The survival of interest groups: evidence from Germany

Abstract

Interest groups are important intermediary organisations that function as a transmission belt between societal interests and political decision-makers. However, while some interest groups survive over decades, others only last a few years. This article argues that the survival of interest groups depends on their ability to mobilise resources which is crucially affected by interest group type and the public salience of an interest group’s policy domain. The theoretical expectations are tested based on a novel dataset mapping the survival of 1699 interest groups registered at the German Bundestag between 1974 and 2012. Using event history analysis, it is shown that interest group type and public salience indeed affect whether interest groups survive. Sectional groups last significantly longer than cause groups, and interest group survival increases with the public salience of their policy area. The results have major implications for our understanding of interest groups and political representation in contemporary democracies.

Keywords:

• Interest groups
• cause groups
• sectional groups
• survival analysis
• interest representation

Why does interest group survival vary across groups? Interest groups are important intermediary organisations that function as a transmission belt between societal interests and political decision-makers. Interest groups aggregate and articulate both the interests of individual citizens but also the interests of corporate and collective actors such as firms and public institutions. Interest groups are therefore an important vehicle for the pluralist representation of all sorts of societal interests. Moreover, interest groups ensure a long-term interest intermediation through formally organised structures. However, while some interest groups represent societal interests over decades, others only last a few years.

As effective interest representation crucially depends on the existence of an organised interest group that lobbies decision-makers on behalf of its constituents, it is highly problematic for the representation of societal interests if their interest associations are unstable and dissolved shortly after being established. At the same time long-term survival could also be problematic for democracy if interest groups representing powerful economic interests not only have advantages in mobilisation, but also with regard to survival, as this could lead to a lock-in of the representation bias. Given that interest groups play a fundamental role in Western democracies as they importantly bridge the gap between society and political decision-makers, it is crucial to understand why some groups last longer than others. Is there for instance a bias towards powerful economic interests? Are interest groups representing powerful economic actors more stable than, for example, NGOs?

A number of scholars have focused on explaining why interest groups organise and mobilise and why the density of interest group systems varies across sectors and across countries (e.g. Berkhout et al. 2015; Gray and Lowery 1996b; Leech et al. 2005; Lowery and Gray 1995; Rasmussen and Carroll 2014). These studies have made an important contribution to our understanding of interest group survival by studying, amongst others, the mortality rate of interest groups at the system level (e.g. Gray and Lowery 1995, 2001). They found that rates of demise are positively affected by interest group density while rates of demise decrease as party competition intensifies. A related line of research investigates interest group mortality anxiety. Gray and Lowery (1997) studied mortality anxiety in the US states and found that mortality anxiety decreases as interest groups rely more strongly on internal sources of revenue, when membership size is large and when policy areas are less conflictual while anxiety increases when interest groups face strong competition. Halpin and Thomas (2012) examined interest group mortality anxiety drawing on a survey of Scottish organisations and found that competition by other groups and shrinking membership size lead to higher anxiety levels while resources and the number of domains covered by groups lead to lower levels of mortality anxiety.

There are only a handful of studies that directly explore the determinants of interest group survival on the level of individual interest groups. Based on a comparison of Danish interest group populations at three time points, Fisker (2015) studied the factors that explain why interest groups have survived or died between 1976 and 2010 and between 1993 and 2010. She found that resources and a privileged position in the decision-making process have a positive effect on interest group survival. Nownes and Lipinski (2005) studied the survival of gay and lesbian rights interest groups in the United States and found that interest group survival is negatively affected by interest group density. Bevan (2013) analysed the survival of voluntary organisations in the US and found that groups that seek political influence last longer than other voluntary organisations and that group resources, professional membership, advocating for a social cause and government attention are associated with longer survival while public salience decreases survival.

This article contributes to this emerging literature and provides a comprehensive analysis of interest group survival that is based on a longitudinal analysis of 1699 interest groups registered at the German Bundestag between 1974 and 2012. Germany is an interesting case due to its corporatist interest intermediation patterns that may lead to comparatively high aggregate levels of survival in comparison to the pluralist US system. It is argued that resource mobilisation is a crucial determinant of interest group survival. More specifically, it is hypothesised that the survival of interest groups depends on their ability to mobilise resources, which is crucially affected by interest group type and the public salience of the policy area in which they are active. Using event history analysis, it is shown that interest group type and public salience importantly determine whether interest groups survive. Sectional groups last significantly longer than cause groups and interest group survival increases with the public salience of the policy area in which interest groups are active.

The results have major implications for our understanding of interest groups and political representation in contemporary democracies. Shedding light on which groups are sustainable in the long run is central for our understanding of representative democracy. Interest groups function as important intermediary organisations that aggregate and articulate the concerns of societal groups towards political institutions. Interest group lobbying ensures that societal interests are heard by politicians so that policy-making is not insulated from the concerns of affected societal groups. However, an important condition for effective interest representation is the long-term survival of interest groups as gaining access and obtaining the trust of policy-makers is a process that simply takes time. As a result, investigating why interest group survival varies and whether there is potentially a bias favouring powerful economic interests is important for understanding which societal interests are effectively represented before government.

Resource mobilisation and interest group survival

According to Beyers et al. (2008: 1106–9), three features must be given to define an actor as an interest group: organisation, political interest and informality. In order to qualify as an interest group, political actors must draw on some sort of organisation, actors must pursue the objective to influence political decision-making (political interest) and interest groups do not compete in elections (informality). I add a fourth characteristic that distinguishes interest groups from companies: Interest groups are organisations that rely on constituents for their survival. Constituents can be individual citizens, companies or institutions that share a common policy objective and provide resources to interest groups. It has to be noted that constituents are not necessarily formal members of interest groups, but they can also merely support interest groups by providing resources.

I assume that the basic concern of interest groups is survival, the most fundamental objective of organisations (e.g. Gray and Lowery 1996a; Lowery 2007). All other goals of interest groups are secondary since survival is the precondition for achieving any of the other objectives. Since constituents are their main resource providers, interest groups compete for members to extract from them adequate resources to ensure their survival (McCarthy and Zald 1977; Schmitter and Streeck 1999). Constituents can be individual citizens or other organisations such as companies, public authorities or other associations. I expect that the ability of interest groups to mobilise resources from supporters is crucially affected by the public salience of the policy they are working in and by interest group type.

Public salience

An important reason why individuals or organisations establish and support interest groups is interest representation. Individual or collective members delegate the representation of their interests to associations which lobby political institutions in order to realise their political interests. Members therefore expect that their interest groups influence legislators so that the policy outcome is as close as possible to their policy preferences. For instance, individuals become members of an environmental group since the protection of the environment is important to them. They provide resources such as financial contributions or labour to the environmental group and expect in exchange that it lobbies legislators in order to improve environmental protection.

Gray and Lowery argue that the mere existence of potential constituents does not necessarily imply that they are willing to equip interest groups with valuable resources (Gray and Lowery 1996b; Lowery and Gray 1995). Even though citizens or firms might deem a policy area very important, they might be satisfied with the status quo and thus do not have any incentives to invest in costly lobbying. It is therefore necessary that there is a latent group of potential constituents that share a common objective, but it is also important that this interest is salient to potential supporters. I accordingly expect that the public salience of the policy area in which interest groups are active importantly affects the ability of interest groups to mobilise resources. If public salience of a policy area is very high, potential constituents are more willing to invest resources for interest group lobbying in that policy area (see also Bevan 2013: 549). In times of high public salience of a policy area, it is likely that governments will respond to the demands in one way or another (Hobolt and Klemmensen 2008; Klüver forthcoming; Wlezien 1995). In such a situation, supporters are willing to provide more resources to interest group lobbying either because they want to defend the status quo in times of heightened attention or they want to seize the window of opportunity and promote policy change. Thus, I expect that the ability of interest groups to mobilise resources to sustain themselves is crucially affected by the public salience of the policy area they are lobbying in.

Hypothesis 1: Interest group survival is positively affected by the salience of the policy area in which interest groups are active.

Interest group type

Interest group scholars distinguish between different types of interest groups that vary in their collective action potential (e.g. Dür and De Bièvre 2007; Giger and Klüver 2016; Klüver 2012; Offe 1969; Olson 1965). Researchers employ varying definitions to capture differences in interest group type such as public vs. private interest groups or specific vs. diffuse interest groups. I distinguish between cause groups and sectional groups (Giger and Klüver 2016; Klüver 2012; Stewart 1958). Cause groups are interest groups which fight for a general belief or principle such as environmental protection or human rights. Anyone who supports that belief can join the group as membership is not restricted. The supporters of cause groups are typically individual citizens and the nature of the interests that cause groups represent is diffuse; that is, their policy goals are only associated with diffuse costs and benefits. Sectional groups by contrast represent the special interests of a specific segment of society, such as the interests of pharmacists, electricity companies or workers in the automobile industry. Their purpose is to represent the special interests of this particular sector and their membership is usually limited to actors in that sector.

Sectional groups represent well-defined homogeneous constituencies with concentrated costs and benefits. Since these interests are typically of primary material concern to their constituents, they are willing to invest money and time in establishing and maintaining an interest group which defends their common interests (Offe and Vale 1972: 86). The benefits of collective action are higher than its costs and potential constituents of sectional groups therefore face strong incentives to engage in collective action. In addition, as sectional groups typically represent economic sectors such as the automobile industry or farmers, potential constituents of sectional groups are on average well-equipped with resources so that they can easily afford to support an interest group that represents their interests (see also Dür and De Bièvre 2007). By contrast, cause groups are not related to the material needs of a small and homogeneous group of constituents, but to diffuse benefits of a large heterogeneous group of citizens (Offe and Vale 1972: 86). The costs of engaging in collective action are therefore higher than its benefits and potential constituents have little incentive to invest resources in lobbying. Moreover, cause groups much more fiercely compete against each other for members since their membership structure is not clearly circumscribed as it is for sectional groups. Their membership is therefore more fluid and the flow of resources more unstable. In addition, unlike constituents of sectional groups which are usually corporate actors that are well-equipped with resources, citizens have on average only a limited amount of resources at their disposal which makes it difficult to spend a lot of money on lobbying. In conclusion, given that the incentives to engage in collective action and the resources available to supporters are considerably higher for potential constituents of sectional groups than for supporters of cause groups, I expect that sectional groups find it considerably easier to obtain resources from their supporters than cause groups.

Hypothesis 2: Sectional groups on average survive longer than cause groups.

Research design

Case selection

In order to study interest group survival, I have selected Germany for analysis for the following reason: the German Bundestag lobbying register was established as early as 1974 and it annually maps the population of interest groups up to the present. According to the rules of regulation of the Bundestag, interest groups can only participate in hearings and obtain a door pass to enter the parliament if they are registered (Deutscher Bundestag 2014: 111). The Bundestag register is therefore an ideal data source to trace the survival of interest groups as it is reasonable to expect that all national-level interest groups have strong incentives to register at the Bundestag while organisations that are only politically active at the regional or local level are excluded. In addition, Germany is an interesting case due to its corporatist interest intermediation patterns that make it very different from the pluralist US system where most of the previous research on interest group survival has been conducted. The corporatist structures in place in Germany may lead to comparatively high aggregate levels of survival as access provision by policy-makers in favour of encompassing groups or government recognition for certain services should enhance the survival of recipient groups over the longer term. It is therefore important to shed light on the determinants of interest group survival in a corporatist setting like Germany.

Dependent variable

In order to study interest group survival, the annually published interest group register was converted into a longitudinal dataset of all interest groups that registered at the Bundestag between 1974 and 2012. I encountered the phenomenon that sometimes different groups merged and in those cases, these groups have been treated as a new interest group reflecting the organisational change that took place.¹ To measure the duration of interest group survival, I obtained the year in which an interest group was established and the year when the interest group last registered in the Bundestag lobby list. Based on this information, I computed the age in years to measure the duration of interest group survival. The age of interest groups varies substantially. The mean age is 45 years, with a standard deviation of 35 years (see Figure 1; and Table A1 in the online appendix). If we distinguish between cause and sectional group, the average age amounts to 42 years for cause groups and 48 years for sectional groups.

Figure 1. The survival of interest groups in the German Bundestag.

Table 1. Results from Cox regression.

	Model 1	Model 2	Model 3
Main
Cause groups	0.381***	0.443***	0.301**
	(0.111)	(0.138)	(0.119)
Public salience	‒0.005*	‒0.010***	‒0.006***
	(0.003)	(0.003)	(0.002)
Government activity		0.000**	‒0.000
		(0.000)	(0.000)
No. of competitors		0.001**	0.001
		(0.001)	(0.001)
No. of new competitors		0.229	‒1.079
		(0.186)	(0.806)
Reunification			4.271**
			(1.891)
TVC × ln(t)
No. of competitors × ln(t)			‒0.000
			(0.000)
No. of new competitors × ln(t)			0.316
			(0.209)
Reunification × ln(t)			‒0.476
			(0.508)
N	25,666	25,666	25,666
Subjects	1,699	1,699	1,699
Failures	339	339	339
Log pseudolikelihood	‒2249	‒2242	‒2139

*** p ≤ 0.01; **p ≤ 0.05; *p ≤ 0.10; Clustered robust standard errors in parentheses.

To measure the year in which an interest group was founded, I made use of several data sources. First, I relied on information retrieved from the websites of interest groups. Second, if there was no information about the founding year on the interest group homepages, a more extensive internet search was conducted primarily relying on Wikipedia and online public affairs directories to identify the founding year of an interest group. Finally, if none of these searches was successful, interest groups were phoned and asked about the year in which their group was established. Altogether, I was able to measure the age and trace the Bundestag registration for 1699 interest groups over the period 1974 until 2012.²

Given that the independent variables change over time, the interest group survival data was recorded annually in the time period of investigation to allow for state changes (Box-Steffensmeier and Zorn 2001). An interest group is accordingly included in the dataset as many times as it registered throughout the 1974 to 2012 period. If an interest group was still registered in 2012 when the register was last coded, it was coded as survival. If an interest group was no longer registered, it was coded as non-survival. Overall, about 20% of all interest groups in the dataset were dissolved during the time period of investigation. Of all the groups that are still in place at the end of our period of analysis (2012), the mean age is 47 years (standard deviation is 36 years) ranging from a minimum of 1 to a maximum of 188 years.

All interest groups included in the dataset were classified into policy areas according to the Policy Agendas Topic Codebook to identify the policy area in which interest groups are active. The Topic Codebook was originally developed by the US Policy Agendas Project (Baumgartner and Jones 1993) and later updated in the framework of the Comparative Agendas Project which maps the policy agenda by manually coding policy documents into policy areas. I relied on a revised version of the codebook designed by Breunig (2013) which takes into account the specificities of the German political system and I moreover slightly adjusted that codebook for interest group data. Altogether, interest groups were classified into 22 policy areas (see Table A2 in the online appendix).³ Human coders classified interest groups into policy areas based on information about their interests and activities that the interest groups provide in the lobbying register and by additionally relying on information retrieved from interest group websites and from phone calls with interest groups. It has to be noted that even though we observe interest groups over several years, none of the interest groups changed its policy profile in such a way that necessitated a different policy agendas code. Reliability tests indicate high levels of correspondence between different coders as the Krippendorff alpha amounts to 0.93 for the major topic codes and 0.80 for the subtopic codes.⁴

Independent variables

In order to test the first hypothesis postulating a positive effect of public salience on interest group survival, I rely on data on the policy priorities of citizens obtained from a monthly public opinion poll called Politbarometer. Ever since 1986, the Politbarometer has included a question that asks respondents about the most important problem which I used to measure the public salience of a policy area. The most important problem question (MIP) is a standard tool for measuring the relative importance of issues to voters. It has to be noted, however, that the MIP question is only a crude proxy for the relative importance of a policy area and particularly does not provide precise information about the importance of an interest group to its constituents. To match the public opinion data with the interest group dataset, the issues named by survey respondents were also classified into policy areas according to the extended version of the Policy Agendas Topic Codebook (see Table A2 in the online appendix). The percentage of respondents who named a major policy area as important serves as a measure of the public salience of that policy domain. Figure A1 illustrates how public salience varies across policy areas and over time. Macroeconomics is by far the most important policy area, followed by civil rights and social welfare. International politics and environmental protection are also policy areas that are increasing in importance, while reunification dominated the agenda in the early 1990s and other policy areas become important at times (e.g. finance during the recent financial crisis).

To test whether interest group type explains how long interest groups survive, interest groups were classified as either ‘cause’ or ‘sectional groups’ (Stewart 1958) based on the nature of the interest they represent and on their membership structure. I relied on information about the nature of the interest and the membership structure that was provided in the lobbying register and on information that was retrieved from interest group websites. Figure A2 in the online appendix provides an overview of the number of cause and sectional groups across policy areas over time. While some policy areas are more densely populated by cause groups (e.g. civil rights, international politics, social welfare) sectional groups are more numerous in other policy fields (e.g. macroeconomics, finance) while in still other policy domains there is not much difference (e.g. labour/immigration, defence and foreign trade). The correlation amounts to 0.58.

Figure 2. The effect of public salience.

As alternative explanatory factors may potentially confound the hypothesised relationships, I include the following control variables in the model. First, I control for government activity, as Lowery and Gray (1995) and Gray and Lowery (1996b) show that interest group density is crucially shaped by government activity. Government activity is operationalised by relying on data on public expenditure (see also Soroka and Wlezien 2010; Wlezien 1995; Wlezien and Soroka 2012). I rely on the public expenditure data released by the German Federal Ministry of Finance which corresponds to the percentage expenditure of all federal public spending. This data covers all public expenditures at the federal level which are directly controlled by the German federal government. The correlation between government spending and public salience only amounts to 0.35 so that they are only moderately related.

Second, I control for the number of competitors and the number of new competitors in a policy area, as several scholars have found that the number of other interest groups competing for members and influence in the same policy sector is positively related to interest group dissolution and rates of demise (Gray and Lowery 1995, 1996b, 2001; Lowery and Gray 1995; Nownes and Lipinski 2005). I accordingly expect that there is a crowding-out effect at play when it comes to interest group mobilisation. As interest groups compete for scarce resources to sustain themselves, competition between interest groups for resources intensifies as the number of interest groups increases in a policy domain. Thus, the larger the number of (new) competitors, the more intensive the competition over supporters and scarce resources. As a result, the growth rate of the interest group population should slow down. I therefore expect that the number of (new) competitors negatively affects interest group survival as an interest group faces more competitors in the struggle for scarce resources. The number of competitors is measured by the overall number of other interest groups in the same policy sector in a given year while the number of new competitors is measured by the number of newly registered interest groups in the same policy domain in a given year. Third, German reunification constitutes an important external shock that fundamentally changed Germany’s interest group system. I include a structural break control to capture this external shock. This variable is coded 0 before German reunification and 1 afterwards. Table A1 in the online appendix provides summary statistics for all variables included in the analysis.

Data analysis

Figure 1 presents a histogram of the age of the 1699 interest groups in the dataset. The figure shows that there is considerable variation across groups and several peaks in the distribution. A first peak can be identified at an interest group age around approximately 22 years. That peak is hardly surprising as a large number of interest groups were newly established after German reunification, such as the Association of German Railway Customers (1990), the Association of German Civil Law Notaries (1991) or the Association of Working Mothers (1990). The second peak can be found at about 63 years, which indicates that a large number of interest groups were established after the Second World War, coinciding with the foundation of the Federal Republic of Germany in 1949; these include the Federation of German Industries (1949), the German Trade Union Confederation (1949) and the German Hotel and Restaurant Association (1949). Finally, a number of interest groups reached an age of about 120‒130 years, indicating that they were founded during the 1870s, the 1880s and the 1890s, shortly after the foundation of the German empire in 1871. For instance, the German Lawyers Association was founded in 1871, the German Chemical Industry Association was established in 1877 and the Workers’ Samaritan Federation was founded in 1888. Thus, the development of the German interest group population is strongly affected by historical developments leading to a considerable transformation of the German interest group system in line with regime changes throughout Germany’s history.

In order to test the theoretical expectations, I rely on event history analysis as interest survival is measured by the number of years an interest group has survived. Event history models are designed to model duration data where the phenomenon of interest is duration until a certain event occurs, in this case the duration until an interest group is dissolved. The unit of analysis is a (unique) interest group that registered at the Bundestag. The dependent variable is the number of years between an interest group’s foundation and its last registration at the Bundestag. During this time frame, it might happen that important changes occur that affect the interest group’s survival (Box-Steffensmeier and Zorn 2001). State changes occur in this study with regard to the number of competitors, the number of new competitors and public salience, which can take on different values from year to year. This implies that an interest group is recorded for as many years in the dataset as it registered at the Bundestag to take into account the state changes of these variables over the years.

I estimate a semi-parametric Cox model as we do not have any a priori assumptions about the specific probability distribution for the time until an interest group ceases to exist (see also Nownes and Lipinski 2005). An important assumption of the semi-parametric Cox model is that the hazard function of each observation follows exactly the same pattern over time. Violations of this assumption might result in biased coefficient estimates and decreased power of significance tests (Box-Steffensmeier and Zorn 2001: 974). I therefore first tested this assumption based on the Schoenfeld residuals (see Tables A3, A4 and A5 in the online appendix). The results show that three of the predictors in model 3 violate the proportional hazards assumption, namely the number of competitors, the number of new competitors and reunification.

Following the procedure suggested by Box-Steffensmeier (2001), I therefore estimated Cox models including time-varying coefficients for the number of competitors, the number of new competitors and reunification in model 3. The incorporation of time-varying coefficients allows for taking into account that the effects of these independent variables vary over time. Table 1 presents the results of the models.⁵ Model 1 only includes the two explanatory variables group type and public salience, while model 2 additionally includes the control variables and model 3 also comprises the structural break control for reunification. The effects are reported as coefficients. Coefficients with a negative value indicate a positive effect on interest group survival while positive values indicate a negative effect on survival.

The results of the Cox regression models provide evidence in line with both hypotheses. First, the public salience of a policy area has a statistically significant positive effect on interest group survival across all three model specifications. The higher the salience of the policy area among the general public, the longer interest groups can on average sustain themselves. Thus, interest groups working in policy areas that are important to the general public find it considerably easier to sustain themselves than interest groups lobbying in policy domains that are of little importance to the general public. The empirical evidence therefore supports hypothesis 1 that the public salience of a policy area positively affects the survival of interest groups. Second, interest group type crucially matters for interest group survival. Cause groups have a much harder time to sustain themselves compared to sectional groups. More specifically, the risk of dissolution is 1.350 times higher for cause than for sectional groups.⁶

As cause groups suffer much more from collective action problems than sectional groups, it is much more difficult for them to mobilise potential supporters and obtain the resources from them that are necessary for their survival. Sectional groups by contrast represent well-circumscribed constituencies that typically fight for concentrated economic interests of their supporters so that resource mobilisation is a much smaller problem.⁷ The effects of public salience and interest group type on interest group survival are further illustrated in Figures 2 and 3.

Figure 3. The effect of interest group type.

These figures show the values for the cumulative hazard function (Cleves et al. 2008) based on the smallest and largest values for public salience (Figure 2) and for the difference between cause and sectional groups (Figure 3) respectively. The y-axis shows the cumulative hazard representing the share of interest groups that are dissolved and the x-axis shows the age of interest groups in years. Figures 2 and 3 clearly show that public salience and interest group type have an important substantive effect on interest group survival. Thus, interest groups lobbying in policy areas that are important to a lot of people in a society can sustain themselves significantly longer than interest groups working in policy areas that are not important to many citizens. Similarly, sectional groups representing the special interests of a specific sector of society such as business associations or trade unions persist significantly longer than cause groups such as environmental or human rights groups.

In order to further test the robustness of the results, I have estimated two additional Cox regression models. First, given that the lobbying register only started in 1974, we simply do not know whether interest groups registered in 1974 had already been lobbying the Bundestag before that. I therefore estimated an additional Cox regression that excludes all interest groups that were registered in 1974 to check whether the results hold when we only analyse those interest groups for which we know the year in which they lobbied the Bundestag for the first time. The results of the additional Cox regression are in line with the presented findings (see Table A6 in the online appendix). While public salience has a statistically significant positive effect on interest group survival, cause groups face a significantly higher risk of dissolution.

Second, as interest groups that are large umbrella associations might last significantly longer due to their superior position in the interest group system, I have also estimated an additional Cox regression controlling for whether an interest group is an umbrella association. The results are reported in Table A7 in the online appendix. While umbrella associations indeed last significantly longer, the effects of the explanatory variables ‒ interest group type and public salience ‒ still hold.

Conclusion

Why do some interest groups exist several decades or even centuries while other interest groups disappear after just a few years? This article has addressed this important question by analysing the survival of 1699 interest groups registered at the German Bundestag between 1974 and 2012. The empirical results show that the survival of interest groups is crucially affected by interest group type and the public salience of the policy area they are working in.

The results have important implications for our understanding of interest group mobilisation and political representation. As it is important for societal interests to be represented by a well-organised interest group that effectively advocates for their preferences in the political arena, understanding why interest groups survive has important implications for the representation of societal interests in Western societies more generally. Basically, the findings indicate that societal interests in highly salient policy areas have a much better chance to be effectively represented. Thus, societal groups in a publicly salient policy domain can much more easily mobilise resources and overcome the hurdles of collective action than societal interests in a policy area that is largely insulated from public interest. While one may view this result positively given that societal interests salient to a large part of the electorate are better represented, one may also raise negative normative concerns as particularly those societal interests that are marginalised in societies cannot organise effective interest representation. However, as the recent Fridays for Future movement has forcefully demonstrated, a previously neglected policy area can come to dominate public debate. Thus, public salience is not exogenous, but the salience of policy areas changes over time amongst others due to the activities of interest groups. Future research should therefore shed light on how interest groups can affect the salience of policy areas.

This study has furthermore demonstrated that interest group type also plays an important role for interest group survival. Sectional groups typically representing economic sectors such as the automobile industry or farmers find it a lot easier to sustain themselves in the long run than cause groups that represent diffuse societal interests such as environmental protection or human rights. As a result, the often voiced concern that interest group systems are biased towards economic interests also holds for interest group survival. Sectional interest groups are systematically advantaged over cause groups as they represent the economic interests of a clearly circumscribed constituency that is typically well-endowed with resources to ensure the long-term survival of the interest groups. Cause groups by contrast constantly face the problem of attracting supporters and obtaining from them the necessary resources to ensure their survival. Diffuse societal interests are therefore not only disadvantaged when it comes to mobilisation, but also when it comes to maintaining an organised interest group fighting for their cause. In conclusion, the often cited bias in the ‘heavenly chorus’ (Lowery et al. 2015; Schattschneider 1960) is partly a result of collective action problems that make it difficult for cause groups to maintain their long-term survival.

While this study has made an important contribution by investigating interest group survival outside the US, there are important avenues for future research. First, the analysis was limited to examining interest group survival in Germany. As Germany shares many similarities with other Western democracies, such as its parliamentary system, its experience with coalition governments, its corporatist interest intermediation patterns and the composition of the interest group population (Leech et al. 2005; Rasmussen 2015; Wonka et al. 2010), it is reasonable not to expect fundamentally different results in other countries. However, external validity is best achieved by comparative research, so future studies should extend the analysis to other countries and other institutional settings. Second, an important avenue for future research is to further shed light on the relationship between interest groups and citizen preferences. Do interest groups provide a balanced picture of societal interests? Further, do interest groups really take up the concerns of citizens or do interest groups also shape citizen preferences?

Acknowledgements

I thank Elisabeth Zeidler, Emanuel Deutschmann, Elisabeth Glück, Marcus Carlsen Häggrot, Clara Heinrich, Christina Hecht, Svenja Krauss, Jens Olesen, Lara Panning, Jochen Rehmert, Jonas Reißmann and David Schmuck for valuable research assistance.

Disclosure statement

No potential conflict of interest was reported by the author.

Additional information

Funding

I am grateful to the Fritz Thyssen Foundation (Grant Az. 10.13.2.109) and the British Academy (Grant SG111433) which generously funded research for this article.

Notes on contributors

Heike Klüver

Heike Klüver is Professor and Chair of Comparative Political Behaviour at Humboldt-Universität zu Berlin. Her research interests include interest groups, political parties, coalition governments, political representation, European politics and quantitative text analysis. She has published her work in, among others, the American Journal of Political Science, the Journal of Politics, Comparative Political Studies, the British Journal of Political Science, the European Journal of Political Research and with Oxford University Press. [[email protected]]

Notes

1 I have estimated an additional model in which I control for whether an interest group has undergone a merger with another interest group (see Table A9 in the online appendix). Having undergone a merger does not significantly affect interest group survival while the main results remain robust.

2 Overall, we identified 3473 distinct interest groups that registered between 1974 and 2012 at the Bundestag. An inspection of the distribution of key variables in the full and the analysis sample presented in Table A11 in the online appendix shows that there is no indication of a selection bias due to missingness (Rubin 1976).

3 Given that the original codebook was developed to classify bills and other policy documents into policy areas, the 225 sub-issues are very fine-grained. Since interest groups compete in broader areas, I only rely on the major policy areas to classify interest groups.

4 The interest group registration data including the policy area classifications based on the Policy Agendas Codebook is made available on the website of the author.

5 In order to check the robustness of the results, I have also estimated two additional specifications in which fixed effects for policy areas and years are included (see Table A8 in the online appendix). The effect of public salience is not statistically significant in the issue-area fixed effects model, but statistically significant in the year fixed effect model. This suggests that it is primarily variation across policy domain that accounts for the public salience effect on interest group survival.

6 The hazard ratio for the cause group coefficient in model 3 is 1.350.

7 One could expect that there is an interaction effect at play between interest group type and public salience such that cause groups benefit more from public salience than sectional groups. In order to test whether this is the case, I have run an additional Cox model in which group type is interacted with public salience (see Table A10 in the online appendix). The results show that the effect of public salience is not conditioned by interest group type so that both cause and sectional groups are similarly affected by public salience.