Participatory Budgeting in Germany –
Towards a More Systematic, Longitudinal Analysis
Sebastian H. Schneider1
Stefan Busse2
Paper prepared for the ECPR General Conference 2015
Université de Montréal, 26-29 August 2015
- Work in progress. Please do not cite! Comments very welcome! –
Abstract
Participatory budgeting is mushrooming internationally in North and Latin America, Asia,
Africa and Europe in the last twenty years. This calls for an evaluation of its functioning and
effects. Hence, research on participatory budgeting is increasing as well. Unfortunately, most
studies use a case study design, focusing on one case or comparing few cases which makes
them susceptible for selection biases. Only recently quantitative studies comparing the effects
of participatory budgeting have been published using large-n datasets for Brazil. For
Germany, research consists mostly of single case and evaluation studies. In this paper, we
present a new dataset for Germany covering participatory budgeting processes from 1998 to
the present and containing a wide range of variables. For example, our dataset contains rich
information on process design, participation rates, political and social context, and the
financial situation of the municipality. This dataset can be regarded as a starting point for
comparative research on participatory budgeting in Germany. To illustrate its possible
applications, we present an analysis of participation rates. Results show a negative effect of
the share of people on welfare, a positive effect of the usage of the internet and huge
differences between Eastern and Western Germany.
Key words: Participatory budgeting, Germany, large-n, dataset, democratic innovation,
participation.
1 Institute for Political Science, University of Münster, Scharnhorststr. 100, 48151 Münster, Germany. E-Mail:
[email protected] 2 International Graduate School "Transformations of Civic Society", Martin Luther University Halle-Wittenberg,
Germany. E-Mail: [email protected]
1
1.Introduction
After its first implementation in Porto Alegre, Brazil, in 1989 participatory budgeting
processes are mushrooming all around the world (Sintomer et al. 2013). In publications, an
estimated number of about 1.500 processes are reported. Despite this enormous expansion,
research is predominantly based on (comparative) case studies (e. g., Abers 1996, Avritzer
2006, Baiocchi 2003, Melgar 2014, Rodgers 2010, Röcke 2013, Ryan & Smith 2012, Talpin
2011, Wampler 2008) and with a strong focus on Brazil and the pioneering example in the
city of Porto Alegre. Only recently, large quantitative and comparative studies building on
aggregate data become available but again only for Brazil (Boulding & Wampler 2010, Spada
2014, Touchton & Wampler 2014, for an early example using comparative survey data see
Wampler 2007). For other countries, research on participatory budgeting is either scarce or
also consists mostly of case studies, evaluation reports and descriptive accounts (for an
overview see, e. g., Herzberg et al. 2012, Sintomer et al. 2008, 2010, 2013).
This scarcity is also true for Germany where participatory was introduced at the turn
of the millennium with the support of the federal state of North Rhine-Westphalia and the
Bertelsmann Foundation (Sintomer et al. 2010: 112-127). From its beginnings in smaller
towns participatory budgeting made its way to the borough of Lichtenberg (about 260.000
inhabitants) in the capital Berlin in 2005 and large cities as for example Hamburg, Cologne,
Frankfurt and Stuttgart. Despite the fact that the online platform www.buergerhaushalt.org
lists 71 running and 52 terminated participatory budgeting processes in 2014 in their annual
status report (Ermert et al. 2015: 7), only few evaluation studies are available (see section 2).
In spite of this scarcity, a rigorous evaluation of participatory budgeting processes is
highly recommend to find out whether they can be regarded as a remedy to the assumed
democratic deficit (Geißel 2013: 8-9), especially since they are sponsored by public funds. To
assess its effects, several evaluation categories have been proposed, as for example inclusive
and meaningful participation, legitimacy, deliberation, effectiveness and democratic
citizenries (Geißel 2013: 16-22). For example, the number and social composition of
participants as well as their motivations to participate, the quality of discussion and the effect
of participatory budgeting on politics (local council and public administration), policy and
citizens are relevant research areas. To investigate these areas, a careful selection of cases
(municipalities) is the necessary point of departure since the pragmatic selection of accessible
cases or research on best practice-cases might lead to serious biases and, hence, invalid
inferences (see, e. g., Geddes 1990, Goertz & Mahoney 2012: Chap. 14, King et al. 1994:
2
Chap. 4).3 Unfortunately, until now for Germany no exhaustive database (sampling frame) to
conduct a proper case selection is available. What is more, as a complement to existing case
studies, we advocate studies using aggregate data and multilevel survey data to enable more
systematic longitudinal analysis (for a similar argumentation see Galais et al. 2012: 67-68).
To partly fill this lacuna, we present a new quantitative database covering nearly all
participatory budgeting processes in Germany from 1998 to the year 2015. The dataset
contains rich information on the design of the process (e. g., used communication channels,
forms of citizens’ input, thematic focus, etc.), participation (number of participants,
participation rates) as well as data on the social, socio-economic and political context within a
municipality. To exemplarily demonstrate the potentials of the database, subsequently, we
present an analysis of participation rates to test hypotheses about the effect of characteristics
of the municipal context and institutional design.
Our paper is structured as follows: Section 2 develops our argumentation for the need
for a comparative database on participatory budgeting in Germany and presents a brief
overview on existing research and data for Germany. In section 3, we present the data
collecting process, data sources as well as the basic structure of our database. The benefits of
this database are demonstrated in section 4, first, with a descriptive analysis of the
longitudinal development of participatory budgeting in Germany and, secondly, with the
multivariate analysis of participation rates mentioned above. In section 5 we summarize our
analyses, discuss the benefits and shortcomings of our database followed by a short glimpse
on further research tasks and possibilities for extending the database with new cases and
additional variables.
2.LiteratureandDataSourcesonParticipatoryBudgeting
In general, literature on participatory budgeting in Germany can be subdivided into three
broad fields: academic studies including students’ theses, evaluation studies and companions
for local councilors and administrative staff. What all three fields have in common is their
focus either on one or few cases (municipalities). The applied research design and methods
cover the whole variety of the social scientific toolbox (surveys, expert interviews, participant
observation, document analysis, etc.). They offer deep and detailed insights into the
implementation, participation process and effects in the municipality under examination.
3 We are aware of the fact that the perils of selection bias, of course, are heavily related to the research purposes
since it can be assumed that different case selection procedures are recommended for explorative, qualitative set-theoretic or quantitative large-n studies (see Goertz & Mahoney 2012: Chap. 14).
3
However, these contributions predominantly are descriptive and/or explorative. Of course,
one should keep in mind that, for example compared to institutions of direct democracy,
participatory budgeting is still a rather young democratic innovation in Germany. Hence, the
number of available cases to investigate was limited.
In case of evaluation studies, pioneering work was done by Klages and Daramus
(2007) with a report on Berlin-Lichtenberg, followed by a series of reports on participatory
budgeting in Oldenburg (Schneider 2011, 2012, 2013), a report for the cities of Cologne
(Taubert et al. 2011), Jena (Kersting et al. 2013), Frankfurt (Geißel et al. 2013) and the
administrative district Mansfeld-Südharz (Franzke & Roeder 2014). Additionally, academic
in-depth studies are available for Potsdam (Franzke & Kleger 2006), Berlin-Lichtenberg
(Röcke 2009)4 and Freiburg (Färber 2009) as well as several students’ theses. Some of the
studies with evaluative character are additionally covered in meta-studies with different
objectives by Busse and Schneider (2014) and Geißel et al. (2015). Furthermore, there are
only two comparative studies available. They deal with the democratic novelty of German
cases (Herzberg 2009) and the aspects of online participation (Masser et. al. 2013).
Table 1: Overview of studies on German cases of participatory budgeting. Single Case Studies (academic, evaluation studies)
Meta-Studies of single case Studies
Comparative Studies Student’s theses (selection)
Färber 2009,
Franzke & Kleger 2006,
Geißel et al. 2013,
Kersting et al. 2013,
Klages & Daramus 2007,
Röcke 2009,
Schneider 2011, 2012,
2013,
Taubert et al. 2011
Busse & Schneider 2014,
Geißel et al. 2015
Herzberg 2009,
Masser et. al. 2003
Eich (2011),
Schnelle (2013)
But since the number of participatory budgeting processes has steadily been growing over the
last years (Ermert et al. 2015: 9), for a better understanding of the functioning of participatory
budgeting, in contrast, a shift from exploration and description towards studies testing causal
hypotheses in our opinion is highly recommended. This recommendation holds on the one
hand for practically orientated evaluation studies, on the other for pure academic research as 4 In her international, comparative study, Röcke (2009) compares the case of Berlin-Lichtenberg with the cases
of Poitou-Charentes (France) and Salford (Great Britain). With regard to the literature on German cases, it should be labelled as a single case study.
4
both utilize an overlapping set of categories to assess democratic innovations in general (e. g.,
Abelson & Gauvin 2006, Geißel 2013, Rowe & Frewer 2000, Smith 2009) and participatory
budgeting in particular (Busse & Schneider 2014, Geißel et al. 2015). Important research
questions are, for example, how citizens can be mobilized to participate, how a good quality
of deliberation can be secured, under which conditions do participatory budgeting processes
have effects on the municipal budget or how positive effects on the attitudes of citizens can be
achieved. Such questions are often based on normative assumptions, which are derived from
theories of participatory and deliberative democracy (see Held 2006) that emphasize the
importance of participation being effective, deliberative and inclusive.
To investigate these questions, several methodological approaches are feasible, of
course depending on the precise research task:
1. Large-n (panel) surveys among citizens in municipalities with and without
participatory budgeting
2. Qualitative comparative case studies
3. Aggregate data studies
4. Experimental studies, either in real-world settings or in the laboratory
Except of pure laboratory experiments, all these approaches require a comprehensive
database, either indirectly for a proper case selection (comparative case study designs) or
directly for aggregate data analysis respectively for adding contextual data to individual level
survey data. When such a case selection is missing, studies become prone for biases and
invalid inferences. For example, beneficial effects of participatory budgeting based on
anecdotal evidence might vanish as soon as systematic comparisons are conducted and
relevant characteristics of the participatory budgeting design and municipal context (e. g.,
political and financial situation) are taken into account. This also helps to demystify certain
assumptions that present German participatory budgeting either in a light that is too positive
or too negative which can easily be achieved by cherry-picking convenient cases.5
An example for such a database can again be found in Brazil. The Brazilian
Participatory Budgeting Census created by Paolo Spada and colleagues contains all cities (N
≈ 500, six time periods) in Brazil with more than 50.000 inhabitants in the period from 1989
to 2012.6 Data was collected by means of internet search and follow-up telephone interviews.
5 Our subjective impression is that on the one hand proponents of the participation industry (consultants, etc.)
select cases with high participation rates, positive media coverage whereas opponents focus on cases with low participation rates and conflicts between key actors.
6 http://www.spadap.com/data-collection-projects/the-brazilian-participatory-budgeting-census/ [retrieved 11.08.2015]
5
Due to available geo-coding variables, census data can easily be merged with other existing
Brazilian socio-economic and political datasets. This approach serves as model for the
creation of our database for Germany presented in this paper.
3.ANewDataset
3.1 Defining participatory budgeting
Whether a municipality is included in our database, depends on the definition of participatory
budgeting used. Sintomer et al. (2013: 3) – probably the most widespread definition in the
literature - define participatory budgeting as “the participation of non-elected citizens in the
conception and/or allocation of public finances”. They add the following five criteria to
specify their definition:
1. “The financial and/or budgetary dimension must be discussed;
participatory budgeting deals with scarce resources
2. The municipal level must be involved or a (decentralised) district with an
elected body and some power over administration (the neighbourhood
level is not enough)
3. It must be a repeated process (one meeting or one referendum on
financial issues are not examples of participatory budgeting)
4. The process must include some form of public deliberation within the
framework of specific meetings/forums (the opening up of administrative
meetings or traditional representative instances to ‘ordinary’ citizens is
not participatory budgeting)
5. Some accountability with regard to output is required” (Sintomer et al.
2013: 3).
In general, we support and adapt the definition above for our database but with a few
modifications. We do not assume that participatory budgeting necessarily must be a repeated
process. Of course, it should be implemented with a long-term perspective but it is also
possible that after the first year it is terminated for whatever reasons. In our opinion, it does
not make sense to retrospectively deny the status of participatory budgeting in such a case. 7
Therefore, we use slightly relaxed criteria to include a municipality in our data set since we
believe that there is a difference between labelling a participatory process “participatory
7 To illustrate our point: An election in an authoritarian regime without competition between candidates, free and
secret voting, etc. would still be called “election”. To further qualify the concept, one could use “democratic elections” including real competition between candidates and parties, free and secret voting, etc.
6
budgeting” and its quality (e. g., used media channels, opportunities for making proposals or
deliberation, etc.). Hence, we include a municipality in our data set as soon as it labels its
participatory process participatory budgeting (in German Bürgerhaushalt or similar
alternatives like Beteiligungshaushalt) without any normative assessment of its quality or
reference to minimum standards. Required are only two attributes:
1. A focus on the municipal budget (criteria 1 and 2)
2. Some form of public input and discussion8 about the budget plan (criterion 4).9
This can be regarded as a minimal definition of participatory budgeting which can radially be
expanded (see Wonka 2007: 70) to the complex definition of Sintomer et al. (2013: 3). The
latter can, in our opinion, rather be regarded as a definition of a normative ideal participatory
budgeting design.
Presumably, the design and quality varies immensely between municipalities, which, in
turn, might influence participation and outcomes. Therefore we capture the institutional
design in great detail. Additionally, we plan to include an expert rating on the design quality
that helps to distinguish cases that might share the same core elements but differ significantly
in the quality of conducting the process.
3.2. Data sources used
As a starting point, we use the data base of www.buergerhaushalt.org which documents nearly
all participatory budgeting processes in Germany and provides information on their
institutional design.10 For example, it provides data on the communication channels used, the
focus of the procedure (whole budget or only certain parts, saving or spending, etc.) and
forms of accountability. Additionally, basic information on population size as well as the
location is provided. Presumably, this database is the most exhaustive source on participatory
budgeting in Germany albeit completeness seems illusory since no government-run official
register is available. Additionally, the quality of information provided by municipalities about
the processes differs greatly. The following table provides an overview of the categories used
to capture the design of a participatory budgeting process.
8 We explicitly avoid the term deliberation because it refers to a discussion in which certain criteria are met (see,
e. g., Bächtiger et al. 2010). 9 For example, some municipalities only collect budget proposals via direct e-mail or postcard to the mayor or
treasurer without any public discussion. 10 Basic data on participatory budgeting in German municipalities can be obtained from their website via open
data-interface. For a similar approach to capture local participation experiences in Catalonia, Spain, see Font and Galais (2011).
7
Table 2: Coding scheme. Category Coding BHH.org 2015 Reduced Coding Scheme Explanation Scope 1. Total budget
2. Selected budget areas 3. Fixed amount 4. Mixed 5. Unclear
- What is the financial scope of participatory budgeting?
Form of Input 1. Expenditures 2. Savings 3. Open 4. Unclear
- What is the financial focus of participatory budgeting?
Form of Participation
1. Proposals 2. Proposals and Feedback 3. Feedback 4. Decision based 5. Mixed 6. Unclear
1. Proposal 2. Feedback 3. Mixed
Can citizens hand in own budget proposals or are they only allowed to discuss and rate official proposals?
Participation Channel
1. Internet only 2. Internet and additional citizen’s forums 3. Citizen’s forums 4. Citizen’s forums and additionally internet 5. Complex 6. Citizen’s forums and internet 7. Unclear
1. Internet only 2. Citizen’s forums only 3. Survey 4. Multichannel 5. By Post 6. Unclear
Which channel is used for participation?
Accountability 1. Detailed 2. Detailed with monitoring 3. Collected 4. None 5. Unclear
1. Yes 2. No 3. Unclear
How does the municipality inform about participation results?
Source: Ermert et al. (2015), own translation. Note: In the category “Form of Participation”, “Decision based” means that citizens have a binding influence on budget decisions whereas “Mixed” means combinations of proposals, feedback and decision-based (Ermert et al. 2015: 13-14). These characteristics are not used very often in German municipalities.
This data base was used to conduct an intensive web search on the websites of the
municipalities, local newspapers as well as private websites and weblogs. Additional
information was obtained from available literature on participatory budgeting in Germany
which - as mentioned before - consists of introductory texts (e. g., Franzke & Kleger 2010,
Günther 2007) as well as comparative case studies (Sintomer et al. 2010), students’ theses and
evaluation studies (see Busse & Schneider 2014, Geißel et al. 2015). This way it was possible
to add further cases and/or time periods of existing cases to the data set – utilizing the
buergerhaushalt.org coding scheme – and to correct some false information of the original
dataset. For instance, the 2013 case of Cottbus simply did not exist, although it is listed in the
database of www.buergerhaushalt.org.
8
In addition to the categories used by buergerhaushalt.org, we found it useful to include
summarized versions of variables like “Form of Participation”, “Participation Channel” and
“Accountability”. One the one hand, we chose this approach because the original codes are
either not distinct enough (e. g., no difference between the second and fifth “Form of
Participation”-code), or too precise in regard to the information available on a particular case
(e. g., “Participation Channel” and “Accountability”). On the other hand, we opt for a less
complex scheme to reduce complexity and make our analyses more accessible.
A vital source of detailed information on the schedule and the design of participatory
budgeting processes can be found in form of the digital information systems of local councils.
What is more, developments and changes over time within a municipality can be
reconstructed. A very nice feature of those websites is the often precise documentation of the
number of participants. This comes as no surprise since it can be used as a simple heuristic to
evaluate the success of a participatory budgeting process which taps its inclusiveness and
consequently its legitimation. One caveat must be noted: In many cases these information
systems only cover the last two legislative periods. Besides, some municipalities do not
provide information on their participatory budgeting process, or even delete the information
that was provided on previous cycles – especially if the project was terminated.
In a subsequent step, the core of the data set was supplemented with rich information on
several aspects of the respective municipality. The following variables are included:
9
Table 3: Context-related variables. Context Variable
Political context Turnout last election for local council (percent) *
Party affiliation of mayor
Number of seats of most important parties
Totals number of seats in local council
Share of left parties council (SPD, Left Party, Green
Party) *
Laakso-Taagepera Index of effective parties*
Socio-Demographic context Population *
Percentage of people on welfare (“SGB II”) *
Percentage of foreigners *
Median age
Old-age dependency ratio
Share of people living in single-person households
Economic/financial context Purchasing power per capita (in Euros)
Share of people working in service sector
Municipal debts per capita (in Euro)
Municipal ways and means advance per capita (in Euro)
Note: * = Included in regression analysis (see section 4.2).
Political context variables can be obtained from municipality websites and the elections
administrators of the federal states. Socio-demographic and economic/ financial context
variables are provided by www.wegweiser-kommune.de from the Bertelsmann Foundation
which provides statistical data on all German municipalities with more than 5.000 inhabitants
from 2003 to 2013.11 Here a first problem becomes obvious: our database contains
participatory budgeting process to the present day. Hence, for every participatory budgeting
process conducted in 2014 the values are set to missing and will be replaced with valid data as
soon as the Bertelsmann Foundation releases data for 2014. Accordingly, for the following
analysis of participation rates, participatory budgeting processes conducted in 2014 and later
will be excluded. A second problem concerns the consistency of provided indicators: Some
major changes in the definitions and computing procedures were conducted in 2005.
Consequently, we restrict our following analyses to the time period from 2006 onwards.
11 During the work on this paper, the Bertelsmann Foundation added data for 2013 and deleted data for the
period 2003 – 2005 from their data base.
10
4.EmpiricalAnalysis
To illustrate the potentials of our dataset, we conduct two separate analyses. First, we take an
in-depth look at the current state of participatory budgeting in Germany by describing the
institutional design and its development over time. Central questions are what channels are
used, which thematic focusses exist and how do citizens participate. For this analysis, we
utilize all available time periods (1998 – 2015). Secondly, we analyze participation in
participatory budgeting processes by means of a theory-guided, multivariate aggregate data
analysis. In contrast to the descriptive part, we restrict the data set to the time periods from
2006 to 2013 for this analysis.
4.1 Descriptive analysis
Table 4: Basic participatory budgeting design over time (n = 310) (percent per row). Year Internet Forum Multi-channel Postal N 1998 0,0 100,0 0,0 0,0 1
1999 0,0 100,0 0,0 0,0 1
2001 0,0 0,0 0,0 100,0 1
2002 0,0 75,0 0,0 25,0 4
2003 16,7 83,3 0,0 0,0 6
2004 0,0 80,0 20,0 0,0 5
2005 0,0 50,0 50,0 0,0 6
2006 12,5 50,0 37,5 0,0 8
2007 0,0 30,0 70,0 0,0 10
2008 0,0 28,6 71,4 0,0 14
2009 33,3 13,3 53,3 0,0 15
2010 31,0 20,7 48,3 0,0 29
2011 33,9 5,4 58,9 1,8 56
2012 40,7 8,5 49,2 1,7 59
2013 41,2 7,8 51,0 0,0 51
2014 38,5 7,7 51,3 2,6 39
2015 0,0 20,0 80,0 0,0 5
Total 30,6 16,8 51,0 1,6 310
Note: Data collection for 2015 is not yet complete.
11
A first look at Table 4 reveals that the number of conducted participatory budgeting processes
steadily increases from the year of the first recording and reaches a peak in 2012.12 From then
on, the number drops. Despite it is too early for definite conclusions, this finding might be a
hint that the trend towards participation in local budgeting is stagnating or even in decline.
Regarding the process design, the descriptive analysis shows that pure online and
multichannel processes are the state of the art in Germany. In contrast, the usage of citizen
forums and the collection of proposals in local facilities were popular in the beginning. In
recent years, the solely usage of this particular design is rare. This is even more the case for
pure postal participatory budgeting processes. A reason might be the connection between the
rising popularity of online tools that made the use of participatory budgeting in large cities
efficient (Sintomer et al. 2010: 138).
Now we turn to the details of the participation process (see Table 5). A first aspect is the
focus of the participatory budgeting process. In all years, open processes dominate. That
means people were allowed to make proposals or rate proposals on expenditures as well as on
savings. Only in the period from 2010 to 2012, for participatory budgeting with a focus on
revenues and savings a peak can be observed which declined in the years after.13 A similar
pattern can be found for the topical focus. In the same period, participatory budgeting with a
focus on a certain topic (e. g., education or green areas) was slightly in vogue but in all years
the processes predominantly covered the whole budget plan. Finally, we have a look at the
possibilities citizens are given to participate. In all years, the most prominent design is
proposal-based participatory budgeting whereas pure feedback-driven (evaluation of
proposals of council and municipal administration) is rare. Interestingly, mixed-designs
(combination of own proposals and evaluation of proposal of council and administration)
were popular at the turn of the last decade but lost ground in recent years. The following table
offers a detailed summary.
12 Compared to the annual reports of www.buergerhaushalt.org, in general, our total number of participatory
budgeting process per year are lower because we look at processes conducted. Therefore, we do not count municipalities with inactive processes or a two-year schedule.
13 To help budget consolidation, local governments (e. g., in the cities of Cologne and Solingen) used participatory budgeting for public discussion and prioritization of saving measures.
12
Table 5: Detailed information on process design (percent per row). Year Scope Form of Input Form of Participation
Total budget
Areas Budget Mixed Expenditures Revenues & Savings
Open Proposals Feedback Mixed
1998 100,0 100,0 100,0
1999 100,0 100,0 100,0
2001 100,0 100,0 100,0
2002 75,0 25,0 25,0 75,0 25,0 25,0 50,0
2003 83,3 16,7 16,7 83,3 66,7 33,3
2004 100,0 100,0 60,0 40,0
2005 83,3 16,7 100,0 66,7 33,3
2006 87,5 12,5 100,0 62,5 37,5
2007 63,6 27,3 9,1 100,0 72,7 9,1 18,2
2008 78,6 14,3 7,1 100,0 50,0 14,3 35,7
2009 66,7 20,0 6,7 6,7 13,3 86,7 46,7 13,3 40,0
2010 72,4 20,7 3,4 3,4 3,4 24,1 72,4 58,6 6,9 34,5
2011 87,5 8,9 1,8 1,8 1,8 26,8 71,4 71,4 5,4 23,2
2012 84,7 10,2 5,1 8,6 25,9 65,5 79,7 3,4 16,9
2013 88,2 7,8 3,9 7,8 11,8 80,4 74,5 5,9 19,6
2014 82,1 7,7 10,3 7,7 7,7 84,6 82,1 2,6 15,4
2015 80,0 20,0 20,0 20,0 60,0 100,0
Note: Data collection for 2015 is not complete yet. In contrast to Ermert et al. (2015), we code mixed designs - consisting of general proposals and proposals for a fixed amount of money - as “Total budget” since they usually affect the total budget plan.14
4.2 Example: analysis of participation rates
Participation rates are a key indicator to evaluate the success or failure of a participatory
budgeting project (e. g., Masser et al. 2013: Chap. 3).15 As is it seems, the basic formula is:
the more people participate, the higher the success respectively the legitimacy of participatory
budgeting processes due to higher inclusiveness. This comes as no surprise since this
information is relatively easy to obtain compared to, for example, the usefulness of budget
proposals, fiscal effects or the quality of deliberation. Hence, the crucial question is what
14 We currently work on a revised coding scheme to remove ambiguities in the existing schemes and to reduce
inconsistencies between different approaches. 15 This analysis is part of Sebastian Schneider’s ongoing PhD project.
13
factors influence citizens’ participation in participatory budgeting?16 To investigate this
question, we use our data set described above and distinguish three blocks of determinants:
socio-demographic, political and institutional. Whereas socio-demographic and political
determinants can be regarded as contextual and beyond the influence of participatory
budgeting practitioners, the institutional design bears considerable leverage for stimulating
participation. The following table (Table 6) provides an overview on these determinants, their
operationalization and the expected effects.
Table 6: Determinants, operationalization, and expected effects. Variable Operationalization Expected Effect
Dependent variable
Participation rate
(Number of participants / population) *
100
Independent variables
Socio-demographic
Population size
Socio-economic heterogeneity
Population in a given year
Welfare quota (“SGB II”)
Share people with migration
background
+/- or reverse U-shaped
+ / -
+ / -
Political
Postmaterialist climate
Effective Number of parties
Financial Situation
Share of councilors from left parties
(SPD, Left Party, Green Party)
Laakso-Taagepera Index
Ways and means advance per capita (in
Euros)
+
+ / -
-
Institutional
Usage of internet
Expenditures possible
Own proposal possible
Dummy variable (0 = no/ 1 = yes)
Dummy variable (0 = no/ 1 = yes)
Dummy variable (0 = no/ 1 = yes)
+
+
+
Control variables
Turnout Last Election
Number of PB processes
Region (Eastern/Western Germany)
Percentage
Absolute number
Dummy variable (0 = west, 1 = east)
+
+/-
+/-
Note: + = positive effect, - = negative effect, +/- = unclear.
16 We are perfectly aware of the fact that quantitative macro-analyses suffer from major limitations (see, e. g.,
Kittel 1999, 2006). However, since no comparative survey data sets on issues of participation in participatory budgeting processes are available, we believe that our approach is a valuable first step for analyzing and evaluating participatory budgeting in Germany. Of course, in the near future studies based on individual-level survey data are highly recommended. To further safeguard our analyses, we focus our argumentation on the level of municipalities and avoid any direct arguments on individual behavior.
14
Before we proceed with the empirical analysis, of course, we should briefly make clear why
we include the variables above and which theoretical mechanisms we assume. Regarding
population size, we follow the classical debate on the relationship between size and
democracy (Dahl & Tufte 1973, for a summary see Kelleher & Lowery 2009). On the one
hand, one could assume that small villages and towns enable the creation of social capital
(dense social networks, generalized trust and norms of duty and reciprocity) which also might
foster participation in quasi-institutionalized forms of civic involvement. What is more,
political matters seem less complex and are therefore more accessible to citizen participation.
On the other hand, large cities might offer larger capacities for influencing political and
societal matters since they are more independent from counties and dispose of larger financial
resources. Additionally, greater societal heterogeneity might provide the base for conflicts
which, in turn, increase participation. A third option might be a reversed U-shaped
relationship with the participation being highest in medium sized municipalities because they
provide a mixture between the closeness of smaller towns and the capacities of large cities.
Hence, to account for socio-economic heterogeneity, we include the quota of people on
welfare as well as the percentage of people with migration background. Again, the expected
effect could either be positive or negative. From a grievance theoretical perspective (see, e. g.,
Gurr 1968), one can argue that heterogeneity (socio-economic and ethnical) facilitates conflict
because of differing preferences and thus leads to higher participation rates, as for example in
participatory budgeting. From a perspective of resource mobilization theory (see, e. g.,
McCarthy & Zald 1977), the effect might be contrary since a high number of people on
welfare and with migration background indicate less resources to participate in politics.
Regarding the political determinants, we first argue that a leftist and post-materialistic
climate within a municipality fosters political participation and thus participation rates in
participatory budgeting. Building on Inglehart’s (1977, 1990, 1997) contributions, we assume
that along with post-materialistic values comes a need for self-expression which includes
political participation within a municipality. What is more, local councilors, mayors and
administrative staff might be more supportive of participatory initiatives (Sweeting & Copus
2013: 135) and undertake more efforts to mobilize citizens. To measure a post-materialistic
climate, we sum up the seats of Social Democrats, Green Party and Left Party in the local
council and relate them to the total number of seats.
Secondly, building on recent contributions to research on the impact of institutions and
political opportunity structures on non-electoral participation (e. g., Dalton et al. 2010, Van
der Meer et al. 2009, Vráblíková 2014), we presume that the general political supply within a
15
municipality has an effect on strategical considerations of its citizens. The more effective
parties in a local council respectively the higher the fragmentation, the higher the incentives to
participate. The mechanism can be described as follows: The more parties have seats, the
more diverse are preferences within the municipality. First, this forces councilors to precisely
listen to public opinion. Secondly, citizens on average see higher chances for a serious
discussion of their budget proposals in municipalities with heterogeneous local councils
compared to municipalities with only few parties and clear majorities. Of course, fragmented
councils could signal complexity and uncertainty which, in turn, depresses participation.
Hence, we include the established Laakso-Taagepera Index of effective parties (Laakso &
Taagepera 1979) in our models.17
Finally, the effect for local debts, measured by ways and means advance per capita, is
rather self-explaining: The higher the debts within a municipality the lower the incentives for
participation since the chances for a budget proposal being successful are low. Furthermore,
given the way that revenue equalization is organized in the German Länder as well as the
wide spread need for consolidation, it seems irrational for citizens to propose the cut-back of
public services since they do not necessarily lead to an increase of expenditures in other
domains.
To finish this section, we now turn to the expected effects of the institutional design of a
participatory budgeting process. Our argument, in general, can be labelled as rationalist since
we assume that citizens (averaged over the whole population within a municipality) take cost
and benefits into their considerations whether to participate or not (see, e. g., Elster 2007:
Chap. 11). Hence, we first expect that the usage of the internet to involve citizens in the
budget planning raises participation rates because it significantly lowers the costs of
participation due to temporal and spatial flexibility (see, e. g., Lupia & Sin 2003: 316).
Secondly, we suppose that the possibility to allow expenditures and formulate own budget
proposals increases participation since this offers a wider array of potential voice and
influence for citizens and thus should lead to higher participation rates.
Finally, we add some control variables. First, to control for the general level of political
activism within a municipality, we include the turnout in the last local election. Secondly, to
account for temporal trends within a municipality, we include the number of participatory 17 The index proposed by Laakso and Taagepera (1979) is computed via the following formula:
1
∑
In this formula p is the normalized proportion of a party’s share of seats in the local council whereas n is the number of parties with at least one seat. The higher the value of the index, the higher the number of effective parties and the higher the fragmentation in a council.
16
budgeting processes in the respective municipality. Thirdly, the analysis contains a dummy
variable indicating whether a municipality is located in Western or Eastern Germany since
research indicates differences in political culture (e. g., Gabriel 2005) as well as the structure
of civil society (e. g., Anheier et al. 2000, Gensicke & Geiß 2010).
Strategy of analysis
The quasi-Time Series-Cross Section structure of the data set (see, e. g., Beck 2001) makes an
adapted modelling strategy necessary. It is obvious that the municipalities included in the data
set are not independent observations since there are in most cases repeated measurements for
several participatory budgeting cycles within a municipality. This leads to a clustered data
structure which, additionally, is heavily unbalanced. Most municipalities conducted
participatory budgeting only once or twice whereas some exceptions conducted up to ten.18
To make things worse: on the one hand there are data gaps within certain municipalities
because they suspended the participatory budgeting process in a given year. On the other, the
dependent variable is completely missing for some or all time periods in some municipalities.
Therefore, imputation methods (e. g., Honaker & King 2010) are, in our opinion, not feasible.
What is more, the chances are high that feedback processes between years are at work. To
illustrate this point: A high participation rate in the first year might, for example, lead to
positive media coverage which signalizes incentives for participation to the citizenry. A low
participation rate instead might instead signalize that it is useless to participate. Hence, some
form of temporal autocorrelation seems plausible. In sum, the assumptions for a pooled OLS
model are not met. In particular, standard errors and confidence intervals – which are
presented to account for uncertainty in estimations - might be biased. Another aspect
regarding data structure has to be mentioned: Most of the independent variables are either
time-invariant or vary only slightly between years. For example, the composition of the local
council or the mayor’s party affiliation are stable for years or more. Likewise, socio-
demographic context variables in many cases only vary in their second decimal places.
These peculiarities lead to the following modelling strategy: Unfortunately, due to the
structure described above, neither panel data (many observations, few time periods) nor
TSCS-data estimators (few observations, many time periods) are feasible because the number
of observations within a municipality in most cases seems too low and data gaps or missing
data cause serious problems. Furthermore, the computation of lagged dependent variables
18 The data set for the regression models utilizing listwise deletion of missing values contains 80 municipalities
with one to six conducted participatory budgeting cycles.
17
would lead to a sharp decline in the number of observations available for analysis since the
first time period - and hence many municipalities with only one completed participatory
budgeting process - will automatically be deleted. Hence, we opt for cluster adjusted standard
errors to account for the temporal clustering of observations within municipalities (for a
similar approach see Touchton & Wampler 2014).19 To check the robustness of our analysis,
we alternatively estimate a random intercept-model with a first-order autoregressive error
structure for longitudinal data which is tabulated in the appendix.20 Such models are harder to
interpret and, additionally, the true three-level hierarchical structure of the data set
(participatory budgeting processes nested in legislative periods nested in municipalities)
cannot be appropriately accounted for due to the restricted number of observations with
complete information. Hence, we prefer clustered standard errors.
Results
Figure 1: Average Participation Rate by Year 2006-2013 (N = 148).
19 Because of missing values on many socio-demographic variables and their different juridical status,
participatory budgeting processes conducted in the boroughs of Berlin are excluded. 20 For clustered standard errors, we use the package plm in R, for random effect-models the package nlme.
Graphics were created with the package ggplot2, regression tables with the package texreg.
18
The line plot above (Figure 1) shows the average participation rate in percent across all
included municipalities in a given year for the period from 2006 to 2013. For the first half of
the period investigated a steady increase of participation rates from below 0.5 percent to 1.75
on average can be seen. In the subsequent years, the average participation rate drops to a
mean of about 1.0 percent in 2013. The lowest participation rate can be found in Duisburg
2012 with 0.0047 percent whereas Luckenwalde 2012 has the highest rate with 9.7658
percent.
An intuitive interpretation of the aggregate pattern above is that participatory budgeting
had its heyday in the years 2009 and 2010 when many municipalities discussed the
implementation or started their process which was accompanied by the support of NGOs,
normative contributions from the fields of political science, public administration and
philosophy as well as positive media reports. This in turn nourished citizens’ interest in
participatory budgeting. In the following years, disenchantment might have set in because
participatory budgeting could only seldom fulfill the expectations. To analyze the influences
of the determinants described in the previous section, we now turn to regression models using
participation rates for a municipality in a given year as the dependent variable.21
21 Table A1 in the appendix presents all cases included for the period under investigation. Descriptive statistics
for all variables included can be found in Table A2. To give a detailed impression of bivariate relationships, Figure A1 presents a scatterplot matrix.
19
Table 7: Regression analysis (clustered standard errors in brackets).
OLS clustered SEs
(Intercept) -8.65*
(3.89) Population -0.00
(0.00)
Share people on welfare („SGB 2“) -0.09*
(0.04) Share People Migration Background 0.02
(0.03) Share of Left Parties 0.02
(0.02) Laakso-Taagepera-Index 0.28
(0.24) Financial Situation 0.00
(0.00)
Internet Use Yes 0.87***
(0.21)
Own Proposals Possible Yes 0.46 (0.54)
Expenditures Possible Yes 0.08 (0.31)
Turnout Last Local Election 0.06 (0.05)
Number PB Processes Conducted -0.01 (0.09)
Eastern Germany 2.19***
(0.65)
R2 0.42
Adj. R2 0.38
N 148 N Municipalities 80 Participatory budgeting cycles 1-6 Notes: ***p < 0.001, **p < 0.01, *p < 0.05
A first look at the regression table (Table 7) reveals that the explained variance with an
adjusted R² of 0.38 is fairly good for a model based on aggregate data. The highly significant
overall F-Test also indicates a good model fit (not tabulated). Most of the assumed effects
cannot be confirmed. Neither the population size22 nor any of the political variables have
significant and substantial relevant effects on the participation rate. Only the share of people
on welfare in a municipality has a significant positive coefficient, the higher the share the
lower the participation rate. An increase of 1 percent goes on average along with a 0.09
percent points lower participation rate. This corroborates established findings from studies on
22 We also ran the model with a quadratic term for population. This did not improve the model fit. Furthermore,
the residuals are approximately normally distributed. Hence, we do not log-transform the skewed dependent variable.
20
political behavior in Germany. For example, Schäfer (2012a) showed that electoral turnout
decreases with the share of people on welfare in German cities. On the micro-level, there is
also a strong relationship between individual resources and political participation (Schäfer
2012b), a finding that holds also for participatory budgeting in Germany as several
quantitative analyses of single cases could show (Schneider 2011, 2012, 2013; Taubert et al.
2011). This leads to the cautious conclusion that participatory budgeting possibly works better
in well-off municipalities.
From the block of institutional design variables, only the usage of online
communication has significant and substantial effect on the participation rate. On average, in
municipalities using the internet in a given year, participation rates are about 0.9 percentage
points higher. This finding comes as no surprise since online participation requires less time
and less spatial mobility. We conclude that using the internet significantly increases citizens’
participation albeit on average the participation rates are rather low across all institutional
variations (see Figure 1).
Finally, we turn to the included control variables. From these, only the east-west
dummy variable indicates a significant and substantially large difference in participation rates
for participatory budgeting processes conducted in Eastern and Western Germany. On
average, the participation rate is about 2.20 percentage points higher in Eastern Germany. At
this point, it remains an open question why this is the case. Ad-hoc, we suppose that German
reunification could not erase differences in political culture and political representation
structures. Presumably, East-Germans value participation more while at the same time civic
societal and representational structures are still not fully established (e. g., civic associations)
(see, e. g., Anheier et al. 2000, Gensicke & Geiss 2010: 5). This in turn could lead to a higher
demand for quasi-institutional means of political influence. Of course, further studies –
preferably based on a survey-data design – need to investigate such differences in greater
detail.
The alternative estimation approach utilizing random intercept-models presented in
appendix Table A3 does not substantially alter the results from the OLS regression with
clustered standard errors. Only the effect of the share of people on welfare within a
municipality slightly misses conventional significance levels with an exact p-value of 0.0701.
What is more, the coefficient for time (number of participatory budgeting processes
conducted) does not vary between municipalities (random slope-model). In sum, we regard
our findings as fairly robust.
21
5.Discussion
In this paper, we present a unique and comprehensive data base on participatory budgeting in
Germany. It covers nearly all cases in Germany albeit we cannot guarantee that there are
municipalities below the medial and scientific attention since there is no state-run official
register on participatory budgeting in Germany. Nevertheless, we are confident to offer as
much variation in cases as possible. What is more, emerging new cases or new time periods
for existing cases can easily be integrated into the data base. Our data base can be used as a
starting point for large-n studies, the selection of cases for qualitative (comparative) case
studies, or nested research designs using the estimation results and residuals of quantitative
analyses (in particular, regression analysis) to systematically choose cases for qualitative in-
depth analyses (see Lieberman 2005, Rohlfing 2008).
To illustrate the opportunities our data base provides, we first presented an overview on
the developments of participatory budgeting in Germany and then addressed an important
explanatory question: Which socio-structural, political and institutional determinants (in
particular, process features) correlate with participation rates? Our descriptive analyses show
that the number of conducted participatory budgeting processes steadily increases up to the
year 2012 and declines afterwards. In recent years, the dominant design is either pure online
or multi-channel processes. In contrast, participatory budgeting relying on citizen forums only
was popular at the millennium turn but steadily lost relevance since then. Regarding the
possibilities of participation, in all years analyzed the dominating design is processes allowing
for own proposals as well as expenditures and, additionally, with no topical restrictions.
Our analyses of participation show that only three factors significantly correlate with
participation rates. First, the higher the share of people living on welfare in a given year, the
lower the participation rate is. This corroborates findings from other studies on the
relationship between social conditions and political participation. Secondly, the usage of the
internet has a positive effect on the participation. This comes as no surprise since it reduces
participation costs. Thirdly, we find a huge difference in participation rates between Eastern
and Western Germany. Ad hoc, differences in political culture and the structure of civil
society might offer an explanation.
One shortcoming of our data set in this respect is that the data available does not allow
for an analysis of temporal trends within a municipality. For example, it is interesting why
some municipalities suffer from declining participation rates while others manage to increase
participation steadily. Unfortunately, this would require detailed information on mobilization
efforts, media coverage etc. as well as more observations in our dataset to secure stable model
22
estimations. Nevertheless, such questions can be explored by using qualitative research
designs.
Further tasks regarding our data set and the quantitative study of participatory budgeting
in Germany are primary located in three areas: First, we plan to add further variables to the
cases included. For example, which actor(s) initiated participatory budgeting (parties,
municipal administration, NGOs, etc.), can critique from certain actors be observed, and
which amount of money was spent for the process in a given year. Of course, further variables
are desirable but in our opinion it seems illusory – as mentioned in the last paragraph - to
quantitatively capture, for example, mobilization efforts or the financial impact of budget
proposals, especially retrospectively. Secondly, we advocate the expansion of the dataset with
municipalities without participatory budgeting in a sufficient number to enable case-control
and matching studies. This provides opportunities for studies analyzing the factors that lead to
the adoption and abolition of participatory budgeting, either by means of a pure quantitative
approach or by selecting cases for comparative qualitative studies with a most-similar-
systems-different-outcomes design. Thirdly, we highly recommend – given appropriate
funding – a survey among participatory budgeting municipalities to fill gaps in our data base
and to validate data (for a similar approach see Galais et al. 2012). Of course, chances are
high that due to turnover in administrative and political staff, lacking official record-keeping,
and limited human cognitive capabilities, knowledge about past events might diminish over
time. Hence, detailed analyses are needed whether missing information (in particular, process
design and participation rates) for certain municipalities is random or systematically related to
characteristics of the municipality. If the latter is true, the results of any analysis might suffer
from biases.
A further task regarding research on participatory budgeting in Germany (and
elsewhere) is the large-scale assessment of its effects on local citizens, politics and public
administration. In our opinion, to complete this task, surveys among citizens with a sufficient
number of observations and purposively selected municipalities, preferably with a case-
control design to account for usually very low numbers of actual participants, are necessary.
The main reason for such an approach is on the one hand the possibility to measure assumed
effects validly and on the other hand the evasion of ecological fallacies. Our data set provides
a nice starting point for such endeavors.
23
Literature
Abelson, Julia./ Gauvin, François-Pierre (2006). Assessing the Impacts of Public Participation: Concepts, Evidence and Policy Implications. Ottawa: Canadian Policy Research Network.
Abers, Rebecca (1996). From Ideas to Practice. The Partido dos Trabalhadores and Participatory Governance in Brazil, in: Latin American Perspectives 23(4), pp. 35-53.
Anheier, Helmut K./ Priller, Eckhard/ Zimmer, Annette (2000). Civil Society in Transition: The East German Sector Ten Years after Unification. London School of Economics and Political Science Civil Society Working Paper 15. London: London School of Economics and Political Science.
Avritzer. Leonardo (2006). New Public Spheres in Brazil: Local Democracy and Deliberative Politics, in: Journal of Urban and Regional Research 30(3), pp. 623-637.
Bächtiger, André/ Niemeyer. Simon/ Neblo, Michael/ Steiner, Jürg/ Steenbergen, Marco R. (2010). Disentangling Diversity in Deliberative Democracy: Competing Theories, their Blind-spots, and Complementarities, in: Journal of Political Philosophy 18(1), pp. 32-63.
Baiocchi, Gianpaolo (2003). Participation, Activism, and Politics: The Porto Alegre Experiment, in: Fung, Archon/ Wright, Erik Olin (Hrsg.). Deepening Democracy. Institutional Innovations in Empowered Participatory Governance. London: Verso, pp. 45-76.
Beck, Nathaniel (2001). Time-Series Cross-Section Data: What Have We Learned in the Past Few Years?, in: Annual Review of Political Science 4, pp. 271-293.
Boulding, Carew/ Wampler, Brian (2010). Voice, Votes, and Resources: Evaluating the Effect of Participatory Democracy on Well-being, in: World Development 38(1), pp. 125-135.
Busse, Stefan/ Schneider, Sebastian H. (2014). Die Evaluation von Bürgerbeteiligung als Politikberatung: Praxis, Herausforderungen und Empfehlungen am Beispiel Bürgerhaushalt, in: Zeitschrift für Politikberatung, Policy Advice and Political Consulting 7(1/2), pp. 3-13 (in press).
Dahl, Robert A./ Tufte, Edward (1973). Size and Democracy. Stanford: Stanford University Press.
Dalton, Russell J./ Van Sickle, Alix/ Weldon, Steven (2010). The Individual-Institutional Nexus of Protest Behaviour, in: British Journal of Political Science 40(1), pp. 51-73.
Eich, Tom (2011). Der Bürgerhaushalt: Partizipation in der kommunalen Haushaltspolitik am Beispiel der Städte Freiburg und Köln. BA Thesis, Fernuniversität Hagen, Germany.
Elster, Jon (2007). Explaining Social Behavior. More Nuts and Bolts for the Social Sciences. New York: Cambridge University Press.
Ermert, Julian/ Pützer Hannah/ Ruesch, Michelle (2015): 8. Statusbericht des Portals Buergerhaushalt.org. Available online at http://www.buergerhaushalt.org/sites/default/files/downloads/Statusbericht_2015_Stand_24052015_aktualisierteVersion.pdf. [retrieved 11.08.2015]
Färber, Christine (2009). Geschlechtersensibler Beteiligungshaushalt. Ergebnisse und Empfehlungen für die Praxis. Weinheim: Dashöfer.
24
Font, Joan/ Galais, Carolina (2011). The Qualities of Local Participation: The Explanatory Role of Ideology, External Support and Civil Society as Organizer, in: International Journal of Urban and Regional Research 35(5), pp. 932-948.
Franzke, Jochen/ Kleger, Heinz (2006). Kommunaler Bürgerhaushalt in Theorie und Praxis am Beispiel Potsdams. Theoretische Reflektionen, zusammenfassende Thesen und Dokumentation eines begleitenden Projektseminars. Potsdam: Universitätsverlag Potsdam.
Franzke, Jochen/ Kleger, Heinz (2010). Bürgerhaushalte: Chancen und Grenzen. Berlin: Edition Sigma.
Franzke, Jochen/ Roeder, Eva (2014). Evaluation des Bürgerhaushaltes im Landkreis Mansfeld-Südharz. Potsdam: Universitätsverlag Potsdam.
Gabriel, Oscar W. (2005). Politische Einstellungen und politische Kultur, in: Gabriel, Oscar W./ Holtmann, Everhard (Eds). Handbuch politisches System der Bundesrepublik Deutschland. 3. Ed. München: Oldenbourg Verlag, pp. 459-522.
Galais, Carolina/ Font, Joan/ Alarcon, Pau/ Sesma, Dolores (2012). Methodological Challenges for the Large N Study of Local Participatory Experiences. Combining Methods and Databases, in: Revista Internacional de Sociologia 70(2), pp. 65-87.
Geddes, Barbara (1990). How the Cases You Choose Affect the Answers You Get: Selection Bias in Comparative Politics, in: Political Analysis 2(1), pp. 131-150.
Geißel, Brigitte (2013). Introduction: On the Evaluation of Participatory Innovations, in: Geißel, Brigitte/ Joas, Marko (Eds.). Participatory Democratic Innovations in Europe. Improving the Quality of Democracy? Opladen: Barbara Budrich Publishers, pp. 8-32.
Geißel, Brigitte/ Kolleck, Alma/ Neunecker, Martina (2013). Projektbericht „Wissenschaftliche Begleitung und Evaluation des Frankfurter Bürgerhaushaltes 2013“. Verfügbar unter: http://www.fb03.uni-frankfurt.de/46461594/Buergerhaushalt-Frankfurt---Evaluationsbericht_Final. [retrieved 11.08.2015]
Geißel, Brigitte/ Neunecker, Martina/ Kolleck, Alma (2015). Dialogorientierte Beteiligungsverfahren: Wirkungsvolle oder sinnlose Innovationen? Das Beispiel Bürgerhaushalt, in: Zeitschrift für Parlamentsfragen 46(1), pp. 151-165.
Gensicke, Thomas/ Geiss, Sabine (2010). Hauptbericht des Freiwilligensurveys 2009. Ergebnisse der repräsentativen Trenderhebung zu Ehrenamt, Freiwilligenarbeit und Bürgerschaftlichem Engagement. Berlin: Bundesministerium für Familie, Senioren, Frauen und Jugend.
Goertz, Gary/ Mahoney, James (2012). A Tale of Two Cultures: Qualtitative and Quantitative Research in the Social Sciences. Princenton: Princeton University Press.
Günther, Albert (2007). Der Bürgerhaushalt: Bestandsaufnahme - Erkenntnisse – Bewertung. Stuttgart: Boorberg.
Gurr, Ted (1968). A Causal Model of Civil Strife: A Comparative Analysis Using New Indices, in: American Political Science Review 62(4), pp. 1104-1124.
Held, David (2006): Models of democracy. 3. ed. Cambridge: Polity Press. Herzberg, Carsten (2009). Von der Bürger- zur Solidarkommune. Lokale Demokratie in
Zeiten der Globalisierung. Hamburg: VSA Verlag.
25
Herzberg, Carsten/ Sintomer, Yves/ Kleger, Heinz (Eds.) (2012). Hoffnung auf eine neue Demokratie. Bürgerhaushalte in Lateinamerika und Europa. Frankfurt/ Main: Campus Verlag.
Honaker, James/ King, Gary (2010). What to do About Missing Values in Time Series Cross-Section Data, in: American Journal of Political Science 54(3), pp. 561-581.
Inglehart, Ronald (1977). The Silent Revolution. Princeton: Princeton University Press. Inglehart, Ronald (1990). Culture Shift in Advanced Industrial Society. Princeton: Princeton
University Press. Inglehart, Ronald (1997). Modernization and Postmodernization. Princeton: Princeton
University Press. Kelleher, Christine A./ Lowery, David Lowery (2009). Central City Size, Metropolitan
Institutions, and Political Participation, in: British Journal of Political Science 39(1), pp. 59-92.
Kersting, Norbert/ Busse, Stefan/ Schneider, Sebastian H. (2013). Evaluationsbericht Bürgerhaushalt Jena. Münster: Westfälische Wilhelms-Universität Münster.
King, Gary/ Keohane, Robert O./ Verba, Sidney (1994). Designing Social Inquiry:Scientific Inference in Qualitative Research. Princeton: Princeton University Press.
Kittel, Bernhard (1999). Sense and Sensitivity in Pooled Analysis of Political Data, in: European Journal of Political Research 35(2), pp.225-253.
Kittel, Bernhard (2006). A Crazy Methodology? On the Limits of Macro-Quantitative Social Science Research, in: International Sociology 21(5), pp. 647-677.
Klages, Helmut/ Daramus, Carmen (2007). Bürgerhaushalt Berlin-Lichtenberg: partizipative Haushaltsplanaufstellung, -entscheidung und -kontrolle im Bezirk Lichtenberg von Berlin. Begleitende Evaluation des ersten Durchlaufs. Speyer: Deutsches Forschungsinstitut für öffentliche Verwaltung.
Laakso, Markku/ Taagepera, Rein (1979). „Effective“ Number of Parties. A Measure with Application to West Europe, in: Comparative Political Studies 12(1), S. 3-27.
Lieberman, Evan S. (2005). Nested Analysis as a Mixed-Method Strategy for Comparative Research, in: American Political Science Review 99(3), pp. 435-452.
Lupia, Arthur/ Sin, Gisela (2003). Which Public Goods are Endangered? How Evolving Communication Technologies Affect The Logic of Collective Action, in: Public Choice 117(3-4), pp. 315-331.
Masser, Kai/ Pistoia, Adriano/ Nitzsche, Philipp (2013). Bürgerbeteiligung und Web 2.0. Potentiale und Risiken webgestützter Bürgerhaushalte. Wiesbaden: VS Springer.
McCarthy, John D./ Zald, Mayer N. (1977). Resource Mobilization and Social Movements: A Partial Theory, in: American Journal of Sociology 82(6), pp. 1212-1241.
Melgar, Teresa R. (2014). A Time of Closure? Participatory Budgeting in Porto Alegre, Brazil, after the Workers' Party Era, in: Journal of Latin American Studies 46(1), pp. 121-149.
Rodgers, Dennis (2010). Contingent Democratisation? The Rise and Fall of Participatory Budgeting in Buenos Aires, in: Journal of Latin American Studies 42(1), pp. 127.
26
Röcke, Anja (2013). Forms of Democratic Innovations in European Participatory Budgeting. The Examples of Poitou-Charentes (France) and Berlin Lichtenberg (Germany), in: Geißel, Brigitte/ Joas, Marko (eds.). Participatory Democratic Innovations in Europe. Improving the Quality of Democracy? Opladen: Barbara Budrich Publishers, pp. 33-52.
Rohlfing, Ingo (2008). What You See and What You Get: Pitfalls and Principles of Nested Analysis in Comparative Research, in: Comparative Political Studies 41(11), pp. 1492-1514.
Rowe, Gene/ Frewer, Lynn (2000). Public Participation Methods: A Framework for Evaluation, in: Science, Technology, & Human Values 25(1), pp. 3-29.
Ryan, Matt/ Smith, Graham (2012). Towards a Comparative Analysis of Democratic Innovations. Lessons from a small-N fsQCA of Participatory Budgeting, in: Revista Internacional de Sociología 70(2), pp. 89-120.
Schäfer, Armin (2012a). Beeinflusst die sinkende Wahlbeteiligung das Wahlergebnis? Eine Analyse kleinräumiger Wahldaten in deutschen Großstädten, in: Politische Vierteljahresschrift 53 (2), pp. 240-264.
Schäfer, Armin (2012b). Consequences of Social Inequality for Democracy in Western Europe, in: Zeitschrift für Vergleichende Politikwissenschaft 6(1), pp. 23-45.
Schneider, Sebastian H. (2011). Bürgerhaushalt Oldenburg. Evaluation. Dokumentation. Oldenburg: Universität Oldenburg.
Schneider, Sebastian H. (2012). Bürgerhaushalt Oldenburg. Evaluation. Dokumentation. Oldenburg: Universität Oldenburg.
Schneider, Sebastian H. (2013). Bürgerhaushalt Oldenburg. Evaluation. Dokumentation. Oldenburg: Universität Oldenburg
Schnelle, Katja (2013). Neue Partizipationsformen auf kommunaler Ebene am Beispiel des Hildesheimer Bürgerhaushalts – Implikationen für eine politische Bildung in Zeiten des gesellschaftlichen Wandels. MA Thesis, University of Hildesheim, Germany.
Sintomer, Yves/ Herzberg, Carsten/ Röcke, Anja (2008). Participatory Budgeting in Europe: Potentials and Challenges, in: International Journal of Urban and Regional Research 32(1), pp. 164-178.
Sintomer, Yves/ Herzberg, Carsten/ Röcke, Anja (2010). Der Bürgerhaushalt in Europa – eine realistische Utopie? Zwischen partizipativer Demokratie, Verwaltungs-modernisierung und sozialer Gerechtigkeit. Wiesbaden: VS Verlag.
Sintomer, Yves/ Traub-Merz, Rudolf/ Herzberg, Carsten (2013). Introduction: Participatory Budgeting – A Global View, in: Sintomer, Yves/ Traub-Merz, Rudolf/ Zhang, Junhua (Eds.). Participatory Budgeting in Asia Europe. Key Challenges of Participation. Basingstoke: Palgrave, pp. 1-26.
Smith, Graham (2009). Democratic Innovations. Designing Institutions for Citizen Participation. Cambridge: Cambridge University Press.
Spada, Paolo (2014). The Diffusion of Participatory Governance Innovations: A Panel Data Analysis of the Adoption and Survival of Participatory Budgeting in Brazil. Working Paper. http://www.spadap.com/app/download/8877410668/The 20Diffusion 20of 20Democratic 20Innovations_sito 20web.pdf?t=1417288702 [retrieved 11.08.2015]
27
Sweeting, David/ Copus, Colin (2013). Councillors, Participation, and Local Democracy, in: Egner, Björn/ Sweeting, David/ Klok, Jan-Pieter (eds.). Local Councillors in Europe. Wiesbaden: VS Springer, pp. 121-137.
Talpin, Julien (2011). Schools of Democracy. How Ordinary Citizens (Sometimes) Become Competent in Participatory Budgeting Institutions. Colchester: ECPR Press.
Taubert, Nils/ Krohn, Wolfgang/ Knobloch, Tobias (2011). Evaluation des Kölner Bürgerhaushalts. Kassel: Kassel University Press.
Touchton, Michael/ Wampler, Brian (2014). Improving Social Well-Being Through New Democratic Institutions, in: Comparative Political Studies 47(10), pp. 1142-1469.
Van der Meer, Tom/ Van Deth, Jan W./ Scheepers, Peer (2009). The Politicized Participant: Ideology and Political Action in Twenty Democracies, in: Comparative Political Studies 42(11), pp. 1426-1457.
Vráblíková, Katerina (2014). How Context Matters? Mobilization, Political Opportunity Structures, and Nonelectoral Political Participation in Old and New Democracies, in: Comparative Political Studies 47(2), pp. 203-229.
Wampler, Brian (2007). Participatory Budgeting in Brazil: Contestation, Cooperation, and Accountability. University Park: Pennsylvania State University Press.
Wampler, Brian (2008). When Does Participatory Democracy Deepen the Quality of Democracy? Lessons from Brazil, in: Comparative Politics 41(1), pp. 61-81.
Wonka, Arndt (2007). Um was geht es? Konzeptspezifikation in der politikwissenschaftlichen Forschung, in: Gschwend, Thomas/ Schimmelfennig, Frank (Eds.): Forschungsdesign in der Politikwissenschaft. Probleme - Strategien - Anwendungen. Frankfurt/Main: Campus, pp. 63–89.
28
Appendix
Table A1: Cases included. Municipality Year Total
2006 2007 2008 2009 2010 2011 2012 2013
Aachen 0 0 0 0 0 2 0 0 2
Ahlen 0 0 0 0 0 0 1 2 3
Altenberge 0 0 0 0 0 0 1 0 1
Artland (Samtgemeinde)
0 0 0 0 0 1 0 0 1
Babenhausen 0 0 0 0 0 1 0 0 1
Bad Hersfeld 0 0 1 0 0 0 0 0 1
Bad Honnef 0 0 0 1 1 0 0 0 2
Bad Vilbel 0 0 0 0 0 0 0 1 1
Bad Wildungen 0 0 0 0 0 1 0 0 1
Bad Woerishofen
0 0 0 0 0 1 0 0 1
Bedburg 0 0 0 0 0 1 1 0 2
Beelen (Gemeinde)
0 0 0 0 0 1 0 1 2
Bergheim 0 0 1 0 0 0 0 0 1
Bernau 0 0 0 0 0 1 1 0 2
Bingen 0 0 0 0 0 0 1 0 1
Bischofsheim 0 0 0 0 0 0 1 0 1
Blankenfelde-Mahlow
0 0 0 0 0 1 1 0 2
Bonn 1 0 1 0 0 1 1 0 4
Borchen 0 0 0 0 0 1 1 1 3
Bottrop 0 0 0 0 0 0 1 0 1
Bruchsal 0 0 0 0 0 1 0 0 1
Buedingen 0 0 0 0 0 0 0 1 1
Celle 0 0 0 0 1 0 1 0 2
Chemnitz 0 0 0 0 0 0 1 1 2
Darmstadt 0 0 0 0 0 0 1 1 2
Diepholz 0 1 0 0 0 0 0 0 1
Dortmund 0 0 0 0 0 0 0 1 1
29
Duisburg 0 0 0 0 0 0 1 2 3
Eberswalde 0 0 0 0 0 0 1 1 2
Eislingen/Fils 0 0 0 0 0 0 1 0 1
Emsdetten 1 1 1 0 2 1 1 1 8
Ennepetal 0 0 0 0 0 1 1 1 3
Erfurt 0 1 1 1 1 0 0 2 6
Essen 0 0 0 0 1 1 0 0 2
Fehmarn 0 0 0 0 0 1 0 0 1
Frankfurt am Main
0 0 0 0 0 1 0 1 2
Freiburg im Breisgau
0 0 1 0 1 0 0 1 3
Fuerstenwalde 0 0 0 0 0 0 0 1 1
Garbsen 0 0 0 0 0 0 1 0 1
Geislingen an der Steige
0 0 0 0 0 1 0 0 1
Gelnhausen 0 0 0 0 0 0 1 0 1
Gera 0 0 0 0 0 0 0 1 1
Gladbeck 0 0 0 0 0 1 0 0 1
Goettingen 0 0 0 0 0 0 1 0 1
Großbreitenbach 0 0 1 1 1 1 0 0 4
Guetersloh 0 0 0 0 1 1 1 0 3
Halle (Saale) 0 0 0 0 0 0 0 1 1
Hamburg 1 0 0 1 0 0 0 0 2
Hattingen 0 0 0 0 0 1 0 0 1
Heidenrod 0 0 0 0 0 0 1 1 2
Heiligenhaus 0 0 0 0 0 1 0 1 2
Henstedt-Ulzburg
0 0 0 0 0 0 0 1 1
Herzebrock-Clarholz
0 0 0 0 1 1 1 0 3
Herzogenrath 0 0 0 0 0 1 0 0 1
Hilden 1 1 1 1 1 1 0 0 6
Hildesheim 0 0 0 0 0 0 1 1 2
Huerth 0 0 0 0 1 0 0 0 1
Iserlohn 0 0 0 0 0 1 0 0 1
30
Jena 0 1 1 1 1 1 1 1 7
Jueterbog 0 0 0 1 1 0 0 0 2
Kaarst 1 0 0 0 0 0 0 0 1
Koeln 0 1 0 0 1 1 0 1 4
Lampertheim 0 0 0 0 0 0 1 1 2
Leipzig 0 0 0 1 0 0 1 1 3
Loehne 0 0 0 0 0 0 0 1 1
Luckenwalde 0 0 0 0 0 0 1 0 1
Luedenscheid 0 0 0 0 1 0 0 0 1
Luenen 0 0 0 0 0 1 0 1 2
Mainhausen 0 0 0 0 0 1 1 0 2
Maintal 0 0 0 0 0 0 0 1 1
Marl 0 0 0 0 0 0 1 0 1
Meerbusch 0 0 0 0 0 0 1 0 1
Moers 0 1 0 0 0 1 1 0 3
Monheim am Rhein
0 0 0 0 0 0 2 1 3
Much 0 1 0 0 0 0 0 0 1
Muelheim an der Ruhr
0 0 0 0 1 1 1 1 4
Muenster 0 0 0 0 0 1 1 0 2
Netphen 0 0 0 0 0 0 0 1 1
Neustadt am Ruebenberge
0 0 0 0 0 0 1 0 1
Norderstedt 0 0 0 0 0 1 0 1 2
Nottuln 0 0 0 0 0 1 1 0 2
Oldenburg 0 0 0 0 1 1 1 0 3
Osnabrueck 0 0 0 0 0 0 1 1 2
Potsdam 1 1 1 1 1 1 1 0 7
Ratingen 0 0 0 0 0 0 1 0 1
Remscheid 0 0 0 0 0 0 1 0 1
Rheda-Wiedenbrueck
0 0 0 0 0 0 1 0 1
Schmoelln 0 0 0 0 0 0 0 1 1
Schoeneiche 0 0 0 1 1 1 1 1 5
31
Schortens 0 0 0 0 0 1 1 0 2
Schwalmtal 0 0 0 0 0 0 0 1 1
Seelze 0 0 0 0 0 1 0 0 1
Senftenberg 0 0 0 0 0 0 1 1 2
Siegburg 0 0 0 0 0 1 2 0 3
Solingen 0 0 0 0 1 0 1 0 2
Steinau an der Strasse
0 0 0 0 0 0 0 1 1
Strausberg 0 0 0 0 1 1 0 0 2
Stuttgart 0 0 0 0 0 0 1 1 2
Taunusstein 0 0 0 0 0 1 1 1 3
Toenisvorst 0 0 0 0 0 1 1 1 3
Trier 0 0 0 1 1 1 1 0 4
Wadersloh 0 0 0 0 0 1 1 0 2
Waldshut-Tiengen
0 0 0 0 0 1 0 0 1
Weimar 0 0 0 0 1 1 0 0 2
Wesel 0 0 0 0 1 1 0 1 3
Westerkappeln 0 0 0 0 0 0 0 1 1
Wiesbaden 0 0 0 1 0 0 0 0 1
Wildeshausen 0 0 0 0 0 1 1 1 3
Wolgast 0 0 1 1 1 0 0 0 3
Worms 0 0 0 0 0 1 1 1 3
Wuppertal 0 0 0 0 0 0 1 0 1
Zwickau 0 0 0 0 0 0 1 1 2
6 9 11 13 26 53 60 50 228
Note: In regression models, the number of valid cases reduces due to missing values.
32
Table A2: Descriptive statistics (N valid = 157 – 228). Variable Mean (SD)
Dependent variable
Participation rate (%)
1,20 (1,60)
Independent variables (metric)
Population
135.824,12 (237.127,93)
Share People on Welfare (“SGB 2”) (%) 10,91 (4,77)
Share People Migration Background (%) 8,43 (4,72)
Share Councilors left parties (%) 46,84 (12,88)
Laakso-Taagepera-Index 3,88 (0,85)
Financial Situation (Euros per capita) 672, 24 (1.096,20)
Turnout Last Local Election (%) 50,54 (6,66)
Independent variables (categorical)
Internet Use
Percentages
No 13,8 %
Yes 86,2 %
Own Proposals Possible No 7,6 %
Yes 92,4 %
Expenditures Possible No 20,9 %
Yes 79,1 %
Region Western Germany 74,6 %
Eastern Germany 25,4 %
Note: Reference category in italics.
33
Figure A1: Bivariate Plots independent vs. dependent variable(s).
Note: Linear regression line included. Analysis of outliers, unusual cases and non-linear relationships not yet completed.
34
Table A3: Random Intercept Model for Longitudinal Data (first-order autoregressive error structure for varying time intervals).
Random Intercept
Random Intercept + Random time slope
(Intercept) -2.45 -2.45 (1.42) (1.42)
Number of PB Processes 0.06 0.06 (0.08) (0.08)
Population -0.07 -0.07 (0.10) (0.10)
Share People on Welfare („SGB 2“) -0.24 -0.24 (0.13) (0.13)
Share People Migration Background -0.12 -0.12 (0.11) (0.11)
Share Left Parties -0.03 -0.03 (0.10) (0.10)
Laakso-Taagepera Index 0.01 0.01 (0.11) (0.11)
Financial Situation -0.08 -0.08 (0.10) (0.10)
Internet Use Yes 1.01**
1.01**
(0.35) (0.35)
Own Proposals Possible Yes -0.28 -0.28 (0.52) (0.52)
Expenditures Possible Yes -0.02 -0.02 (0.28) (0.28)
Turnout Last Election -0.02 -0.02 (0.12) (0.12)
Eastern Germany 1.79***
1.79***
(0.36) (0.36)
SD Intercept Municipality (Residual)
0.000294609 (1.301418)
0.0002850372 (1.301418)
SD Number of PB Processes - 0.000000007
AIC 525.84 529.84
BIC 572.32 582.13
Log Likelihood -246.92 -246.92
N 148 148
Municipalities 80 80 Notes: ***p < 0.001, **p < 0.01, *p < 0.05. All metric variables were group mean-centered and z-standardized to avoid estimation problems. Restricted Maximum Likelihood Estimation. (REML).