Sweet-talk or policy concessions? Interest group influence in consultations

1

Sweet-talk or policy concessions?

Interest group influence in consultations

Paper prepared for the 84th Annual Meeting of the Southern Political Science Association

Orlando January 3-5 2013

(Not to be quoted)

Anne Skorkjær Binderkrantz ([email protected])

Peter Munk Christiansen ([email protected]) and

Helene Helboe Pedersen ([email protected])

Department of Political Science and Government, Aarhus University

2

Abstract

Governmental proposals are routinely sent out for consultation with interested parties.

Interest groups may make their voices heard and affect public policy. This paper

investigates factors that affect the success of groups in consultations. The analyses draw on

a dataset containing all consultations on Danish bills in the parliamentary year 2009/2010.

1,692 groups replied to 209 consultations carried out across the full set of policy areas.

Three sets of factors that potentially affect influence are tested: Group level variables turn

out not to be as important as often assumed. Variables related to the issues and to the

nature of mobilization of groups in the consultation have a profound influence on the

likelihood of success. Success is also found in regard to specific regulation and bills that do

not cut public expenditure. Groups who are generally positive towards the proposed

legislation are able to gain more accommodation than groups who oppose a proposal.

Finally, influence is likely if many groups are mobilized and opposition to the bill is low.

1. Introduction

In the preparation of policies, governments consult with interested parties. Informal as well as

formal consultations serve to involve organized interests in policy making. Interest groups value

the opportunity to participate in consultations and the sheer amount of responses proves that

groups are willing to spend resources on their participation (Binderkrantz, 2005; Furlong &

Kerwin, 2004; Halpin & Binderkrantz, 2011). Consultations allow affected interests to be involved

in policy making and may therefore enhance the quality and legitimacy of policies. Involving

groups in preparation of political decisions may, however, also lead to unbalanced influence as

some interests may be more adept than others at utilizing this instrument. Investigating the extent

to which interest groups are successful in shaping policies through consultation processes can

enhance our understanding of the functioning of democratic politics.

Our knowledge of the outcome of government consultations is limited. It may be that major policy

concessions are made in the consultation process, but it could also be that little change happens

after initial drafts of bills have been prepared. In this case, all interest groups gain from

consultations could be administrative “sweet-talk”. A few studies have demonstrated that

3

consultations do indeed lead to change. Using quantitative text analysis Klüver (2009;

forthcoming) has found that EU policies change in ways that may be linked to interest group

responses in consultations. US studies of administrative rule-making have shown that groups are

indeed able to gain policy concessions in the notice and comment period. In these studies interest

group comments and suggestions have been directly linked to agency alterations of rules (S. W.

Yackee, 2005: 3). Agencies are found to alter rules in response to the most dominant side in

consultations and more often than not business groups come out as winners (McKay & Yackee,

2007; J. W. Yackee & Yackee, 2006).

We investigate the effects of interest group responses to consultations on bills before these are

introduced in the Danish parliament. In mapping influence we follow Yackee et al. (2006) and link

group responses to agency reactions. The setting is different, however, as our study focuses on a

parliamentary democracy – Denmark – and investigates consultations on bills rather than on

administrative regulations. The analyses draw on a dataset covering all bills introduced in the

Danish parliament during the parliamentary session 2009/10. In Denmark almost all bills (and in

the year under scrutiny all bills that were eventually passed1) are introduced by the government.

Draft bills are routinely sent out for consultation before being presented to parliament. We have

registered all interest group responses in these consultations and related them to ministerial

reactions. We have benefitted from a practice of drawing up consultation reports detailing

responses to comments received in the consultation. Consultation-response data are linked to

survey data about group resources and other factors at the group level. We are therefore able to

investigate the effects of three sets of factors previously argued to affect influence: 1) group

characteristics such as group type and group resources, 2) characteristics of the policy issue in

question and 3) factors related to the mobilization of groups in a specific consultation.

1. Except one bill regarding housing facilities for MPs put forward by the parliament’s executive committee and

enacted unanimously.

4

2. Shaping bills to be proposed to parliament

2.1 Government consultations as a group arena

Political influence is a fundamental to interest groups. Studying influence empirically is, however,

notoriously difficult and many scholars have shied away from the issue and devoted their

attention to more manageable questions (Dür, 2007). Recent scholarship has taken up the challenge

of studying group influence empirically. Two main approaches can be identified in this literature.

Some focus on goal attainment by linking policy positions of groups to political outcomes

(Baumgartner et al., 2009; Bernhagen, 2011; Klüver, 2009; Mahoney, 2007). Others focus on interest

group influence at specific stages of the policy process (for a review see: Leech, 2010). This latter

approach has the advantage that it links the actions carried out by groups and their effects in terms

of political influence. It only captures a part of the policy process in a given study, but the hope is

that scholars can collectively examine “the development of policies from cradle to grave” (Furlong,

1997: 341).

The present study focuses on group influence in consultations on bills before they are introduced

to parliament. Much scholarly attention has been devoted to group attempts to affect the fate of

legislation once it has reached parliament. In many countries, this latter stage of the decision

making process is, however, only one of several stages where influence may be sought. Generally,

groups target different venues and use a variety of tools in their attempts to affect politics

(Baumgartner & Leech, 1998; Beyers, 2004; A. Binderkrantz, 2005). A particularly crucial venue in

regard to affecting bills is the bureaucratic entity responsible for drafting legislation. Groups may

be involved in informal or formal consultations with the bureaucracy and while responding to

consultations is only one among many instruments, surveys indicate that groups regard this as

important (A. Binderkrantz, 2005; Furlong & Kerwin, 2004: 366). Our focus on influence exercised

through formal consultations enables a systematic comparison of interest group consultation

responses and their effects across a large number of groups and issues.

In mapping the effects of group responses we follow the approach used first by Golden (1988) in a

study of selected rules and later by Yackee et al. (2006) in a broader study of US administrative

rule-making. In contrast to studies linking group positions to policy outcomes (Klüver, 2009) we

5

restrict the analysis to instances where the documentary material clearly documents a direct

relation between a group response and an agency reaction. In contrast to asking groups about their

effectiveness in for example a survey, we do not need to deal with problems of under- or over-

reported influence (Golden, 1998: 248). A key limitation is, however, that we focus on a narrow

part of the policy process. Much may have happened before the consultation, much may happen

later. Nevertheless, this study can provide a part of understanding group influence.

In this setting, influence is defined as the ability to change a proposed bill. In their responses to

consultations, groups typically provide ministries with suggestions for improvement (or they

simply advise the ministry to drop the bill). These suggestions may be neglected, they may be

partly accommodated, or the group may get all it wants. The standard of evaluation is thus the

group’s own consultation response – does the group get what it asks for? Based on this, influence

is analyzed in each specific consultation and is aggregated over the full set of consultations in

order to see whether some groups are successful across a wide range of bills. Before we explain the

research design in more detail, the next section discusses the factors that possibly shape influence

in consultations.

2.2 What factors shape group influence?

Generally, we expect ministries to be attentive to group comments. A group response to a

government consultation can be seen as the transmission of information. Groups may signal

potential problems or inform the administration about possible unintended consequences of the

bill. They may also send signals of a more political character indicating that negligence of the

group’s point of view may lead to public criticism and negative attention (S. W. Yackee, 2006: 728).

To the extent that such signals are deemed relevant, ministries may consider adjustments,

although with two limitations: different groups’ comments may be conflicting; and groups’

comments may contradict the ministry’s policy. In balancing these different considerations three

sets of factors may be relevant: Influence may be related to group level variables such as group

type and group resources; policy characteristics may affect the patterns of influence; and the

pattern of mobilization of groups in the specific consultation may matter for the likelihood of

success.

6

Different groups, different rates of success?

A dominant perspective in the literature argues that some groups are simply more likely to be

politically successful than others. Business interests have repeatedly been singled out as

particularly well represented in the political system and better equipped to affect policy making

than other types of groups (Lindblom, 1977; Lowery et al., 2005; Schattschneider, 1975 [1969];

Schlozman et al. 2012). Theories of agency capture emphasize how business interests may assume

a very dominant role vis-à-vis regulatory agencies (Furlong, 1997: 328; Stigler, 1971). Participation

in administrative consultations has indeed been found to be highly skewed with business interests

being more active than other types of interests (Furlong & Kerwin, 2004: 361; Golden, 1998: 255; J.

W. Yackee & Yackee, 2006: 129).

High mobilization by business interests may be related to the saliency of policies for different

types of groups. Businesses participate in more rule making than other groups because they are

affected directly by more rules than other groups (Furlong & Kerwin, 2004: 361). High levels of

participation may in themselves lead to more influence as agencies take stock of the interest

advocated in consultations and adapt policies accordingly (J. W. Yackee & Yackee, 2006: 136). Also,

business groups are in many instances able to supply bureaucrats with relevant technical

information about proposed regulation and in return gain political influence (A. S. Binderkrantz,

Pedersen, & Christiansen, 2012; Bouwen, 2004). In the corporatist literature, the ability to affect

societal production has been singled out as a crucial resource possessed by both business groups

and trade unions (Rokkan, 1966). In contrast to US studies, this – mainly European – literature

emphasizes the role played by major trade associations rather than individual businesses. In

countries with a corporatist heritage – e.g. Denmark – the objects of interest are therefore the large

organized groups of businesses and their level of representation and success.

The results on whether business interests are indeed more successful than other groups are mixed.

Yackee and Yackee (2006: 135) conclude that: “Agencies appear to alter final rules to suit the

expressed desires of business commenters, but do not appear to alter rules to match the expressed

preferences of other kinds of interests”. In contrast, earlier US studies found no strong support to

traditional theories of capture although trade associations seemed to have some advantage

(Furlong, 1997: 340; Golden, 1998). Klüver (forthcoming) concludes in her study of EU policy

7

making that lobbying success does not vary systematically across group type. Conclusions about

the role of business thus seem to vary according to institutional setup and research design.

The role of money and other resources for lobbying success has also attracted considerable

attention. Even though recent research has shown that financial resources may be of less relevance

than usually assumed (Baumgartner et al., 2009), the issue is hardly settled. Rather than finances as

such it may be important how groups utilize their resources. Based on reanalysis of the data

originally compiled by Heinz and associates, McKay (2011: 13) concludes that: “While the general

picture shows little measurable effect of organizational wealth on group’s policy success, the data

suggest that how that money is spent can affect groups’ ability to get what they want”. Notably,

some groups dedicate resources to build up large secretariats capable of interacting professionally

with decision makers. This indicates emphasis on achieving political influence as well as

possession of resources to interact professionally with decision makers – for example by supplying

high quality comments in consultations (Furlong, 1997: 327; S. W. Yackee, 2005: 3).

H1: Business groups are more influential than other groups

H2: Groups with high levels of professional resources are more influential than other groups

Different policy characteristics, different chances of influence?

Lowi (1964: 688-89) famously argued that political relationships are determined by the type of

policy at stake. Policies must be defined in terms of their impact or expected impact on society, and

different policies lead to different patterns of mobilization and policy making. The most elitist

patterns of policy making were found in distributional policies, while regulatory and

redistributive policies were characterized by respectively pluralist and conflictual patterns of

interest mobilization (Lowi, 1964: 713). In a parallel argument, Wilson (1980) points to the

distribution of policies’ costs and benefits as a central policy variable. Groups are particularly

inclined to react on policies with concentrated benefits or costs. Theories of agency capture also

emphasize how agencies concerned with regulation of specific branches have been particularly

prone to capture (Golden, 1998; Stigler, 1971).

8

Regulatory policies satisfy – at least theoretically – some features that may attract the attention of

business groups: They are often related to the business sector and most often have concentrated

costs or benefits. The alleged high influence of business groups may therefore be related to the

type of policies they lobby on. In Golden’s (1998) study of consultations on eleven rules, variation

across policy areas was found with extremely skewed participation in regard to rules issued by

regulatory agencies. Here, business interests dominated and only few public interest or citizen

advocacy groups were found. Furlong (1997: 337-338) also includes variables capturing policy

types in his study of group influence on rule-making, but they turn out not to affect the perceived

effectiveness of groups.

A dimension related to the general discussion of policy areas is whether bills entail increased

spending or cuts on public expenditures. Lowi’s (1964) distributional policies were home to client

groups who could happily interact with agencies, while relationships were less amiable in areas

where specific groups were harmed by proposed regulation. Wilson (1980) hypothesizes that only

groups favored by policies with concentrated benefits or fighting policies with concentrated costs

will mobilize in favor of or against policies. McDermott (2004) points to the particular processes

related to losses. Due to the ‘negativity bias’ groups facing losses are particularly prone to

mobilize. Consequently interest groups are expected to mobilize in situations where government

proposals impose increased costs or reduced benefits on societal interests (Christiansen &

Nørgaard, 2003: 169). Since civil servants and ministers are familiar with and relatively resistant to

such protests, we expect them not to be very successful. In conclusion, we test a hypothesis related

to the policy area as such and one concerning the characteristics of the specific proposal:

H3: Influence is more likely in regard to specific regulation than other policy areas

H4: Influence is less likely in regard to bills involving increased state revenue or reduced expenses

Different patterns of mobilization, different patterns of influence?

A final set of factors relate to the pattern of mobilization in the consultation process. Several

authors have argued that characteristics of the issue context may be as important as or even more

important than group level characteristics (Klüver, 2011; Mahoney, 2007). One such factor relates

9

to the salience of a proposal in general or in regard to the number of groups mobilized (Nixon et

al. 2002: 68). S. W. Yackee (2006) has found that congressional attention constrains group influence,

while Klüver (2011: 497) focuses on the interests mobilized and finds that the number of

submissions received in consultations positively affects the likelihood of influence.

Interest groups may have very different overall opinions about proposed legislation. Some groups

are largely in favor, others are fierce opponents. Golden (1998: 61-2) argues that agencies tend to

favor supporters of its rules over critics. it is not very likely for a group who strongly opposes the

bill to have its – supposedly major – suggestions for revisions accepted. On the other hand, a

group who is generally in favor of the bill may be more effective in getting its suggestions

accommodated. The aggregate stance of participating groups in regard to proposals may also

matter. Consensus among commenters and repetition of a comment might increase the agency’s

probability of agreement (Nixon et al., 2002: 63). Mckay and Yackee (2007: 337) also find that the

balance of opinion matters as agencies are less likely to alter rules when lobbying is competitive,

and more likely when one side of a policy issue dominates the lobbying effort. However, their

study does not take into account the general stance of groups on the proposed legislation, but

examines in which direction proposals are changed. We combine the insight about the balance of

opinion with the argument that agencies are most likely to favor supporters. When bills are not

contested, agencies are more prone to accommodate comments because the stakes are smaller. We

propose three hypotheses about the effects of group mobilization:

H5: Influence is more likely when many groups respond to the consultation.

H6: Influence is more likely when opposition to the bill is low.

H7: Influence is more likely for groups who generally support the bill.

3. Research design

In this paper we focus on consultations about bills carried out by Danish ministries. In Denmark,

legislation is almost always prepared by the administration. With a few exceptions, a consultation

10

is routinely carried out before a bill is presented to parliament.2 Consultations are published online

and sent to a set of groups – mainly interest groups and public authorities – chosen by the ministry

in charge of the consultation. The lists of groups invited to respond are very inclusive and anyone

can reply to the consultation – if they find out about it. After the response date has passed the

ministry decides to what extent the proposed legislation will be changed. Also, a consultation

report containing all responses as well as the ministerial reaction to responses is prepared and

forwarded to the relevant parliamentary committee. This report is publicly available on the

parliament’s homepage.

3.1 Establishing a measure of group success

In order to obtain a relevant set of consultations, all bills introduced to parliament in its 2009-10

section – a total of 225 – have been registered. Consultation was carried out for 20, and two rounds

for six bills. For one consultation it was not possible to find the relevant material, and the

remaining 209 consultations related to 203 bills therefore form the basis of the analysis. Two of

these were postponed to later sessions; the rest were eventually passed by parliament.

The advantage of our strategy of case selection is that it encompasses all bills, which allows us to

include a wide range of policy areas and both high-salience and low-salience issues in the analysis

(S. W. Yackee, 2005: 8). One caveat is that some bills sent out for consultation may never make it to

parliament. After all, an often acknowledged effect of influence group activity is to hinder political

proposals in being put into effect (Baumgartner et al., 2009). This may happen regularly in some

political systems, but in Denmark it is very rare that bills are withdrawn once they have reached

the consultation stage. To check for occurrence, we visited the website ‘høringsportalen.dk’

(https://bdkv2.borger.dk/Lovgivning/Hoeringsportalen/Sider/Soeg.aspx), which registers all bills

sent for consultation in 2009-10 and established whether they were eventually proposed in

parliament. The number of bills sent out for consultation and not proposed to parliament makes

up less than a handful.

2. Main exceptions are financial bills, bills on tax agreements between Denmark and other countries, and bills on

naturalization.

11

Our methodology for studying influence exercised in consultations is similar to the one used in

studies of US administrative rule making (Golden, 1998; Nixon et al., 2002: 65; S. W. Yackee, 2005).

The unit of analysis is response by an interest group to a government consultation. For the 209

consultations we have coded all consultation responses from interest groups. Letters without

substantial content (for example just acknowledging receipt of the proposed bill) were not

registered. We found 1,691 responses sent by 415 different groups.

Establishing whether letters have led to changes in proposed bills is a time consuming process

(Furlong & Kerwin, 2004: 364). In constructing a measure on whether a group got what it wanted,

we were aided by a ministerial practice of drawing up a ‘hearing report’ detailing the ministry’s

responses to the consultation. For each letter, we have checked the report to establish whether the

group’s demands were met.

To establish the dependent variable – groups’ success – we registered whether a group’s proposed

changes were: 1) fully or almost fully followed by the ministry, 2) partly followed by the ministry

or 3) rejected. In the multivariate analyses we constructed a dichotomous measure, where groups

are assigned ‘1’ if their proposals were partly or fully accommodated and ‘0’ if rejected or in

opposition to the bill, but did not suggest specific amendments. In many cases the evaluation of

success was not straightforward as ministries sometimes strive to seem accommodating while

keeping the bill substantially unchanged. Therefore the coding process was difficult and the

instruction of the coders very detailed. In order to obtain reliable results we instructed the coders

to consult with the coding responsible in cases of doubt. We obtained intercoder-reliability

(Cohen’s Kappa) of .721 in a final sample of 66 cases.

3.2 Measures of independent variables

Group level variables

All interest groups responding to at least one consultation were given a unique ID number

allowing us to trace their pattern of activity across all consultations. Groups were coded into types

of groups including: 1) business groups, 2) trade unions, 3) professional groups (such as groups of

teachers, doctors or technicians), 4) associations of authorities and institutions, 5) identity groups

12

(such as patients, students or senior citizens), 6) hobby groups3, and 7) public interest groups. The

coding was based on group names and descriptions of groups on their websites. The coding was

done by the authors with a reliability test of 100 groups resulting in a Cohen’s Kappa of 0.906.

To obtain data on group resources, a survey was administered to all identified groups. Out of 415

groups active in consultations, 303 – corresponding to 73 percent – responded to the survey. This

was part of a larger survey among Danish interest groups and the 1,109 groups who responded to

the survey are used as a standard of comparison when we analyze the types of groups appearing

in the consultation data. The full questionnaire as well as the frequency distributions of group

answers may be found at: http://interarena.dk. The questionnaire included questions about group

resources including annual group income and number of employees working with politics broadly

(contacts to bureaucrats, politicians or reporters, conducting analyses and monitoring the political

process). These different measures exhibit a high level of multicollinearity (the tolerance level for

political employees is 0.225 when group income is included alongside the other variables in the

multivariate analysis of influence). Only the measure of employees working with politics is

therefore included in the analyses because personnel resources are supposed to represent a more

specific resource in relation to consultations than monetary resources. To obtain better linearity the

measure was logarithmically transformed.

Issue level variables

The policy area of all consultations was registered and the areas were recoded depending on

whether the area could best be classified as: 1) public production (health, education, culture,

defense etc.), 2) specific regulation (agriculture, labor market, housing etc.) or 3) general regulation

(macroeconomics, environment, foreign affairs etc.).

For each bill we also registered its public expenditure consequences. Ministries routinely report a

bill’s consequences for public finances. This information is rigorously checked by the Ministry of

Finance and may be seen as a rather reliable measure. Based on this we established a dichotomous

variable registering whether the bill included increased revenue or reduced expenses for the state

(including bills with both increased revenues and costs).

3. Only two religious groups appeared in the material. They were grouped with hobby groups.

13

Mobilization variables

Two variables were established based on the full set of responses to consultations. The first simply

registers the number of groups replying to the consultation to operationalize the salience of the bill

to the interest group community. The second measure registers the share of groups who were

generally against the bill.

The last variable combines the group and issue level in registering the group’s general stance on

the bill. This was coded based on group consultation responses. Responses were coded in the

following three categories: 1) The group overwhelmingly supports the bill (objections are

negligible), 2) The group supports the bill, but has significant objections and 3) The group is

overwhelmingly against the bill. Since this coding is open for judgment the coders were instructed

carefully and told to contact the coding responsible in any cases of doubt. In a final sample of 66

we obtained a Cohen’s Kappa of .65.

3.3 Methods

Two types of multivariate analyses are conducted in order to shed light on the factors affecting

group influence. Prior to these analyses all variables have been recoded to range from 0-1. First, we

perform an OLS regression of the success rate of each group across all consultations they

participated in. Second, we analyze group-issue dyads to investigate the factors affecting whether

a group is successful in a specific instance. Here, a logistic regression is performed. Because many

groups are active in the same consultation, robust clustered standard errors in regard to

consultations are used. The analyses have been repeated with similar results clustering the

standard errors in regard to unique groups.

4. Analyzing group participation in consultations

4.1 The mobilization of groups in consultations

How many groups and other actors respond when bills are sent for consultation? Are groups

typically engaged in only one or a few consultations or do they spread their attention more

widely? And which patterns of group mobilization can be found in different policy areas? These

questions can be addressed by descriptive analysis of the consultation data. Figure 1 gives an

14

overview of the extent of participation in consultations by national interest groups and by other

external actors.

Across the 209 consultations included in the study 2,169 replies were received, of which 1,692 –

corresponding to 78 percent – were from national interest groups. On average 10 actors replied to

each consultation and 8 of these were interest groups. Measured by the number of responses the

most salient issue was the proposed establishment of a new Center for Energy Savings. This

consultation attracted replies from 54 different actors of whom 32 were interest groups. At the

other end of the spectrum we find three consultations that did not attract a single reply.

Figure 1: The Distribution of consultation responses from interest groups and other actors

(numbers/consultation)

Compared to activity levels in other studies, the level we found appears low. In her study of 40

consultations held by US agencies, Yackee (2005: 8) found an average of 36 organized interests

participating although her definition of organized interests is broader than ours (S. W. Yackee,

2005: 8, 10). Klüver (forthcoming: 11) studied replies to 56 EU consultations and found almost 50

replies from associations per consultations. Lundberg’s (2012: 11) study of 33 consultations found

30 replies from voluntary organizations per consultation. The relatively low activity level may be

explained by the fact that the time frame for replying to consultations is often very short and that

15

the practice of holding consultations for almost all bills means that even bills of little substantial

interest to external actors are sent for consultation. The activity level in our study could also

indicate that our consultations represent an ‘insider’s game’, where a limited number of groups

seek to affect bills.

Interestingly, the dominance of interest groups in Danish consultations contrasts findings from a

longitudinal study of Swedish government consultations, where voluntary organizations

accounted for only 25.5 of replies to consultations held between 2000 and 2009 (Lundberg, 2012:

11). This demonstrates that even in countries with relatively similar overall institutional

arrangements for incorporating groups in decision making (Christiansen et al., 2009), variation

may be found depending on the specific setup of consultations. The Swedish consultations studied

by Lundberg were held on – often major – policy issues with a report prepared by a committee

being sent out for consultation (Lundberg, 2012). This seems to attract responses from a wider

circle of actors than the Danish consultations held in the last phase of administrative bill

preparation and on almost all bills.

Table 1: Number of consultations participated in, percentages

Number of consultations Number of groups Percentage

1 180 43.4

2 82 19.8

3 40 9.6

4 30 7.2

5 18 4.3

6-10 30 7.2

11-15 18 4.3

16-20 5 1.2

21-75 12 2.8

All groups 415 100

Among those who do respond, activity is rather unequally distributed. Across all consultations we

identified 415 unique groups and Table 1 shows the number of consultations these groups

participated in. It is most common to reply to just one consultation, which 43.4 percent of groups

16

did. About 20 percent participated in two consultations and a little less than 10 percent in three

consultations. Thus, more than 70 percent of all groups are active from one to three times. We also

find some rather active groups as about 8 percent are active in more than ten consultations and

almost 3 percent participate in more than 20 consultations.

The single most active group is the association of local communities ‘Local Government Denmark’,

who participated in no less than 75 consultations. The ‘Confederation of Danish Industry’ comes in

second with participation in 63 consultations followed by three other major trade associations.

Among the most active groups we also find ‘The Danish Confederation of Trade Unions’, ‘Danish

Regions’ and ‘Disabled Peoples Organization Denmark’. These groups are all well known political

players and have a significant presence in other political arenas (A. S. Binderkrantz et al., 2012).

Table 2: Group type and policy area, column percentages

Public

production

Specific

regulation

General

regulation All bills

‘Population’

as measured in

group survey

Business groups 22.1 56.7 46.1 41.0 25.3

Trade unions 25.5 18.8 16.3 20.3 14.2

Associations of

authorities and

institutions

26.3 6.8 9.6 14.9 6.3

Professional groups 1.8 1.2 3.9 2.4 11.5

Identity groups 15.9 4.6 7.6 9.8 14.4

Hobby groups 4.4 0.9 5.0 3.7 12.7

Public interest groups 4.1 7.9 11.5 7.8 15.6

Total 100.0 100.0 100.0 100.0 100.0

All groups 616 457 618 1,691 1,109

Note: Chi-square = 279.3, significant at the 0.001 level. Cramér's V = 0.2874

The more than 200 consultations span a wide variety of policy areas. Previous research has found

different patterns of group participation depending on the nature of the issue (Golden, 1998). Table

2 shows the pattern of group mobilization overall and in respect to policies related to public

production, specific regulation and general regulation. For each type of policy, groups are

distributed according to their type. As a standard of comparison, the table also shows the

17

distribution of groups in the survey of all national interest groups, which may be seen as a proxy

to the interest group ‘population’.

Overall, the data supports the pattern from US studies (J. W. Yackee & Yackee, 2006) that business

interests are very active participants. Across all consultations they account for 41 percent of replies.

Trade unions and associations of authorities and institutions are also rather active. Together, these

three types of labor market related groups account for more than 75 percent of all consultation

replies (compared to a share of 45 percent of the group ‘population’). All other group types carry

less numerical weight in the consultations than in the group ‘population’ as measured in the

survey. Professional groups and hobby groups are almost absent in consultations even though

they each represent about 12 percent of all groups. Identity groups and public interest groups fare

somewhat better with 10 percent and 8 percent of the replies respectively. Other studies have

found even lower levels of participation by public interest groups (Lundberg, 2012: 12; J. W.

Yackee & Yackee, 2006: 133), although Klüver (forthcoming: 25) reports that 20 percent of the

groups participating in EU consultations are cause groups.

The attention from different group types is not evenly distributed across policy areas. While

business groups are most present in specific as well as general regulation, institutional groups are

most active in bills concerning public production. These replies come mainly from two groups –

Local Government Denmark and Danish Regions – who organize the local and regional authorities

that represent two thirds of public consumption. We also find identity groups – typically users of

public service – to be active in regard to public production. Public interest groups are, on the other

hand, most often found to respond to hearings regarding general regulation such as environmental

issues. The patterns of activity are systematically different depending on policy issue as illustrated

by the significant chi-square test.

We cannot assume that all consultation replies carry equal weight. Notably, in some areas the

organization of interests is more concentrated than in others and umbrella organizations may

speak for a wide spectrum of underlying organizations. Another approach to comparing the

relative weight of different group types is therefore to see how many consultations each group

type collectively appears in (numbers not shown). Again, business groups come out as the most

dominant with at least one business group participating in 80 percent of consultations, 57 percent

18

of consultations have trade union participation and 51 percent have participation from at least one

association of authorities and institutions. Identity groups and public interest groups are present in

respectively 32 and 39 percent of consultations, while professional groups and hobby groups are

only active in 15-17 percent. Overall, this matches the pattern found when we analyze the relative

percentages of replies.

4.2 How groups evaluate bills – and what they get

Groups may have very different overall opinions about proposed legislation. Some see bills as

largely positive even though they may have specific reservations; others are fierce opponents and

seek to have a bill altered considerably or even withdrawn. Table 3 illustrates how this plays out

across different types of groups as all replies have been coded as either ‘supportive’, ‘somewhat

supportive’ or ‘opposing’.

Table 3: Group type and evaluation of bill, row percentages

Supportive Somewhat supportive Opposing N

Trade unions 30.0 51.3 18.7 343

Business groups 32.9 57.3 9.8 693

Associations of authorities and institutions 30.3 57.4 12.4 251

Professional groups 27.5 47.5 25.0 40

Identity groups 12.8 66.5 20.7 164

Hobby groups 12.9 69.4 17.7 62

Public interest groups 15.9 59.1 25.0 132

All groups 27.8 57.3 14.9 1,685

Note: Chi-square = 71.7, significant at the 0.001 level. Cramér's V = 0.1459

Different group types generally have different opinions about the set of bills sent for consultation.

The chi-square test shows that the two variables are associated. Business groups stand out as the

most positive. In less than 10 percent of the cases, replies from business groups were in opposition

to the bill proposed. Associations of authorities and institutions are also rather positively inclined.

The fiercest opponents of government bills are found among public interest groups with 25

percent against and only 16 percent in favor of proposed bills. Professional associations and

identity groups are also among the groups who are least positive towards proposed legislation.

19

This pattern may reflect that the government at the time of data collection was Liberal-

Conservative and thus traditionally pro-business. Another reason may be that some of the positive

groups have been included in the policy preparation at an earlier stage of the policy process and

thus been given early policy concessions.

Some groups simply state their general view of the legislation and advise the government to either

promote or drop the proposed bill. Others provide more specific suggestions for change. Table 4

shows the distribution of all consultation replies according to whether they contained specific

suggestions and if so whether the suggestion was fully met, partly met or neglected.

Table 4: Group type and accomodation, row percentages

No

suggestions

Full

accommodation

Partial

accommodation Neglected N

Trade unions 27.7 8.5 25.1 38.8 343

Business groups 24.4 8.2 31.3 26.1 693

Associations of authorities

and institutions

31.0 7.5 24.2 37.3 252

Professional groups 25.0 12.5 30.0 32.5 40

Identity groups 18.8 4.9 28.5 47.9 165

Hobby groups 21.0 12.9 35.5 30.7 62

Public interest groups 21.2 3.8 28.0 47.0 132

All groups 25.1 7.8 28.6 38.5 1,687

Note: Chi-square = 31.3, significant at the 0.05 level. Cramér's V = 0.0786

First, we notice that about 25 percent of the groups do not have specific suggestions. For groups

that oppose the bill this means that their suggestion was simply to drop the bill and we can

conclude that they were unsuccessful since all of the analyzed bills were eventually put forward to

the legislature. In contrast we cannot make any conclusions regarding the success of groups that

made no suggestions but were positively inclined towards the bill, since there is no way for us to

link the support of a specific group to the advancement of the bill.

Second, it is interesting that change does happen. In about 8 percent of the cases a group makes a

suggestion that is fully accommodated and in almost 30 percent of the cases group suggestions are

partly accommodated. US studies report ample evidence of change taking place. Yackee (2005: 3)

20

concludes that: “the notice and comment period is an important political arena where the

bureaucracy frequently alters and adapts public polices to better match the preferences of interest

group commenters”. The present study indicates that bills are indeed altered in the Danish

consultation process, although the magnitude of change may not always look impressive.

Third, there is variation across different types of groups (although the magnitude is not impressive

as indicated by the statistical test). Business groups and hobby groups have the highest success

rate when it comes to suggestions not being neglected. Identity groups and public interest groups

are less likely to have their wishes accomodated. The types of groups that are generally most

positive towards proposed bills are thus also the most influential.

4.3 Explaining influence: Group type, resources or contextual factors?

The next step involves testing the hypotheses about influence. The issue of influence may be

analyzed in different ways. One strategy is to ask which groups are most successful across the

range of bills they seek to influence. Another strategy is to see each consultation reply as an

attempt to affect a bill and seek to explain whether such an attempt succeeds. Both approaches will

be applied here. In both analyses we define success as having a proposal fully or partly

accomodated. Lack of success is defined as having a proposal neglected or being in opposition to a

bill that is eventually put forward to parliament.

The most active participants are – not surprisingly – also the most succesful in terms of affecting a

large number of bills. For example, in 35 instances Local Government Denmark’s suggestions were

fully or partly accommodated. Other groups have higher success rates, but based on less activity.

Quite a few groups have a success rate of 100 percent but most participated only once. Here, two

groups representing major economic interests – The Danish Shipowners Association and the

Danish Mortgage Banks' Federation – stand out: Both participated six times and in all instances

their suggestions were at least partly accommodated.

Table 5 shows the result of a set of multivariate analyses. The first column presents the analysis of

success rates. Here, unique groups are the units of analysis and only variables at the group level

are incorporated. The two other columns present an analysis of group success in individual

consultations with group bill dyads as units of analysis. This allows us to include issue and

21

consultation related variables alongside group level variables in the analysis. The dependent

variable is dichotomous with ‘1’ representing at least partial accommodation and ‘0’ representing

neglect. Results are shown in table 6, which reports coefficients and robust standard errors

clustered on unique groups. The analysis is conducted in two steps and the first step includes only

group level variables.

Table 5: Multivariate analysis of succes (OLS and Logistic regression)

Overall

success rate

Case-wise success

Model I

Case-wise success

Model II

Group

factors

Business groups Ref. Ref. Ref.

Trade unions -0.015 (0.677) -0.363* (0.220) -0.038 (0.221)

Associations of authorities

and institutions

-0.159* (0.079) -0.225 (0.219) 0.081 (0.260)

Professional groups 0.022 (0.157) -0.066 (0.495) 0.049 (0.642)

Identity groups -0.082 (0.074) -0.288 (0.260) 0.020 (0.274)

Hobby groups 0.133 (0.102) 0.223 (0.397) 0.581 (0.381)

Public interest groups -0.077 (0.084) -0.520* (0.284) -0.269 (0.303)

Political employees 0.032 (0.130) 0.394 (0.279) 0.411 (0.308)

Policy

factors

Public production Ref.

Specific regulation 0.519* (0.271)

General regulation 0.204 (0.283)

Expenditure cuts -0.531* (0.216)

Mobilization

factors

Number of responses 1.404** (0.471)

Share of opposing groups -2.835*** (0.620)

Group positive Ref.

Group somewhat positive -0.375 (0.118)

Group opposing -0.967** (0.311)

Constant 0.476*** (0.054) -0.144 (0.196) -0.186 (0.335)

R2/Pseudo R2 0.007 0.009 0.106

N 280 1,106 1,105

Note: Levels of significance: *=0.1, **=0.01, ***=0.001. Standard errors are clustered with respect to

consultations.

22

Group level factors have little leverage in explaining success in consultations. In the analysis of

success rates the only significant effect found is that associations of institutions and authorities are

less successful than business groups, while the first step of the case-by-case analysis shows that

trade unions and public interest groups are less likely to achieve their goals than business groups.

In both models little variation is explained by the group-level factors included, none of the group

level variables are significant when policy related variables are included. We cannot confirm the

two hypotheses (H1 and H2) related to group level factors.

The most important variables all relate to the characteristics of and the mobilization of groups in

the specific consultation. First, more influence is attained in policy areas related to specific

regulation (H3). Second, influence is more common when bills do not aim to increase public

revenue or cut public expenditures (H4). Third, many responses to a consultation make influence

more likely (H5). Fourth, if a large share of groups is in opposition to a bill influence is less likely

(H6). Fifth and finally, groups who are negatively inclined towards a bill are less likely to affect its

content (H7). All hypotheses about factors related to an issue or mobilization of groups are thus

supported by the analyses.

Figure 2: Predicted probabilities of success (all other variables kept at mean)

0

0,1

0,2

0,3

0,4

0,5

0,6

23

In terms of the magnitude of effects, the predicted probabilities of success for the categorical

variables are illustrated by Figure 2, while Figure 3 illustrates the effect of number of groups

responding to the consultation (with all other variables kept at their mean). The difference between

being active in public production and specific regulation amounts to moving from a predicted

success rate of 0.39 to one of 0.52. If groups lobby in instances of budget cuts (or revenues for the

state raised) their chance of success is predicted to be 0.34, compared to 0.47 in other situations.

The largest difference is found in the overall stance on the bill, where supportive groups have a

predicted success rate of 0.55, while opponents are successful in only 0.32 of the cases. In Figure 3

we see the difference between being the only active group and being one among many. Over the

full spectrum of level of activity the predicted probability goes from 0.31 to 0.65, i.e. the chance of

success doubles. The effects of the remaining variables are generally lower than of those illustrated

here.

Figure 3: Predicted probabilities of success with increased levels of group participation (all other

variables kept at mean)

Note: The bars illustrate the 95 percent confidence interval

.2.4

.6.8

Pre

dic

ted p

roba

bili

ties

0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1Number of responses

24

5. Conclusion

Now, what do these findings tell us generally about influence in consultations? A main conclusion

is that group level variables are not as important for influence as sometimes assumed. Among

those who participate in consultations business groups are not more likely than others to achieve

their goals and wealthy groups are not more influential than others. This does not mean that group

type and resources are inconsequential. Across all consultations a business group accounted for 41

percent of the replies and in matters related to specific regulation the share of business replies was

even .57. Even though they were not more likely than other groups to get what they wanted in any

specific consultation the high level of business mobilization means that business views were

accommodated in more cases.

The results fit a description of Danish bill consultations as an insider’s game. The level of

participation in consultations is moderate compared to some other countries and the main

participants are well known groups representing major societal interests. We find no evidence that

success is related to large-scale mobilization of opposition to bills. Rather, groups who are

generally positive towards bills but have suggestions for improvements are able to affect the

content. And, if this is combined with interest in the bill from other groups, influence is more

likely. It is also more likely that groups are influential in regard to bills on specific regulation that

do not propose cuts in public expenditure.

The study has focused on one specific stage of the policy process. The picture of influence can only

be partial. Much may have been won prior to a bill being sent for consultation, and changes also

happen after the bill is introduced to parliament. However, we are confident that what we capture

is indeed an important part of politics. When bills are sent out for consultations, groups reply and

the fact that major interest groups are among the most active participants testifies to the

importance these groups attach to being active in consultations. Also, group comments do make a

difference as accommodations are made in response to many of the comments received. These may

be of varying importance, but there is clearly something to be gained from consultation

participation.

In comparing the results with other studies we find many differences. Fewer groups reply to an

average Danish bill consultation than to US consultations on administrative rule-making, EU

25

Commission consultations and Swedish consultations on public white papers. Also the pattern of

mobilization differs, as organized interests are more dominant in Danish consultations (Klüver,

2011; Lundberg, 2012; J. W. Yackee & Yackee, 2006). While this may be connected to overall

differences in group populations, institutional arrangements for incorporating groups into public

decision making, and the type of objects that groups react to, it is also likely to be related to the

specific institutions governing the consultations. Consultations held at different stages of the

policy process and with different set-ups for replying are likely to lead to different patterns of

mobilization. From a normative perspective this is encouraging because it points to the options

politicians and bureaucrats have to induce more diverse participation by shaping institutional

arrangements.

26

6. References

Baumgartner, F. R., Berry, J. M., Hojnacki, M., Kimball, D., & Leech, B. L. (2009). Lobbying and policy

change. who wins, who loses, and why. Chicago: The University of Chicago Press.

Baumgartner, F. R., & Leech, B. L. (1998). Basic interests. the importance of groups in politics and in

political science. Princeton, N.J.: Princeton University Press.

Baumgartner, F. R., & Leech, B. L. (2001). Interest niches and policy bandwaggons: Patterns of

interest group involvement in national politics. The Journal of Politics, 63(4), 1191-1213.

Bernhagen, P. (2011). Who gets what in british politics - and how? an analysis of media reports on

lobbying around government policies, 2001-7. Political Studies, , 1-21.

Beyers, J. (2004). Voice and access. political practices of european interest associations. European

Union Politics, 5, 211-240.

Binderkrantz, A. (2005). Interest group strategies: Navigating between privileged access and

strategies of pressure. Political Studies, 53, 694-715.

Binderkrantz, A. S., Pedersen, H. H., & Christiansen, P. M. (2012). Accessing political arenas.

interest group access to the administration, parliament and media. Paper,

Bouwen, P. (2004). Exchanging goods for access: A comparative study of business lobbying in the

european union institutions. European Journal of Political Research, 43, 337-369.

Christiansen, P. M., & Nørgaard, A. S. (2003). Faste forhold - flygtige forbindelser. stat og

interesseorganisationer i Danmark i det 20. århundrede . Aarhus: Aarhus Universitetsforlag.

Christiansen, P. M., Nørgaard, A. S., Rommetvedt, H., Svensson, T., Thesen, G., & Öberg, P. (2009).

Varieties of democracy: Interest groups and corporatist committees in scandinavian policy

making. Voluntas, 21, 22-40.

Dür, A. (2007). The question of interest group influence. Journal of Public Policy, 27, 1-12.

27

Furlong, S. R. (1997). Interest group influence on rule making. Administration & Society, 29, 325-344.

Furlong, S. R., & Kerwin, C. M. (2004). Interest group participation in rule-making: A decade of

change. Journal of Public Administration Research and Theory, 15(3), 353-370.

Golden, M. M. (1998). Interest groups in the rule-making process: Who participates? whose voices

get heard? Journal of Public Administration Research and Theory, 8, 245-270.

Halpin, D., & Binderkrantz, A. S. (2011). Explaining breadth of policy engagement: Patterns of

interest group mobilization in public policy. Journal of European Public Policy, 18(2), 201-219.

Klüver, H. (2009). Measuring interest group influence using quantitative text analysis. European

Union Politics, 10(4), 535-549.

Klüver, H. (2011). The contextual nature of lobbying: Explaining lobbying success in the european

union. European Union Politics, 12(4), 483-506.

Klüver, H. (Forthcoming). Biasing politics? interest group participation in european policy-

making. West European Politics.

Lindblom, C. (1977). Politics and Markets: The World's Political Economic Systems. New York: Basic

Books, 1977.

Leech, B. L. (2010). Lobbying and influence. In S. L. Maisel, & J. M. Berry (Eds.), The oxford handbook

of american political parties and interest groups (pp. 234-567).

Lowery, D., Gray, V., & Fellowes, M. (2005). Sisyphus meets the borg. economic scale and

inequalities in interest representation. Journal of Theoretical Politics, 17(1), 41-74.

Lowi, T. J. (1964). American business, public policy, case-studies, and political theory. World

Politics, 16(4), 677-715.

Lundberg, E. (2012). Changing balance: The participation and role of voluntary organisations in

the swedish policy process. Scandinavian Political Studies, Online.

28

Mahoney, C. (2007). Lobbying success in the united states and the european union. Journal of Public

Policy, 27(1), 36-56.

McDermott, R. (2004) Prospect Theory in Political Science: Gains and Losses From the First

Decade, Political Psychology 25(2): 289–312.

McKay, A. (2011). Buying politics? The effects of lobbyists' resources on their policy success.

Political Research Quarterly, Preview.

McKay, A., & Yackee, S. W. (2007). Interest group competition on federal agency rules. Ameritcan

Politics Research, 35, 336-357.

Nixon, D. C., Howard, R. M., & DeWitt, J. R. (2002). With friends like these: Rule-making comment

submissions to the securities and exchange commission. Journal of Public Administration

Research and Theory, 12(1), 59-76.

Rokkan, S. (1966). Norway: Numerical democracy and corporate pluralism. In R. A. Dahl (Ed.),

Political oppositions in western democracies (). New Haven: Yale University Press.

Schattschneider, E. E. (1975 [1969]). The semisovereign people. A realist's view of democracy in america.

US: Thomson Learning.

Schlozman, K. L., Verba, S., & Brady, H. E.,. (2012). The unheavily chorus. unequal political voice and

the broken promise of american democracy. Princeton University Press: Princeton.

Long, J. S. & J. Freese (2006). Regression Models for Categorical Dependent Variables Using Stata,

Second Edition. College Station, Texas: Stata Press.

Stigler, G. J. (1971). The theory of economic regulation. The Bell Journal of Economics and Management

Science, 2(1), 3-21.

Wilson, James Q. (1980). The Politics of Regulation, pp. 357-447 i James Q. Wilson (ed.). The

Politics of Regulation. New York: Basic Books.

29

Yackee, J. W., & Yackee, S. W. (2006). A bias towards business? assessing interest group influence

on the U.S. bureaucracy. The Journal of Politics, 68(1), 128-139.

Yackee, S. W. (2005). Sweet-talking the fourth branch: The influence of interest group comments on

federal agency rulemaking. Journal of Public Administration Research and Theory, 16(1), 103-124.

Yackee, S. W. (2006). Assessing inter-institutional attention to and influence on government

regulations. British Journal of Political Science, 36, 723-744.