast, frequently support regulation, and face nearly certain
gains compared to the status quo once the Commission becomes active. Moreover,
business’ ability to defend its interests may suffer further when the policy episode is
marked by controversy and publicity. Hence we expect business to be successful in
defending its preferences only when legislation is discussed with little controversy.
An original and unprecedented data set with the positions of 1,043 interest
groups with respect to 112 controversial issues included in 70 different Commission
proposals introduced between 2008 and 2010 allows us to assess this argument.
Relying on a variety of approaches, we show that empirical support for our argument
is very robust. In fact, our results indicate that on much of EU legislation the
Commission together with citizen groups and the European Parliament want to
change and business interests defend the status quo. The outcome ends up relatively
far away from business preferences.
Our research builds on, and speaks to, a growing literature on interest group
influence and lobbying success, both in national political systems (Alexander et al.
2009; Baumgartner et al. 2009; Bernhagen 2012; McKay 2011; Nelson and Yackee
2012) and at the EU level (Dür and De Bièvre 2007; Mahoney 2007; Schneider et al.
2007; Klüver 2013). While only 15 years ago the study of interest group influence
could be considered an area of “confusion” in the lobbying literature (Baumgartner
and Leech 1998: 13), these recent studies have contributed to a much better under-
3
standing of conditions for interest group influence. The present paper adds a novel
argument, a new data set, and new findings to this literature.
The paper’s contribution to scholarly debates is not limited to the interest group
literature. As put by James March (1955: 432) half a century ago, “[i]nfluence is
to the study of decision-making what force is to the study of motion – a generic
explanation for the basic observable phenomena.” The findings from our research,
therefore, are of interest to the broad literature on policy-making in the EU. In
particular, our study adds a crucial dimension – namely the role of non-state actors
– to the burgeoning literature on EU decision-making (see, for example, Thomson
2011). Finally, our findings are of major importance for an assessment of the EU’s
democratic legitimacy. We show that citizen groups are surprisingly successful, and
business interests surprisingly unsuccessful, thus casting some doubts on prior claims
that policy outcomes in the EU are biased in favor of business interests.
Explaining success and failure in business lobbying
Actors are politically influential to the extent that they succeed in obtaining policies
they prefer while averting policies they dislike, even if the latter are preferred by
other actors (cf. Dahl 1957). In the context of national policymaking, much of the
existing literature suggests that business is frequently more influential than other
actors. This may be so, first, because business has superior access to politically
useful resources such as time, money and expertise, that can be traded for access
to political decisonmakers and favorable legislation (Hillman et al. 2004). Second,
4
business may enjoy a structurally privileged position in the policy process (Lindblom
1977; Przeworski and Wallerstein 1988). Because major investment decisions in a
capitalist system are usually private, pubic officials are very perceptive to the view
of business actors (Lindblom 1977: 170–88). While the classic accounts of business’
structural power are found mostly in the context of redistributive taxation and
social-democratic welfare politics (e.g., Quinn and Shapiro 1991; Swank 1992), the
same kinds of constraints apply to regulatory policy fields. Dryzek (1995: 15), for
example, asserts rather generally that “policies that damage business profitability
– or are even perceived as likely to damage that profitability – are automatically
punished by the recoil of the market.” Thsi has been been argued for the case of
the EU by Chari and Kritzinger (2006).
Third, business actors may be able to influence public policy by virtue of their
control over relevant information (Ainsworth 1993; Austen-Smith 1993; Grossman
and Helpman 2001). While all interest groups may try to influence policy by means
of strategic information transmission, business actors are particularly well positioned
for this. In the course of performing their everyday activities individual firms constantly accumulate knowledge about relevant policy issues (Laffont and Tirole 1986).
Corporations are also more likely to engage in research than other actors (Schlozman
and Tierney 1986). Jointly, these explanatory strands point to the expectation that
policy outcomes are more often in line with the preferences of business actors than
with the preferences of other organized interests.
Nevertheless, this account has not gone uncontested. Business actors may be
better resourced to influence policy than other organized interests, but many non-
5
business interests can muster considerable material resources to mobilize politically
and attempt to influence decision-making. While capitalist political-economic systems give considerable political leverage to business corporations, Galbraith (1954)
has emphasized how “countervailing powers” are mobilized by trade unions and citizen groups. More generally, liberal democracies have not been shy when it comes
to regulating business. As Wilson (1995, 143) points out for the US, “business firms
have been subjected to, on the whole, rising levels of taxation, increased regulatory
supervision, and stronger injunctions against unfair labor practices.”
Our analysis focuses on the time between the formulation of a new regulation
or directive by the European Commission and the final decision on this proposal
by the European legislative institutions. Hence we do not analyze the “seconddimensional” power relations that prevent some interests from successfully defining
issues and placing them on the public agenda (Bachrach and Baratz, 1962). In the
EU as elsewhere, lobbyists would prefer to shape policy before it gets to the stage
of public debate (Dunleavy 1991). However, not all policy advocates manage to
have their voices heard – and accommodated – from the start of the policy process.
While some organized interests may influence the Commission’s policy agenda and
formulation, they still have incentives to lobby in favor of delivery at the later stages,
not least so as to counter lobbying from interests that come out to oppose the policy
proposal in question. Furthermore, as Richardson and Jordan (1979) have argued
for the British political system, policy communities and issue networks can fail to
settle issues early on.
Our argument to explain lobbying success at the decision-making stage rests on
6
three key assumptions. First, we assume that societal interests can exert influence
on the positions held by member states and members of the European Parliament
(MEPs). This assumption is plausible because member state governments and MEPs
want to retain office, and thus strive to avoid concentrated opposition to their policies. We expect both business actors and citizen groups to be influential within
member countries and at the European Parliament. Second, we assume that the
European Commission is a strategic actor that in the EU, legislation can only be
passed if a qualified majority in the Council (and a majority of the EP in cases in
which the ordinary legislative procedure applies) support the proposal. Finally, we
assume that the Commission is concerned about a loss of reputation in case that its
legislative proposal fails.
Whenever these assumptions apply, the Commission should assess support for
and opposition to a change in the status quo before making a proposal on a specific
issue. It will only propose a change to the SQ if there is enough societal support
(relative to opposition) to make a positive response by the Council and the EP likely.
What we should find on the legislative agenda of the EU, therefore, are proposals on
which a move away from the SQ is likely given the constellation of societal interests.
Since the completion of the single market in the 1980s, most legislative activity in the EU is about regulating the markets in which business firms are active
(Majone, 1994; Pedler and Van Schendelen, 1994). While on these regulatory issues
non-business groups representing environmental, health, consumer or labor interests often seek policy change, business actors frequently find themselves defending
the status quo (cf. Long and Lörinczi, 2009; Boessen and Maarse, 2009). Business
7
is unlikely to take a common stance on such regulatory issues, however. Common regulatory standards, even if on a relatively high regulatory level, will reduce
competitive disadvantages for some firms and facilitate intra-European trade. On
average, however, we expect firms to reject stricter regulation. The European Working Time Directive, for example, has met resistance by employers throughout the
EU for many years after it has been passed by the EU institutions in 2003.
With the SQ changing with high probability once the Commission puts forward
a legislative proposal, business interests that want to preserve the SQ are likely to
lose and citizen groups are likely to win. We thus hypothesize:
H 1 Business actors are less successful than citizen groups in influencing EU legislation.
We do not expect business actors to be equally unsuccessful on all legislative
proposals in the EU. Following Schattschneider (1960), many scholars emphasize
how business interests attempt to shape the policy process quietly, avoiding conflict. Thus, Salisbury (1984) found that, while citizen groups tend to play more
prominent roles in publicly visible arenas and newspaper reporting on organized interest representation, firms and trade associations seem to be more active in the less
visible arenas (Salisbury 1984: 74–5). Business groups are more likely to adopt an
inside strategy, meaning direct contacts with decision-makers, than citizen groups
(Dür and Mateo 2013). Thus, much political influence seeking by business actors
may take place behind the scenes. As Danielian and Page (1994, 1076) point out,
“most interest groups fare best when they can work in the dark, when visibility
is low and the scope of conflict is narrow... In contrast, when the spotlight is on
8
and the public gets involved, political equality tends to prevail and special interests
lose.” This is a fortiori the case for business actors. In V.O. Key’s words, the
lobbyists for electrical utilities are “eternally on the job”, while the lobbyists for the
consumers are “conspicuous by their absence” (Key 1964: 166). In their analysis of
lobbying in Washington D.C., Baumgartner and Leech (2001) found that in cases
where only one or two interest groups were involved in an issue, participation was
predominantly by businesses and trade associations. By contrast, unions and citizen
groups were more likely to be involved in the relatively more open and conflictual
processes involving larger numbers of participants (Baumgartner and Leech 2001:
1204). To the extent that the factors causing firms and business groups to dominate
the quiet avenues of lobbying at the expense of non-business groups apply to the
political system of the EU, we expect that,
H 2 Business actors are more successful in influencing EU legislation the less conflictual the policy episode.
Research Design
The unique dataset that we use to assess our theoretical expectations comprises information relating to 70 EU legislative proposals (43 directives and 26 regulations,
with 13 proposals decided by consultation, 51 by co-decision/the ordinary legislative
procedure and 5 by another procedure) that the European Commission put forward
between 2008 and 2010.2 The proposals were selected by means of a stratified ran2
We started with a larger sample of 125 proposals, but unfortunately were not able to conduct or successfully complete interviews for 41 of the 125 legislative proposals (15 remain to be
9
dom sample of the population of all legislative proposals, which was subsequently
filtered on the basis of a minimum level of public visibility.3 Our sample, however,
also includes six proposals for which this latter criterion was removed. The primary
responsibility for the proposals was held by 17 different Commission directorate generals (DGs), indicating that we cover a large number of policy fields. Nevertheless,
no fewer than 13 proposals were initiated by DG Internal Market and Services.
For our sample of legislative proposals, we conducted semi-structured interviews
with Commission officials (often policy officers or heads/deputy heads of unit) that
had particular responsibilities for the relevant legislative proposal at the time of its
initiation, as well as a detailed knowledge of the negotiations that took place at this
time. The interviews lasted on average 70 minutes.
The interviews were used to establish conflict dimensions (issues) within the
proposals and policy positions taken by stakeholders on the issues.4 In total, we
identified 106 conflict dimensions, or about 1.5 issues per proposal. This disaggregation enables us to capture more fully the portion of a proposal that interest
group took an active stance on, whilst facilitating a spatial conception of policymaking. Additionally, the single dimension of policy bargaining allows comparisons
to be made across issues, both within the proposal and across proposals, and, by
extension, across policy fields.
investigated).
3
Public visibility was defined through the number of occurrences in six leading national and panEuropean newspapers (Agence Europe, Euractiv, European Voice, Financial Times, Frankfurter
Allgemeine Zeitung and Le Monde). This approach enabled the purely procedural proposals to
be filtered out, whilst avoiding the overestimation of policy conflict associated with populations
defined entirely by policy actors.
4
The precise question asked respondents to “...identify up to three distinct issues within the
proposal” that involved at least one non-state actor.
10
The officials were asked to spatially locate the policy actors along an issue continuum ranging from 0-100. For this, we first asked for the position of the two most
divergent positions with respect to the original proposal of the Commission. These
policy positions were recorded at either end of a visual representation of the issue
continuum (0 to 100). The officials were then asked to locate the policy alternatives
initially favoured by the other non-state policy advocates on the policy continuum.
Again these were recorded to complete the picture of interest group lobbying. Next,
the officials were asked to locate the policy positions favoured by the European Commission, the those member states that took a clear position on the issue’, and by
the party groups in the European Parliament.5 With the actor positions completed
the officials were asked to locate the position of the final legislative outcome. Finally, to complete the spatial model representation of the policy episode the officials
were asked to locate the reversion point.6 Figure 1 provides an example of the 106
constellations that were derived from these questions.
5
Positional values for governmental actors could be awarded outside the 0-100 range.
The reversion point is the point where the outcome would be located, if no agreement on new
legislation were to be found. In nearly all cases, the reversion point is equal to the status quo. The
question used to find the reversion point was phrased: “If the Commission, Council and Parliament
had failed (were to fail) to reach an agreement on the issue, where would you locate the outcome
of such a situation on the continuum?”
6
11
Figure 1: An illustration
Should there be nutritional information labelling on alcoholic beverages?
Outcome
Wine
Spirits industry
Commission & France Consumers
0
20
RP
No additional labelling
70
100
Mandatory labelling
Through the interviews, we identified a total of 1,043 non-state advocates, of
which 652 are business interests. The percentage of business interests in our data
of 63 percent is thus close to the business share among the registrants to the EU’s
Joint Transparency register (56 percent after professional lobby firms have been
excluded). The mean number of groups per issue is 9 and the maximum number
of interest groups identified on a single issue is 28. Some groups were identified as
active on multiple issues (maximum of 16 occurrences), leaving 461 unique non-state
actors in our data set. We found that business groups are on average active across
fewer issues (2.3) than citizen groups (2.7). There are missing values for 12 out of
112 reversion points; and 10 out of 112 legislative outcomes. As a consequence of
the missing reversion points and legislative outcomes, when calculating a measure of
success we lose about 10 per cent of the observations. Figure 2 shows the positions
we identified for a subset of issues.
12
Figure 2: Our data (sample)
011_001
019_001
024_001
●
●
●
●
●
●
●
●
●
0
20
40
60
80
100
0
Business
Citizen
Commission
Outcome
Reversion point
20
40
Position
026_001
●
●
●
●
●
●
●
●
●
●
60
80
●
●
●
●
100
0
20
40
60
Position
Position
027_001
027_002
80
100
80
100
●
●
●
●
●
●
●
●
●
0
20
40
60
80
100
0
20
40
Position
032_001
●
60
80
100
0
20
40
60
Position
Position
035_001
035_002
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
●
0
20
40
60
80
100
0
20
●
40
60
80
100
0
●
●
●
20
40
60
80
100
The identification of issues and the positional data rely heavily on the information provided by the 70 Commission officials. To ensure that this method has not
biased our findings an additional validation of these results has been made. For
half of the 70 proposals a formal consultation was held for which the Commission
received written responses from interest groups. In each of these consultations, submissions were received from at least one interest group (the maximum number of
groups that submitted a consultation paper and are contained in our data is 11).
The texts of these overlapping documents required manual coding, given the highly
nuanced nature of the language. The initial results from this independent verification indicate that issues identified by the Commission officials were also live at
13
this time of the consultation, and that the policy positions defined in the interviews
correlate well with the positions identified in the documents submitted. Interviews
with 70 interest group officials involved in these campaigns further confirm both the
issue dimensions and the actor lists from the Commission interviews.
Measuring success
The approach that we use rests on the assumption that interest groups have single
peaked and symmetric preferences. In other words, an actor has an ideal point on an
issue; and it prefers outcomes that are close to this ideal point to outcomes that are
further away. Figure 3 illustrates these assumptions, which are common to spatial
models of politics. In this figure, Actor A has an ideal point of 20% on this issue. It
prefers this point to 30%, but is indifferent between 10% and 30%. The figure also
shows that Actor A reaps a utility gain from any decision that is located between
the reversion point (RP, in this example at 35%) and Actor’s ideal point minus the
distance to the RP, in this case 5%.
Given these assumptions, the simplest approach at measuring success is to calculate the absolute difference between an actor’s ideal point and the outcome. The
closer an actor is to the final outcome, the greater her success (e.g. Verschuren and
Arts 2004).7 The disadvantage of this measure of success is that it does not take
into account the location of the reversion point (Aksoy 2010). Imagine the following
scenario: actor A has an ideal point of 20 and actor B of 50. The outcome is located
7
This approach is also used by studies of member state bargaining power in the EU - see Golub
2012. We use the term “success” here, as our measure contains both power and “luck” (Barry
1980). For a full discussion of the difficulties of measuring influence, see Dür 2008. For a longer
treatment of our specific approach, see Bernhagen et al. 2013.
14
Figure 3: A spatial approach to understanding interest group success
Utility gain
Single peaked preference (Actor A)
0%
10%
Less taxes
20%
30%
Actor A
40%
RP
50%
More taxes
at 35, so following this simple measure actors A and B are equally successful. Now
suppose that the reversion point is located at 20: actor B actually managed to avert
a much worse outcome, whereas the result moves the policy away from the ideal
point of A. For most purposes, in this case B should be considered more successful
than A.
Our measure of success (Success), therefore, builds on the idea that an actor is
more successful, the more it manages to pull the outcome closer to its ideal point
relative to the reversion point. If the distance to the reversion point was large, and
the distance to the outcome is small, an actor should be considered highly successful
(and vice versa if the distance to the reversion point was small but the distance to
the outcome is large, the actor was little successful). Formally,
sij =| (xij − RPj ) | − | (xij − Oj ) |
(1)
where the subscripts i and j denote the actor and the issue, respectively, s is
15
the success measure, x is the ideal point, and O is the outcome. The larger sij ,
the greater the success of actor i on issue j. Since we asked Commission officials to
identify only those non-state actors that played a prominent role in a policy debate,
we can expect salience to be a constant in our data set. Therefore, we do not weigh
this measure of success by salience.
A potential objection to our measure is that distances may not be comparable
across issues. That is, a gain of 50 on one issue may not be the same as a gain
of 50 on another issue, as the 0-100 scales used in the interviews are determined
separately for each issue. On one issue, what separates positions 0 and 100 may be
a relatively minor aspect of a policy, whereas on another the difference may be much
more important.8 Since we asked respondents only about highly controversial issues,
we do not think that this is a major concern for our data.9 Furthermore, as gains
and losses are always relative to what is possible, our standardized spatial model
maximizes comparability across issues. Nevertheless, to ease these concerns it makes
sense to compare results with this measure with a dichotomous success measure
(Success (di)). This dichotomous measure equals 1 if | (xi − RP ) |>| (xi − O) |, and
0 otherwise.
Figure 4 shows the distribution of both Success and Success (di) using our data.
As can be seen, Success ranges from -100 to +100, with a peak at 0. It has a mean
of 3.2 and a median of 0. This is about a standard deviation lower than the success
score for an actor with random position, given the outcomes and reversion points
8
There are also some issues where only the extreme positions of 0 or 100 make sense (dichotomous issues).
9
In fact, users of the Decision-making in the European Union (DEU) dataset (Thomson et al.
2006) also treat spatial distances as comparable across issues (see Aksoy 2010; Golub 2012).
16
included in our data set.10 For Success (di) we find that a slight majority of groups
is not successful.
Figure 4: Distribution of the success measure
Success (di)
250
200
100
150
0.010
0
0.000
50
0.005
Density
0.015
300
0.020
350
Success
−100
−50
0
50
100
0
1
Success
Note: because of missing observations for the outcome and the reversion point, we
lose 326 observations.
Predictors and control variables
Hypothesis 1 stresses actor type as the main explanatory variable. We distinguish
three types of actors: business, citizen groups and other (including institutions, professional associations, labor unions and governmental actors). We include Business
and Other as dummy variables in the analysis. In one variant of the analysis, we
break down Business in firms (F irm) and business associations (Bus.associations).
Hypothesis 2 emphasizes the degree of conflict on a specific issue. We operationalize
conflict by calculating the standard deviation of interest group positions for each
10
A simulation with 1000 runs gives a mean of 9.7 and a standard deviation of 4.3.
17
issue (IG Conf lict). This variable ranges from 0 to 71, with a mean of 37 and a
median of 42.
In our multivariate analysis below, we also include a series of control variables.
For one, it can be expected that the larger the number of groups that are pulling
in the direction of the position taken by a specific actor, the greater the (likelihood
of) success of that actor will be (Side, see also Klüver 2013). We operationalize this
variable using the number of groups that coincide with the position of the actor, with
the positions being left (0-20), middle (20.1-79.9) and right (80-100). On average,
3.6 groups take a left position, 1.0 a middle position, and 4.7 a right position.
A group’s success may also depend on its endowment with resources, in particular information. Our operationalization of this variable is based on a question put
to our respondents about the technical knowledge possessed by the interest groups
identified in the interviews.11 Respondents could rate that knowledge on a fivepoint scale from very low to very high (IG Knowledge). The modal value in our
dataset is 5, meaning that the Commission officials evaluated most groups as highly
knowledgeable. Nevertheless, there is some variation across groups, with 128 actors
evaluated as having very low or low technical knowledge.
Success may also depend on the distance between an actor’s ideal position and
the European Commission’s position. As agenda setter, the Commission plays a
crucial role in determining EU policies. Acting strategically, the Commmission
11
The exact wording of the question is: “Non-state actors such as firms, interest groups and
regional representations differ from each other regarding the technical knowledge they possess.
By technical knowledge we mean detailed information on, and an in-depth understanding of, the
substance of the proposal. Thinking about each non-state stakeholder, can you please tell me the
level of technical knowledge each organisation holds with respect to this policy proposal? Please
use a five point scale, ranging from very low, to low, medium, high and very high.”
18
produces policy proposals that are likely to find the suport of the required majorities
in the Council of Ministers and in the EP. Because many societal actors try to shape
the the Commission’s agenda early on in the policy process, proposals are also likely
to be supported by a variety of organized interests. As a result, outcomes are likely
to be close to the Commission position. To account for the Commission’s role in
shaping outcomes, we include a measure of the absolute distance between an actor’s
ideal point and the Commission position (Distance Com). This variable varies from
0 to 100, with a mean of 46 and a median of 50. On average, therefore, groups are
relatively far from the Commission’s position. A reason for this may be that those
groups that are close to the Commission see little value in lobbying and rely on the
Commission to defend their interests. We expect groups to be less successful the
further they are away from the Commission’s position.
We also include a variable measuring media attention in our model (M edia).
This is the number of times that six newspapers reported about a specific proposal.
As we expect media attention to have a marginally decreasing effect on the policy
decision, we use the natural log of that number. Finally, we include a nominal variable coded 0 for directives and 1 for regulations (P roposal type). Our expectation
is that groups are more successful when lobbying on regulations than on directives,
because the national transposition constraint that applies to the latter may allow
for less giving-in to interest group demands at the European level. For directives,
we have 651 interest group positions, and for regulations 376 positions.
As our argument rests on the assumption that the positions of the Council and
the European Parliament are influenced by lobbying, and thus reflect interest group
19
positions, we do not control for them in our main models. When controlling for
these positions, we rely on data from our interviews on the Council majority, the
member states’ positions, and the Parliament majority, all of which we have on the
same scale as the one used to locate the interest groups. On the one hand, we calculate the absolute distance between a group’s position and the Council majority
(Distance Council). As we have a large number of missing values for the Council majority, we imputed these observations using the median position across the
member states for which we have a position. On the other hand, we calculate the
absolute distance to the European Parliament majority (Distance EP ). While on a
average the Council tends to side with business interests, the European Parliament
tends to be on the same side as citizen groups. Table A-1 in the Appendix shows
descriptive statistics for these variables.
Empirical Analysis
Descriptive analysis
Our argument resulted in two hypotheses: 1.) that business groups are less successful
than citizen groups and 2.) that business success is higher on less controversial issues
with little conflict between interest groups. To facilitate interpretation we begin with
a descriptive analysis of these questions. First, we look at the median positions of
business and citizen groups relative to the median outcome and the median reversion
20
point (see Figure 5).12 The figure offers some interesting evidence: as expected in
the argument, for a large segment of the EU’s legislative agenda, the reversion point
is located to the left (little regulation at the EU level). Business groups tend to be
close to the reversion point. The Council is relatively close to business. On the other
extreme, citizen groups pull towards more regulation, and are supported in that aim
by the European Commission and especially the European Parliament. The median
outcome is located exactly in the middle of the scale. For the question of success, this
means that citizen groups gain a lot compared to the reversion point, and business
loses, even if business groups are closer to the final outcome than citizen groups.
Figure 5: Median positions
RPBusiness
0
10
Council Outcome Commission
38
50
Less regulation
70
EP
90
Citizen
100
More regulation
Note: RP=reversion point; EP=European Parliament.
Next, in Table 1 we compare the mean values for business groups and citizen
groups across our two measures of success and find support for our argument. Citizen
groups turn out to be substantially more successful than business groups.13 The
12
We use the median rather than the mean to avoid having the values influenced by occasional
outliers.
13
And also the “random player” mentioned above in footnote 10 is more successful than business.
21
difference for Success is more than half a standard deviation. For Success (di), the
proportion of successful citizen groups is nearly double the proportion of successful
business groups.
Table 1: Mean success of business and citizen groups
Success
Success (di)
Business
-3.47
0.38
Citizen groups
26.59
0.73
Multivariate analysis
We now turn to a multivariate analysis of our data, relying on ordinary least squares
regression for Success and logistic regression for Success (di). Across all our models,
we take into account that groups are nested in issues and proposals. We therefore
estimate hierarchical models with random effects at both the issue and the proposal
level (Gelman and Hill 2007).14
Table 2 summarizes the results, with citizen group as the base category for group
type. Looking first at Models 1a and 1b, the coefficient for business is negative
and highly statistically significant. According to the estimates for Model 1a, the
disadvantage of being a business actors is a bit smaller than in the descriptive
analysis, but still substantially important: business has a success score that is about
a quarter of a standard deviation lower than the one of citizen groups. The effect
is larger in the model using Success (di) (Model 1b): here the predicted value on
14
Likelihood-ratio tests indeed show that hierarchical models are appropriate.
22
the dependent variable is 0.72 for a citizen group and 0.34 for business (keeping all
other variables at the mean or the mode).
Interestingly, the coefficient for the category of “other actors”, which includes
institutions, think tanks but also EU and foreign governmental actors, is also negative and statistically significant. Several of the control variables have the expected
effect. Media attention seems to go hand-in-hand with more business success, but
this effect is not statistically significant once controlling for clustering by issue and
proposal. Greater conflict among interest groups is related to less success. The
type of legislative act put forward by the Commission does not affect success. By
contrast, actors’ distance to the European Commission has a strongly statistically
significant negative effect on success. The models have a reasonable fit: Model 1b,
for example, correctly predicts 77 per cent of the observations.
In Model 2, we show that our results are not sensitive to controlling for the positions of political institutions (see Table 2). Concretely, we control for the distance
between an interest group’s position and the Council majority and the European
Parliament majority. Doing so does not change our main finding, namely that the
coefficient for Business is negative and statistically significant. As could be expected, however, greater distance from the political institutions reduces the success
of groups.
Hypothesis 2 suggests that business success should be conditional on the degree
of conflict on a specific issue. Concretely, we expect that business success will be
lower, the greater conflict. Model 3 in Table 2, which includes an interaction effect
between Business and IGConf lict, strongly supports this expectation. When in-
23
Table 2: Multivariate analysis
Model 1a
−20.98***
(4.07)
−14.04***
(4.88)
−0.92*
(0.51)
1.38
(3.06)
0.90
(1.35)
−0.53***
(0.18)
−1.47
(5.61)
−0.81***
(0.04)
Business
Other
Side
Media
IG Knowledge
IG Conflict
Proposal type
Distance Com
Distance Council
Distance EP
Business * IG Conflict
Constant
76.72***
(12.81)
Observations
No. of proposals
No. of issues
Log Likelihood
Var. (Issue)
Var. (Proposal)
Var. (Residual)
***
p < 0.01,
**
836
57
89
−4223.64
20.88
0.00
35.17
Model 1b
Model 2
Model 3
−2.88*** −17.50*** 16.03
(0.50)
(5.06)
(10.00)
−2.40*** −17.78*** −11.61**
(0.59)
(5.88)
(4.87)
0.09
−0.04
−1.06**
(0.06)
(0.62)
(0.51)
0.69
1.55
2.09
(0.47)
(4.42)
(3.01)
0.32**
2.39
1.15
(0.15)
(1.67)
(1.34)
−0.05*
−0.24
0.05
(0.03)
(0.23)
(0.23)
0.48
−2.69
−3.30
(0.86)
(7.73)
(5.51)
−0.06*** −0.61*** −0.80***
(0.01)
(0.06)
(0.04)
−0.14***
(0.05)
−0.39***
(0.06)
−0.95***
(0.24)
2.49
69.19*** 51.73***
(1.89)
(16.84)
(14.02)
836
57
89
−327.68
3.17
0.65
p < 0.05, * p < 0.1
24
562
33
55
−2812.31
23.33
0.00
33.21
836
57
89
−4215.57
20.39
0.00
34.88
terest group conflict is low, business actors are more successful than citizen groups
at attaining preferred policy output. As the interaction effect shows, however, the
greater interest group conflict, the lower business success. Figure 6 shows the interaction effect included in Model 3 graphically. At low levels of interest group
conflict, business and non-business groups cannot be distinguished in terms of lobbying success. Only as the level of conflict increases, the effect of business turns
negative.
-50
Success (predicted)
0
50
Figure 6: Business success and interest group conflict
0
10
20
30
40
IGConflict
Citizen groups
50
60
70
Business
Robustness checks
The findings are highly robust to changes in operationalization of key variables. For
one, we check whether our results are sensitive to the specific operationalization of
Success that we used. In particular, it may be argued that a measure of success
should not only capture the change of the outcome relative to the reversion point,
but also take into account how far the outcome is from the position of an actor. In
25
a hypothetical example with two actors A and B with ideal points at 50 and 100
respectively, using the operationalization of Success in equation 1 above, both gain
50 if the reversion point is at 0 and the outcome is at 50. But actor A may be seen
as having been more successful, as the final outcome coincides with its ideal point.
To account for this, we create a second measure of success, which is calculated as
follows (see also Bernhagen et al. 2013):
sij =
| (xij − RPj ) | − | (xij − Oj ) | +100
| (xij − Oj ) + 100
(2)
Theoretically, this measure can range from 0 to 2. In the hypothetical example, the
formula gives a value of 1.5 to actor A and a value of 1 to actor B.
In Model 4 in Table 3, we re-run the analysis with this alternative measure of
success. The result is substantially the same as for our first measure. The coefficient
for Business is negative and statistically significant. Business thus is less successful
than citizen groups in EU legislative decision-making.
Furthermore, Model 5 in Table 3 shows that disaggregating the business category
to firms and associations does not change our results. The coefficients for both
firms and business associations are negative and statistically significant. Clearly,
therefore, we do not see powerful firms and weak associations, but both types of
business actors with relatively little success in EU legislative decision-making. The
other coefficients are very similar to those reported in Model 1a.
In Model 6 we show that our results are robust to assuming that all issues for
which we are missing an outcome measure have failed, that is, that the outcome
equals the reversion point. This adds seven proposals and nine issues to our data,
26
Table 3: Robustness checks
Model 4
Model 5
Model 6
Model 7
Business
−0.09**
(0.04)
Firm
−17.03***
(4.78)
−21.73***
(4.16)
−0.08
−13.46*** −13.18*** −17.29***
(0.05)
(4.86)
(4.81)
(6.18)
−0.01**
−0.98*
−0.99**
−0.51
(0.00)
(0.51)
(0.50)
(0.61)
0.01
1.37
1.76
0.39
(0.03)
(3.04)
(2.95)
(3.26)
0.01
0.46
0.34
1.75
(0.01)
(1.37)
(1.29)
(1.77)
−0.00*** −0.54*** −0.53***
0.42
(0.00)
(0.18)
(0.17)
(0.31)
−0.02
−1.65
−2.24
1.73
(0.05)
(5.57)
(5.19)
(6.01)
−0.01*** −0.81*** −0.76*** −0.82***
(0.00)
(0.04)
(0.04)
(0.06)
1.41*** 78.56*** 75.85*** 33.51*
(0.12)
(12.83)
(12.36)
(17.10)
Business association
Other
Side
Media
IG Knowledge
IG Conflict
Proposal type
Distance Com
Constant
Observations
No. of proposals
No. of issues
Log Likelihood
Var. (Issue)
Var. (Proposal)
Var. (Residual)
***
p < 0.01,
**
836
57
89
−343.68
0.20
0.00
0.34
−20.26*** −21.41***
(3.98)
(4.70)
836
57
89
−4223.29
20.68
0.00
35.19
p < 0.05, * p < 0.1
27
866
61
95
−4373.03
19.93
0.00
35.18
553
32
50
−2803.32
12.41
7.12
36.91
but without this changing the results.
Finally, not all issues are lobbied on by business and citizen groups at once.
Repeating the analysis for issues on which both business and citizen groups were
active shows that the results are robust to this restriction (see Model 7 in Table 3.
Business success, however, is higher on issues with no lobbying by citizen groups
(for these 37 issues, the mean on Success for business groups is 13.5 as compared
to -2.2 for the full sample). This indicates that what our analysis captures is the
lobbying of citizen groups offsetting the lobbying of business groups.
Discussion
Popular and academic concern about business influence frequently assumes that
business actors are both successful at promoting their own agendas and manage
to avert policies that might be socially desirable but costly for business, such as
environmental or labor regulation. Our findings suggest that the latter is rarely the
case. We have argued and empirically shown that, at the decision-making stage,
business interests are less successful in EU legislative politics than citizen groups.
With business interests mostly defending the status quo and citizen groups together
with the European Commission and the European Parliament pushing for policy
change, the former tend to be in a defensive position with respect to much legislative
activity in the EU. Business success is both bigger and more likely on less conflictual
policy episodes, when business interests face limited opposition from other actors.
These findings are robust to changes in the operationalization of success and the
28
inclusion of numerous control variables.
The findings have major implications for our understanding of interest group
influence and decision-making in the EU. With respect to the former, it counters
arguments that business interests tend to be more influential because of the resources
they possess, their greater mobilization, their privileged access to decision-makers or
their ability to overcome collective action problems. With respect to the latter, we
show that policy outcomes in the EU tend to be a compromise between a coalition
comprising citizen groups, the European Commission and the European Parliament
and another coalition comprising most member states and business interests.
What does all of this say about business influence? It is of course possible that
business is highly influential at the agenda-setting stage. What we observe at the
decision making stage that is the subject of our analysis is then merely the struggle
over the last chips on the table. Yet, the Commission proposals in our data are often
diametrically opposed by business interests, suggesting that the agenda stage too
might be influenced by interests other than business. Another qualification might be
that business actors may be influential on the non-salient and non-conflictive issues
that did not make it into our sample but that would be marked by a considerable
degree of loophole lobbying by business. However, compared to the US, lobbying
for loopholes is a less important aspect of interest group politics in the EU.
A more plausible interpretation would suggest that the entire project of European economic and monetary integration has been a great political success for European business. According to Scharpf (2001), the agenda of European integration
leading to the Single European At and the Maastricht Treaty consisted of “market-
29
creating policies” that were strongly supported by business interests and that led
to the single market and currency union. By contrast, the subsequent multitude of
“market-enabling” and “process regulations”, even if beneficial to the operations of
European business actors, often cause costs for business and are frequently opposed
by business actors. These are the policies that make up our sample, and on these
business often has to give way to consumer and environmental interests siding with
the European Commission. Thus, while the European Union might be seen as a
project whose main purpose has been to serve the interests of European business,
when it comes to regulating the way in which business is done in Europe, business
actors are substantially less influential than is often perceived.
Bibliography
Aksoy, Deniz (2010) ‘It Takes a Coalition’: Coalition Potential and Legislative Decision Making. Legislative Studies Quarterly 35 (4): 519–42.
Alexander, Raquel, Stephen W. Mazza, and Susan Scholz (2009) Measuring Rates
of Return on Lobbying Expenditures: An Empirical Case Study of Tax Breaks
for Multinational Corporations. Journal of Law and Economics 25 (4): 401–58.
Austen-Smith, David (1993) Information and Influence: Lobbying for Agendas and
Votes. American Journal of Political Science 37 (3): 799–833.
Bachrach, Peter, and Morton S. Baratz (1962) Two Faces of Power. American
Political Science Review 56 (4): 947–52.
Barry, Brian (1980) Is It Better to be Powerful or Lucky? Part 1. Political Studies
30
28 (2): 183–94.
Baumgartner, Frank R., and Beth L. Leech (2001) Interest Niches and Policy Bandwagons: Patterns of Interest Group Involvement in National Politics. Journal
of Politics 63 (4): 1191–1213.
Baumgartner, Frank R., and Beth L. Leech (1998) Basic Interests: The Importance
of Groups in Politics and Political Science (Princeton: Princeton University
Press).
Baumgartner, Frank R., et al. (2009) Lobbying and Policy Change: Who Wins,
Who Loses, and Why (Chicago: University Of Chicago Press).
Beer, Samuel H. (1965) Modern British politics: a study of parties and pressure
groups (London: Faber).
Bernhagen, Patrick (2012) Who Gets What in British Politics – and How? Analyzing
Media Reports on Lobbying around Government Policy, 2001–2007. Political
Studies 60 (3): 557–577.
Bernhagen, Patrick, Andreas Dür, and David Marshall (2013) Measuring Interest
Group Success Spatially, unpublished paper.
Boessen, Sandra and Maarse, Hans (2009) A Ban on Tobacco Advertising: The Role
of Interest Groups, in David Coen and Jeremy Richardson (eds) Lobbying the
European Union: Institutions, Actors, and Issues (Oxford: Oxford University
Press), 212–32.
Bunea, Adriana (2013) Issues, preferences and ties: determinants of interest groups’
preference attainment in the EU environmental policy. Journal of European
Public Policy 20 (4): 552–70.
31
Chari and Kritzinger 2006
Coen, David (1997) The Evolution of Large Firm Political Action in the European
Union. Journal of European Public Policy 4 (1): 91–108.
Dahl, Robert A. (1957) The Concept of Power. Behavioral Science 2 (3): 201–15.
Danielian, Lucig H., and Benjamin I. Page (1994) The Heavenly Chorus: Interest
Group Voices on TV News. American Journal of Political Science 38 (4): 1056–
1078.
Dryzek, John (1995) Political and ecological communication. Environmental Politics
4 (4): 13–30.
Dunleavy, Patrick (1991) Democracy, bureaucracy and public choice: economic explanations in political science (Harvester: Wheatsheaf).
Dür, Andreas (2008) Measuring Interest Group Influence in the EU: A Note on
Methodology. European Union Politics 9 (4): 559–76.
Dür, Andreas, and Dirk De Bièvre (2007) Inclusion without Influence? NGOs in
European Trade Policy. Journal of Public Policy 27 (1): 79–101.
Dür, Andreas, and Gemma Mateo (2013) Gaining access or going public? Interest group strategies in five European countries. European Journal of Political
Research.
Galbraith, John K. (1954) Countervailing Power. The American Economic Review
44 (2): 1–6.
Gelman, Andrew, and Christopher Hill (2007) Data Analysis Using Regression and
Multilevel/Hierarchical Models (Cambridge: Cambridge University Press).
Golub, Jonathan (2012) How the European Union does not work: national bargain-
32
ing success in the Council of Ministers. Journal of European Public Policy 19
(9): 1294–1315.
Greenwood, Justin (2011) Interest Representation in the European Union (Houndmills: Palgrave).
Hillman, Amy J., Gerald D. Keim, and Douglas Schuler (2004) Corporate Political
Activity: A Review and Research Agenda. Journal of Management 30 (6):
837–857.
John, Peter (1998) Analysing Public Policy (Continuum International Publishing
Group).
Kautto, Petrus (2007) Industry-government interaction in the preparation of a new
directive: Nokia, industry associations and EuP. European Environment 17 (2):
79–91.
Key, Valdimer O. (1964) Politics, Parties and Pressure Groups (5th ed) (New York:
Crowell).
Kingdon, John W. (1984) Agendas, alternatives, and public policies (Boston: Little,
Brown).
Klüver, Heike (2013) Lobbying in the European Union: Interest Groups, Lobbying
Coalitions, and Policy Change (Oxford: Oxford University Press).
Laffont, Jean-Jacques, and Jean Tirole (1986) Using Cost Observation to Regulate
Firms. Journal of Political Economy 94 (3): 614–641.
Lindblom, Charles E. (1977) Politics and Markets: The World’s Political-Economic
Systems (New York: Basic Books).
Long, Tony, and Larissa Lörinczi (2009) NGOs as Gatekeepers: A green vision.
33
In David Coen and Jeremy Richardson (eds) Lobbying the European Union:
Institutions, Actors, and Issues (Oxford: Oxford University Press).
Mahoney, Christine (2007) Lobbying Success in the United States and the European
Union. Journal of Public Policy 27 (1): 35–56.
Majone, Giandomenico (1994) The Rise of the Regulatory State in Europe. West
European Politics 17 (3): 78–102.
March, James (1955) An Introduction to the Theory and Measurement of Influence.
American Political Science Review 49 (2): 431–51.
McKay, Amy (2011) Buying Policy? The Effects of Lobbyists’ Resources on Their
Policy Success. Political Research Quarterly 65: 908–923.
Nelson, David, and Susan Webb Yackee (2012) Lobbying Coalitions and Government
Policy Change: An Analysis of Federal Agency Rulemaking. Journal of Politics
74 (2): 339–53.
Olson, Mancur (1965) The Logic of Collective Action (Cambridge: Harvard University Press).
Pedler, Robin H., and Rinus Van Schendelen (eds) (1994) Lobbying in the EU:
Companies, Trade Associations and Issue Groups (Dartmouth: Aldershot).
Polk, Andreas (2002) How Special Interests Shape Policy: A Survey (Zürich: Socioeconomic Institute, University of Zürich).
Przeworski, Adam, and Michael Wallerstein (1988) Structural Dependence of the
State on Capital. American Political Science Review 82 (1): 11–29.
Quinn and Shapiro (1991)
Richardson, Jeremy John, and Grant A. Jordan (1979) Governing Under Pres-
34
sure: The Policy Process in a Post-Parliamentary Democracy (Oxford: Martin
Robertson, Limited).
Sabatier, Paul (1999) The Need for Better Theories. In Paul Sabatier (ed) Theories
of the Policy Process (Boulder, CO: Westview).
Salisbury, Robert H. (1984) Interest Representation: The Dominance of Institutions.
The American Political Science Review 78 (1): 64–76.
Schattschneider, Elmer E. (1960) The Semisovereign People (New York: Holt, Rinehart, and Winston).
Schlozman, Kay Lehman, and John T. Tierney (1986) Organized Interests and American Democracy (New York: Harper & Row).
Schneider, Gerald, Daniel Finke, and Konstantin Baltz (2007) With a Little Help
from the State: Interest Intermediation in the Domestic Pre-Negotiations of EU
Legislation. Journal of European Public Policy 14 (3): 444–59.
Sunstein, Cass (1998) Health-Health Trade-offs. In Jon Elster (ed) Deliberative
Democracy (Cambridge: Cambridge University Press).
Swank (1992)
Thomson, Robert (2011) Resolving Controversy in the European Union (Cambridge:
Cambridge University Press).
Thomson, Robert, Frans N. Stokman, Christopher Achen, and Thomas Koenig, eds
(2006) The European Union Decides (Cambridge: Cambridge University Press).
Verschuren, Piet, and Bas Arts (2004) Quantifying Influence in Complex Decision
Making by Means of Paired Comparisons. Quality & Quantity 38 (5): 495–516.
Weber, Max (1978) Economy and Society: An Outline of Interpretative Sociology
35
(Berkeley: University of California Press).
Wilson, James Q. (1995) Political Organizations (Princeton: Princeton University
Press).
36
Appendix
Table A-1: Descriptive statistics
Statistic
Success
Success (di)
Business
Firm
Bus. association
Other
IG Conflict
Side
Media (logged)
IG Knowledge
Proposal type
Distance Com
Distance Council
Distance EP
N
Mean
St. Dev.
Min
Median
Max
924
924
1,043
1,043
1,043
1,043
1,043
1,043
1,028
939
1,043
1,027
1,018
636
5.05
0.45
0.63
0.22
0.41
0.19
36.97
5.71
2.59
3.98
0.35
45.47
52.02
43.38
50.70
0.50
0.48
0.41
0.49
0.39
15.13
3.70
0.89
1.17
0.48
35.35
39.49
36.83
−100.00
0
0
0
0
0
0.00
0
0.00
0.00
0
0.00
0.00
0.00
0.00
0
1
0
0
0
41.93
5
2.71
4.00
0
50.00
50.00
40.00
100.00
1
1
1
1
1
70.71
14
4.69
5.00
1
100.00
100.00
100.00
37