Analysing Policy Proximity Through Media Reporting
Christoph Knill, Yves Steinebach & Bastian Buitkamp
Published in Der Moderne Staat (DMS)
Please cite as:
Knill, C., Steinebach, Y. & Buitkamp, B. (2022). “Analysing Policy Proximity through media
reporting.” dms – der moderne staat – Zeitschrift für Public Policy, Recht und Management,
online first, 15(2-2022), 1-18. https://doi.org/10.3224/dms.v15i2.02
Analysing Policy Proximity Through Media Reporting
Policy changes in one subsystem can easily spill over to other subsystems. An approach that
addresses these interconnections is the concept of ‘policy proximity’. This concept posits that
different policy issues share common features that make them more or less likely to change
together. Unfortunately, however, we have no systematic knowledge of the proximity between
policy areas. In this article, we address this shortcoming by proposing a novel measurement
concept of policy proximity that captures the proximity between different policy issues based
on their joint appearance in media reporting. To do so, we conduct a relational content analysis
of all media reports aired by the German news broadcast ‘Tagesschau’ between the years
2013 to 2021. We show that policy issues substantially differ in their connectivity with other
subjects and identify for each subsystem the closest ‘neighbors’. We conclude by discussing
our results in light of existing policy change theories.
policy proximity; policy change; relational content analysis
Eine Analyse von ‚Policy Proximity‘ durch Medienberichterstattung
Veränderungen in einem Policy Subsystem können sich leicht auf andere Subsysteme
auswirken. Ein Ansatz, der sich mit diesen Zusammenhängen befasst, ist das Konzept der
‚Policy Proximity‘. Dieses Konzept geht davon aus, dass verschiedene politische Themen
gemeinsame Merkmale aufweisen, die es wahrscheinlicher oder unwahrscheinlicher machen,
dass diese sich gemeinsam verändern. Leider haben wir bisher keine systematischen
Kenntnisse über die exakte Nähe zwischen verschiedenen Politikbereichen. Dieser Artikel
adressiert diese Forschungslücke, indem wir ein neuartiges Messkonzept der politischen Nähe
vorschlagen, welches die Nähe zwischen verschiedenen politischen Themen auf der
Grundlage ihres gemeinsamen Auftretens in der Medienberichterstattung erfasst. Dazu führen
wir eine relationale Inhaltsanalyse aller Medienberichte durch, die von der deutschen
‚Tagesschau‘ zwischen 2013 und 2021 ausgestrahlt wurden. Wir zeigen, dass sich politische
Themen in ihrer Verknüpfung mit anderen Themen deutlich unterscheiden und identifizieren
für jedes Teilsystem die engsten "Nachbarn". Abschließend diskutieren wir unsere Ergebnisse
mithilfe bestehender Theorien zum Politikwandel.
Policy Proximity; Politikwandel; relationale Inhaltsanalyse
In response to the pandemic, governments have developed bundles of new policy measures
and reforms (Goyal & Howlett, 2021). What started as a public health issue, has quickly
cascaded into policy reforms in all kinds of areas. The public health measures required a
shutdown of the economy and public life. The shutdown, in turn, required compensatory
measures for both businesses and employees as well as massive investments into the digital
infrastructure to facilitate home-schooling and distance learning.
This simple observation indicates that patterns of policy change and stability do not
occur independently across different policy subsystems: Policy development in one policy
subsystem might directly or indirectly affect the patterns of policy change in other domains.
Yet, existing theories of policy change are strikingly silent when it comes to the analysis of this
phenomenon. This is due to the ‘subsystem bias’ that characterizes the existing policy change
scholarship. Policy changes are studied with a strong focus on individual policy subsystems.
These fields are the central analytical point of reference. Everything that happens in fields
other than the one under investigation is deliberately neglected.
One of the few public policy concepts that takes account of the interdependencies
across different policy subsystems is the concept of policy proximity developed by Nohrstedt
and Weible (2010). Policy proximity captures the similarity of policies with regard to their
“statutes, laws, and policies, including the instruments, ideas, and symbols therein” (p. 20). As
a result of this policy proximity, some policy subsystems are expected more likely than others
to change together in response to common crisis events and other political developments. The
concept of policy proximity has become widely used in empirical (mostly qualitative) studies.
Unfortunately, however, the proximity of different policy issues is typically determined in an ex-
post assessment. In other words: policy subsystems are said to be proximate because they
had changed together – not because scholars had a pre-existing knowledge about policy
subsystems’ interconnections and commonalities (but see May et al., 2013).
In this article, we address this shortcoming and propose a novel measurement concept
of policy proximity that captures the ‘distance’ between different policy subsystems based on
their joint appearance in media reporting. To do so, we conduct a relational content analysis
of 27.000 media reports aired by the German news broadcast ‘Tagesschau’ between the years
2013 to 2021. We show that policy issues differ in their connectivity to other issue areas.
Moreover, we identify the ‘closest’ neighbours to each policy subsystem and briefly discuss
our findings in light of the dominant theories on policy change.
Our contribution is a first step towards improving our scientific capacities in predicting
and explaining policy change events within and across different policy subsystems. The
remainder of this article is structured as follows. In the following section, we briefly discuss the
existing literature on policy change and the ‘subsystem bias’ characterizing the existing
literature (section 2). Thereafter, we introduce our novel measurement approach of policy
proximity and specify the context of analysis (media outlets) as well as the policy issues under
scrutiny (section 3). In section 4 we present the empirical patterns of policy proximity.
Moreover, we check how our findings are influenced by the occurrence of major crisis events
such as the 2015 and 2016 refugee crises and the 2020 and 2021 COVID-19 pandemic.
Section 5 discusses how we can theoretically account for different degrees of policy proximity
across subsystems, building on central theoretical frameworks of policy change. The last
section concludes with a broader discussion of how our measures of policy proximity can help
to improve the study of policy change.
2. Existing Account of Policy Change and the ‘Subsystem Bias’
The literature knows several theoretical frameworks for the analysis of policy change (for an
overview see Weible, et al. 2017). According to the seminal piece by Capano (2009) on
“Understanding Policy Change As An Epistemological and Theoretical Problem”, the most
important ones are the Advocacy Coalition Framework (ACF) (Sabatier & Weible, 2007), the
Multiple Streams Approach (MSA) (Kingdon, 2003), and the Punctuated Equilibrium Theory
(PET) (Baumgartner & Jones, 1993, Baumgartner et al., 2009).
The frameworks consider
different factors as particularly relevant for the explanation of policy change (for an overview
see Capano, 2009).
Despite the differences in the explanation of policy chance, all three theoretical
frameworks consider so-called ‘policy subsystems’ as the central unit of analysis (Fernandéz-
í-Marin et al., 2020; Knill & Steinbach, 2022). Policy subsystems are defined by a substantive
issue area (domain), a geographical scope, and a relatively stable set of actors that interact
within well-defined institutional boundaries (Cairney & Weible, 2015). The rationale behind this
strong subsystem focus is that the typical mode of policy-making is considered to be driven by
dynamics that are endogenous to the subsystem under scrutiny. The policy agenda is usually
controlled by stable networks of politicians, bureaucrats, and interest group representatives
operating in a given domain (Burstein, 1991). As a result, policies display high stability over
time and undergo only incremental change. It is only rarely the case that this isolated everyday
life of policy subsystems is distorted.
Yet, regardless of the prevalence of stability or punctuations, the crucial point of
reference in all policy studies is the subsystem level (Jochim & May, 2010). Anything that
happens in other subsystems is considered analytically irrelevant as long as there are no
straightforward spill-over effects between policy subsystems (May et al.. 2011). In short, the
hardly challenged assumption in the studies of policy change is that policy subsystems do ‘live
next to each other’ and do hardly meet and influence each other in practice. The possibility
that policy developments across subsystems are connected has largely been neglected in
public policy research (but see Fernández‐i‐Marín et al., 2019; May et al., 2013). Considering
this state of the literature, McGee and Jones (2019) emphasize that the focus on policy
subsystems is “sorely in need of updating in light of current developments in the policymaking
Capano (2009) also discusses path dependence but admits that “it does not strictly belong to
the specific literature on policy change” (p. 19).
process (…). Issues have become more complex as governments address more problems and
these problems interact with one another” (p. 139).
A promising starting point for such a perspective is the concept of policy proximity
(Nohrsted & Weible, 2010). The authors posit three aspects. They assume that (1)
“[s]ubsystem boundaries are artificial constructs” (Nohrsted & Weible, 2010, p. 8); that (2)
“policy subsystems are interdependent” (ibid: 8); and (3) that they jointly change in response
to common stimuli such as crisis events (ibid: 9). The extent to which policy subsystems are
connected, in turn, depends on the degree of policy proximity, i.e. “the degree that policy
subsystems share policy design components—such as a subsystem’s statutes, laws, and
policies, including the instruments, ideas, and symbols therein” (ibid: 20). Put simply,
environmental policies, for instance, will always be more likely to involve and trigger reforms
in energy policy than in the subsystem of migration or defence.
The concept of policy proximity has become widely used in empirical (mostly
qualitative) studies (see e.g. Chui et al., 2014; Sichling, 2020; Hurka, 2017). Yet, it is not the
case that the (level of) proximity is assessed on the basis of some subsystem features. Rather,
subsystems are assessed as proximate because they had changed together. Given these
problems of endogeneity, the explanatory and predictive potentials of the concept are limited
in practice. To improve the applicability of the concept of policy proximity, we need (1) a
broader and more general understanding of the interconnections between different policy
subsystems and (2) one that goes beyond the assessment of individual policy change events.
In the following, we propose a novel measurement concept of policy proximity that captures
the ‘distance’ between different policy subsystems based on their joint appearance in media
3. Measuring Policy Proximity
There are multiple ways to assess (thematic) proximity and relations (Armborst, 2017). In this
article, we determine proximity by analysing the frequency and patterns of thematic joint
appearances. This approach assumes that issues that are frequently and regularly mentioned
in the same thematic and temporal context share common features. Our underlying measuring
approach is thus a ”relational content analysis” (Bos & Tarnai, 1999; Armborst 2017). A
relational content analysis involves (1) the definition of ‘objects’ and (2) an assessment of their
connections. A relational content analysis is not a mixed methodology (in the strict sense of
the term) as it does not necessarily require the collection of both quantitative and qualitative
data (Creswell & Plano Clark 2011). Yet, the ‘mix’ part comes in as it requires the detection
and interpretation of qualitative information and its transformation into values, i.e. the numeric
indication of the co-occurrence of different ‘objects’.
For the qualitative assessment, it is necessary to take two analytical decisions: First,
we must specify the exact context in which to assess the (co-)occurrence of different policy
issues. Second, we need to decide on which policy subsystems the analysis focuses on and
how these issues can be systematically distinguished from one another.
3.1 Data Source
In the abstract, there are multiple possibilities in which the thematic (co-)occurrence of different
policy issues can be studied, ranging from, for instance, speeches in the parliament to reports
in media outlets. A potential challenge posed by the first option is that speakers (Members of
the Parliament) from varying party backgrounds might differ in the way they discuss and thus
connect different policy issues. This is not a problem per se but can create difficulties (biases)
for the analysis if the speaking time on the floor is not equally distributed across the different
parties in parliament (Proksch & Slapin, 2012; Döring, 1995). In the context of this analysis,
we thus decided to focus on reports in media outlets. Here, we concentrate on TV news. A key
benefit of reports in TV news – compared to newspaper articles – is that they are usually quite
concise and of comparable length.
This study analyses news items from the main German public-service news bulletin,
the ARD’s Tagesschau. The main edition of the Tagesschau is shown at 08:00 p.m., lasts
fifteen minutes, and typically involves about ten individual reports. From the beginning, the
Tagesschau was considered the German news program that pays particular attention to
political topics. In comparison to other German TV news, the Tagesschau (still) has the highest
percentage of political topics among its reports (Krüger & Zapf-Schramm, 2017: 65). This
makes the Tagesschau the most appropriate news outlet for our analysis.
Our initial idea was to access the individual reports via the online media library of the
Tagessschau. Yet, this approach has proven difficult given that (1) the online media library is
slightly incomplete and (2) watching all news reports turned out to be a very time-consuming
activity. In consequence, we contacted the scientific department of the ARD and asked for their
support. The department gratefully shared different datasets containing information on each
show and report for the years 2013 to 2021. These datasets contain information on, amongst
others, the date and length as well as a short description of the content of each report. This
way, we have been able to access and assess a total of about 27.000 news reports. We deem
our investigation period appropriate for two reasons. First, we cover a total of nine years. This
allows us to get a comprehensive picture that goes beyond some short-time co-occurrences.
Second, we cover the most recent time period. As discussed in more detail below, the
interpretation of policy issues might change over time and in response to varying contextual
conditions. Focus on the years 2013 to 2021 thus gives us an impression of the ‘current’
connections between issues and issue areas.
We are well aware of the fact that Tagesschau reports are not a completely objective
source and measure of policy proximity given that, obviously, the ARD editors might have their
own (political) incentive to present policy issues in a certain way. However, we think that the
relatively short length of the individual reports limits the editors’ ability to frame issues in their
own sense. In other words, while the interconnection observed (see below) might indeed reflect
the editors’ own interpretations of policy issues, these interpretations must focus on the most
straightforward and obvious connections to other policy issues.
3.2 Identifying Relevant Policy Subsystems
Overall, we focus on a list of 20 policy issues. We rely on the main issue categories used in
the Comparative Agenda Project (CAP) codebook developed by Bevan (2019). This ensures
the compatibility of our approach with other scholarly contributions and thus the later
applicability. Table 1 provides a list of all policy issues under scrutiny as well as a short
description and example.
Table 1: List of all policy issues under scrutiny including short description and example.
Reports on agricultural trade, subsidies to Farmers, food inspection & safety,
marketing & promotion, animal and crop disease, fisheries & fishing, R&D on
Nitrate pollution in groundwater and slurry regulations of the Minister of Agriculture
Reports on minority/gender/age/handicap discrimination, age discrimination,
voting rights, freedom of speech, right to privacy, anti-government
Tightening state Corona measures in the area of contact restrictions due to rising
Reports on culture-related aspect (e.g. reports on museums, exhibitions, etc),
religious celebrations (e.g. Hajj), and commemoration ceremonies
First newly built mosque in Thuringia (13.11.2018)
Reports on alliances, intelligence, nuclear arms, military aid, personnel issues,
procurement, installations & land, reserve forces, civilian personnel, contractors,
foreign operations, claims against, military
Taliban invasion of Kabul and evacuation of personnel by Western states (15.08.2021)
Reports on banking, securities & commodities, consumer finance, insurance
regulation, bankruptcy, corporate management, small businesses, copyrights
and patents, disaster relief, tourism, consumer safety, sports regulation
Lufthansa Executive Board approves conditions for government financial aid
Reports on higher education, student loans and education finance, elementary
and primary schools, school reform, safety in schools, efforts to generally
improve educational standards and outcomes, education of underprivileged
students, vocational education for children and adults and their impact, special
education and education for the physically or mentally handicapped, education
excellence, including efforts to increase the quality of specific areas, such as
math, science or foreign language skills.
Promotion and expansion of state education projects for illiterate people (28.11.2016)
Reports on nuclear, electricity, natural gas & oil, coal, alternative & renewable,
energy conservation, R&D on energy
Approval of the reform of the Renewable Energies Act by the Bundesrat and Bundestag
Reports on drinking water, waste disposal, hazardous waste, air pollution,
recycling, indoor hazards, species & forest, conservation, climate change
Community of nations adopts climate agreement at Paris climate summit (12.12.2015)
Reports on health care reform, insurance, drug industry, medical facilities,
insurance providers, medical liability, manpower, disease prevention, infants and
children, mental, long-term care, drug coverage and cost, tobacco abuse, drug
and alcohol abuse, R&D
Corona Infection incidence in Germany and government hygiene measures (18.08.2020)
Reports on community development, urban development, rural housing, rural
development, low-income assistance, elderly homeless
Decrease in social housing with fixed prices in Germany and expansion of subsidies for
Reports on foreign aid, resources exploitation, developing countries, international
finance specific country, human rights treaties, conventions and violations,
human rights, organizations
terrorism, and diplomats
European Union heads of government discuss future relationship with the United
Kingdom after Brexit (15.10.2020)
Reports on worker safety, employment training, employee benefits, pensions,
and retirement accounts, including government-provided unemployment
insurance, labour unions, fair labour Standards, youth employment, child labour
and job training for youths, migrant and seasonal workers
Labour market figures, unemployment rate and measures to boost employment
Law & Crime
Reports on agencies, white collar crime, illegal drugs, court administration,
prisons, juvenile crime, child abuse, family issues, civil code, crime control,
Differences in the federal government with regard to an application for a ban on the NPD
Reports on interest rates, unemployment rate, , monetary policy, national budget,
tax code, industrial policy, price control
The Federal Constitutional Court rules that the European Central Bank did not disregard
European law when creating the European Banking Union (30.07.2019)
Reports on migration, refugees, and citizenship
Discussion on internal security and migration as well as deportation of convicted
Reports on national parks, forest management, water resources, territorial and
dependency issues and devolution
Search for suitable areas for nuclear waste disposal sites (28.09.2020)
Reports on low-income assistance, elderly assistance, disabled assistance,
volunteer associations and childcare
Ministers' assessment of the family support program "Elterngeld Plus” (10.01.2018)
Reports on space, commercial use of space, science transfer,
telecommunication, broadcast, weather forecasting, computers
Government strategy for artificial intelligence at the digital summit (04.12.2018)
Reports on trade agreements, exports, private investments, competitiveness,
tariff & imports, exchange rates
Merkel's trip to Beijing to enforce German interests in Sino-German trade relations
Reports on mass transport, highways, air travel, railroad travel, maritime,
infrastructure, R&D on new means of transportation
The Bundestag approves a toll law revised in response to pressure from the EU
Source: Own descriptions and examples based on Bevan (2019).
Our final coding approach was thus to go through all 27.000 news reports and to
manually assign them to one or more of the thematic issues identified. We have split the
workload between the authors to make the high workload manageable. This process comes
with two challenges. The first one is that different coders may assign the same report to
different policy issues. To address this issue, we randomly selected 50 news reports and coded
them independently by different coders. We then checked for intercoder reliability. Overall, 87
percent of the issues were assigned consistently, indicating a high to a very high level of
agreement (please note that a single report might address multiple issues). The second
challenge is that the short summaries provided by the Tagesschau might give a wrong
impression about the full content of the news report. In consequence, we randomly selected
five news reports, accessed them through the Tagesschau media library, fully transcribed
these reports, and coded them on the basis of our transcription. In the last step, we compared
this coding with our initial one based on the short summaries. Based on the full transcription,
we identified 13 different issue areas. 11 of these 13 issue areas (85 percent) have already
been captured by our initial coding based on the summaries. In essence, this implies that the
short summaries give a fairly good impression of the full content of the news reports.
Figure 1: Total nr. of occurrences of different policy issues in media reports
Source: Own data.
Overall, the policy issue most present in news reports is the issue of “International Affairs”. The
policy issue with the lowest frequency in news reports is “Housing” (see Figure 1 for an
overview). 6875 (25.2%) of the reports assessed concerned a single policy subsystem, 6733
(24.6%) two policy subsystems, 3464 (12.7%) three policy subsystems, and 4051 (7%) four or
more subsystems. In the case of 8411 reports (30%), no policy subsystem could be identified
(mostly sports news or daily events without direct political relevance). A good example of
multiple coding is the report on “Lockdown of the Economy due to COVID-19” shown on the
10th January 2021. It is coded as “Macroeconomics”, “Civil Rights”, “Health”, and “Domestic
Commerce”. The report discusses restrictions on the personal freedoms of all citizens due to
the outbreak of COVID-19 as well as support measures for the economy. Due to the restrictions
of personal freedoms, which are introduced with the explicit justification to contain the COVID-
9 pandemic, the report falls into the categories of “Health” and “Civil Rights”. The issues of
“Macroeconomics” and “Domestic Commerce” are included as the article also reports on
additional government borrowing (“Macroeconomic”) as well as on compensatory measures
for national enterprises (“Domestic Commerce”).
4. Empirical Patterns of Policy Proximity
In the previous section, we briefly described how often certain policy issues are mentioned in
media reports and whether they generally tend to be discussed alone or in combination with
others. The next analytical step is to combine this information and check whether policy issues
vary in their connectivity to other subsystems. To assess this relationship, we count all
combinations in which a policy subsystem has been mentioned together with another issue
area and divide it by the number of single occurrences, i.e., reports in which only a single policy
subsystem was discussed. This latter step is necessary to avoid issues that exhibit a higher
level of connectivity simply because there are generally mentioned more frequently in news
reports (for the differences in the frequency of occurrences please consult again Figure 1).
Figure 2 shows the respective results. Here, a value greater than ‘1’ indicates that issues of a
given policy subsystem occur more often in combination with others than alone. For values
smaller than ‘1’ the exact opposite is true.
Figure 2: Total connectivity per policy subsystem
Source: Own data.
Using a visual inspection, we can roughly identify four groups of policy subsystems
based on their total connectivity score, i.e. the ratio between multiple and single occurrences.
The first group with the greatest connectivity are “Energy” (2.21), “Labour” (2.19), “Foreign
Trade” (2.17), and “Domestic Commerce (2.11). This group is followed by a second, larger
group that comprises “Technology” (1.97), “Agriculture” (1.88), “Macroeconomics” (1.88),
“Education” (1.85), “Social Welfare” (1.84), “Housing” (1.81), “Civil Rights” (1.78) and “Health”
(1.73). At the lower end, we find a third group that includes “Transportation” (1.65), “Public
Lands” (1.61), “Law and Crime” (1.53), “Environment” (1.47), “Defense” (1.43), “International
Affairs” (1.40), and “Migration” (1.28). Remarkably, however, it is only “Culture” (0.38) that is
mentioned substantially more often alone than in combination with issues from other policy
subsystems. For all other issue areas, the ratio between pairwise and single occurrences is
constantly greater than “1”.
While Figure 2 provides us with information on the general connectivity of different
policy subsystems, it tells us little about which exact (pairs of) policy subsystems can be
considered proximate to one another. Table 2 thus lists the ‘Best’, ‘2nd Best’ and ‘3rd Best
Buddies’ of each policy subsystem, i.e., the policy subsystems with the first, second, and the
third greatest number of joint appearances in news reports. A brief look at Table 2 reveals that
‘International Affairs’ and ‘Law and Crime’ have the most joint appearances with issues of other
policy subsystems. Remarkably, however, these two fields only rank in midfield when it comes
to their total connectivity (see again Figure 2). These diverging observations essentially
indicate that these policy subsystems typically emerge only in combination with one (or some)
rather than multiple policy subsystems.
While the proximity of some subsystems is intuitive, it is less clear why others occur as
"best buddies". While no general answer can be given on why some areas are connected, a
more detailed view of less straightforward cases might allow controlling for the general
plausibility of our approach. A connection that does not seem very plausible at first sight, for
instance, is the connection between “Social Welfare” and “International Affairs”. This is
especially the case as welfare state programs are typically not considered a matter of
international concern (in contrast to, for instance, “Energy”, “Foreign Trade”,
“Macroeconomics”, and “Migration”). When looking at the respective connections, however, it
becomes clear that the majority of connections between “Social Welfare” and “International
Affairs” have occurred in the context of EU cooperation programs on youth unemployment
well as in the context of the international negotiations involving the European Commission
(EC), the European Central Bank (ECB) and the International Monetary Fund (IMF) on the
austerity measures in Spain, Portugal and especially Greece.
Overall, this finding highlights
that special circumstances can easily link otherwise rather disparate policy areas.
See e.g. report on “German-Spanish agreement against youth unemployment” (21st May
2013); report on “EU Brussels: Youth Unemployment Summit” (27th June 2013)
See e.g. report on “Greece: Austerity measures and aid package” (6th March 2015); report
on “German reactions to the failure of negotiations with Greece” (27th June 2015).
Table 2: Policy proximity per policy issue
2nd Best Buddy
3rd Best Buddy
International Affairs (82)
Law and Crime (1715)
International Affairs (936)
International Affairs (97)
Law and Crime (96)
Civil Rights (90)
International Affairs (2689)
Law and Crime (1564)
Civil Rights (572)
Law and Crime (228)
Civil Rights (80)
International Affairs (172)
Public Lands (434)
International Affairs (650)
Law and Crime (855)
International Affairs (683)
Civil Rights (552)
Law and Crime (52)
Civil rights (19)
Law and Crime (1649)
Civil Rights (378)
Domestic Commerce (201)
Law and Crime
Civil Rights (1715)
International Affairs (1649)
International Affairs (1093)
International Affairs (895)
Law and Crime (405)
Civil Rights (181)
International Affairs (118)
International Affairs (110)
International Affairs (534)
Law and Crime (451)
Law and Crime (310)
A potential critique of our measurement approach could be that our results are driven by single
crisis events and that, in consequence, the observed interconnections presented in Table 2 do
only hold true for parts of the observation period. The dominance of the COVID-19 pandemic
in the years 2020 and 2021, for instance, might bias the interconnections between “Health”
and other policy issues. In other words, the occurrence of the COVID-19 pandemic might give
the impression that “Health” is generally connected to “Law and Crime”, “International Affairs”
and “Civil Rights”, while, in reality, these findings are primarily driven by the peculiarities of the
COVID-19 pandemic as an international health crisis that had strong repercussion on individual
civil liberties. The same can be said about the issue of “Migration” for the refugee crisis in the
years 2015 and 2016. We thus decided to replicate our analysis by dropping all occurrences
of issues of “Migration” in the years 2015 and 2016 and issues of “Health” in the years between
2020 and 2021. In total, we excluded about 2200 observations (about eight percent of our total
Remarkably, this process has no influence on the policy subsystems considered most
closely connected to migration matters. Yet, it alters our findings with regard to health issues.
When dropping the years 2020 and 2021, “Education” is no longer linked to “Health” but to
“Labour”. Likewise, “Health” is less discussed in the context of “Civil Rights” and “International
Affairs” but as a matter of “Environment” and “Macroeconomics”. Moreover, “Domestic
Commerce” is no longer linked to “Health” but to “Law and Crime”. These changes essentially
reveal two aspects: On the one hand, they demonstrate the plausibility of our approach as the
COVID-19 pandemic has significantly affected the education sector due to the closing of
schools and – during the lockdowns – the citizens’ civil liberties and national enterprises. On
the other hand, the changes also demonstrate that the connections between policy
subsystems can be influenced by distinct crisis events and thus might change and vary over
time. Any assessment of policy proximity must thus take account of (or control for) the temporal
embeddedness and context (see discussion section).
A potential criticism of the approach could be that entirely dropping “Migration” and
“Health” might exclude issues in the respective years that are not crisis-related but still relevant.
In consequence, we checked again all Migration and Health occurrences in the years 2015
and 2016, and 2020 and 2021 respectively, and coded whether the respective issues have
been discussed in the context of either the COVID-19 pandemic or the refugee crisis using a
simple dummy variable. This procedure leaves our previous results unchanged. In the years
2015 and 2016, 85 percent (1251 out of 1481 mentions) of all issues on “Migration” occurred
in the context of the refugee crisis. For health matters, this observation is even more
pronounced for the years 2020 and 2021. Here, 94 percent (749 out of 798) of all “Health”
issues were discussed in the context of the COVID-19 pandemic.
In sum, the findings correspond quite well to the starting point of this paper, namely that
boundary-spanning policy problems and change events are the rule rather than the exception.
At the same time, however, we also see policy subsystems substantially differ in their
connectivity to other subsystems (see again Figure 2) and that policy issues align with some
issues more than with others (see again Table 2). As a consequence of these findings, we can
expect that some areas are more prone to cause changes in other areas or to change with
others in response to a common stimulus. Following the above analysis, changes in “Energy”
and “Labour”, for instance, are just far more likely to involve broader policy changes than
culture-related aspects. Moreover, our findings provide some indications about where exactly
we can actually expect subsystems’ interconnections. Based on our findings, scholarly
contributions that intend to explain changes in “Agriculture” policy, for instance, might find it
helpful to check for the latest developments in “Health”, “International Affairs”, and the
“Environment” and consider policy dynamics in these areas as a potential explanation.
5. Theoretical Perspectives on Policy Proximity
As discussed above, the connectivity of policy subsystems constitutes a blind spot of
conventional theories of policy change, such as the ACF, the MSA, or the PET. This is simply
due to the fact that these frameworks are deeply rooted in a subsystem perspective. Yet,
despite their analytical neglect of policy proximity, it might still be possible to ‘distil’ potential
factors from these frameworks that serve as a starting point for developing a more systematic
theoretical understanding of the determinants of policy proximity. This exercise might help us
to both (1) discuss our empirical findings in light of the dominant change theories as well as to
(2) complement the respective approaches to explicitly consider interconnections across policy
5.1 Policy proximity from the ACF perspective
In the ACF, the central theoretical gateway for cross-subsystem influences is the composition
of advocacy coalitions. Depending on the extent to which advocacy coalitions consist of
members whose interests and beliefs are related to a broader spectrum of policy subsystems,
interconnections of change across subsystems might be more or less likely.
In this context, Bandelow (2015) argues that especially in parliamentary systems, party
leaders and parliamentary group leaders often exert a lasting impact on decisions in policy
subsystems. These ‘generalists’ have the ability to combine issues from different policy
subsystems. Moreover, the formation of coalitions does not necessarily have to be oriented
along lines of conflict within a subsystem. Party-political coalitions might follow their own rules,
which cannot be limited to the content of a single policy issue or field (Bandelow, 2015: 308).
Based on these considerations, one can argue that policy areas become interconnected when
they are subject to broader political interests and conflicts. This is particularly the case if policy
areas ‘fall’ on the same similar party-political cleavages. This insight might explain why we find
strong interconnections in our analysis between those policy areas that are typically referenced
in the context of the left-right divide such as labor, social welfare, and macroeconomics.
Another factor that might explain cross-subsystem interdependencies within the ACF
is the role of policy brokers. Policy brokers are considered well-connected actors that seek
stability and “can be identified as a path to learning processes across coalitions in conflict
situations” (Ingold & Varone 2012: 3). While policy brokers are typically considered subsystem
actors, it is well possible that they also try to find compromises across subsystem boundaries.
As highlighted by Leifeld et al. (2021), cross-sectoral advocacy and engagement provide
actors with an informational advantage to position themselves strategically in different debates.
A key prerequisite for engaging in cross-sectoral advocacy is that policy brokers have access
to multiple policy subsystems at the same time. We can expect that ‘common’ subsystem
membership of policy brokers becomes more likely when subsystems share similar
professional norms. For instance, there are several policy subsystems in which economic
experts play a pronounced role. This might, for instance, explain the strong connections that
we can observe in our data between the field of macroeconomics on the one hand, and
domestic commerce, environment, labor, social welfare, and transport on the other.
Second, we can expect that cross-sectoral policy brokerage is facilitated by common
institutional or organizational affiliations. This is the case, for instance, when competencies for
different policy subsystems are bundled in one ministry or agency. In Germany, for instance,
competencies of social issues and labor are traditionally organized under the same ‘roof’. It
was only in the 2002 to 2005 period when labor was part of the ministry for the ‘Federal Ministry
for Economic Affairs and Labour’ while social issues were located in the same ministry as
health issues (‘Federal Minister for Health and Social Security.’). Moreover, in the German
context, energy – especially the issue of nuclear safety – has always been organized in the
same ministry (‘Federal Ministry for the Environment, Nature Conservation, Nuclear Safe’).
These strong institutional links are also reflected in our ‘best buddy’ pairs observed above.
5.2 Policy proximity from the MSA perspective
Contrary to the ACF, which at least provides a basis for developing arguments about
interconnections between policy subsystems, the MSA framework offers less leverage for
developing such expectations. The reason for this lies in the fact that – contrary to all other
theories of policy change – the main assumption of the MSA is that policy change is contingent
and the perception and interpretation of political decision-making are generally ambiguous
(Herweg, 2015). The emphasis placed on contingency and ambiguity at the same time implies
that from an MSA perspective any potential interconnections between policy subsystems are
similarly random and might be perceived differently by different actors involved.
According to this basic conception, the main conclusion is that, by definition, there
should be no systematic patterns in which policy subsystems are linked. Policy proximity is
subject to random variation. This implies that there should be no “best buddies” of policy
subsystems, nor should we find any systematic patterns in the degree to which subsystems
vary in their connection to other subsystems. Although Kingdon (2003) refers to spill-over
effects, he essentially refers to such effects within rather than between policy subsystems. For
him, such effects describe the phenomenon that a policy change that has already taken place
can have an impact on the policy status of issues that are interpreted as problematic on the
basis of the same assessment criterion (Herweg, 2015: 334).
At the same time, however, the MSA implicitly emphasizes that the opening of the
famous policy window is facilitated or even presupposes the cross-sectoral framing of policy
issues. On the one hand, policy entrepreneurs might engage in coupling problem and policy
streams by selectively emphasising aspects of an ambiguously perceived problem (Kingdon,
2003; Zahariadis, 2008). On the other hand, policy entrepreneurs may seek to couple the
politics and policy streams, attempting to exploit changing political opportunity structures by
advocating their favoured policy as a solution for a given problem. Both of these strategies –
problem-focused advocacy and problem-surfing (Boscarino, 2009) – might entail the cross-
sectoral framing of policy issues. In this regard, frame bundling and cross-subsystem
involvement has been characterized as an essential element of policy entrepreneurship (Faling
et al., 2019). In sum, from an MSA perspective, we can derive two main conclusions about
subsystem interconnections. First, there should be no systematic pattern of interconnections.
Second, at the same time, such interconnection should be observed very frequently and might
even be the rule rather than the exception.
Our data only partially support these considerations. While we cannot exclude the
possibility that the observed connections are sometimes ‘random’, the simple fact that we see
strong variation across the policy subsystems suggests that there are systematic differences
between the areas and that these differences go beyond mere chance. A possible explanation
for this (unexpected) finding could be the existence of so-called “policy communities” (for
broader discussion see Herweg 2016). According to Kingdon (2002) “[p]olicy communities are
composed of specialists in a given policy area (…) scattered both through and outside of
government”. In other words: although policy processes might be highly contingent, the policy
solutions, alternatives and thus interconnections considered might still be ‘structured’ around
rather established and constant ties both within and across different issue networks.
5.3 Policy proximity from the PET perspective
Departing from the PET perspective, the main source of potential cross-sectoral
interconnections is the subsystem competition for macrolevel attention. The theoretical
argument of PET rests on the idea that while the segmentation of policies into policy
subsystems promotes routine decision making and incrementalism, major policy changes
(punctuations) require political attention at the level of the entire political system. Yet since the
attention space at the system level and the cognitive capacities of policymakers to process
incoming information are limited (Jones & Baumgartner, 2005), trade-offs in attention are an
inevitable consequence (May et al., 2011; Fernandéz-í-Marín et al., 2020). In view of these
considerations, trade-offs in political attention at the system level should be most pronounced
when an external shock hits the political system. Such external shocks can be of a rather short-
term nature and come in the form of a single focusing event (Kingdon, 2003), or they can exert
their impact over a longer period, like in the case of economic crises.
Consequently, external shocks will affect the relative position of subsystem demands
in the competition for political attention at the system level. When an external shock hits a
political system, the system should react by channeling its resources into the fight against the
causes of the shock or at least into the alleviation of its most severe consequences. This entails
that political systems experience pressure to redirect their focus of attention toward those
policy subsystems that are closely associated with and proximate to the shock at the expense
of other policy subsystems that are regarded as more remote. Policy proximity across
subsystems hence emerges from the fact that different policy subsystems share the
characteristic of being affected or relevant for the design of policy responses to the external
shock experience. These subsystems display a higher probability of receiving attention at the
system level. While political systems hardly ever devote the same amount of attention to all
policy subsystems even without the presence of an external shock, these external shocks
should incentivize a pronounced redistribution of political attention at the system level.
Yet, while PET scholars only very recently have begun to systematically analyze such
cross-system interdependencies (Knill & Steinebach, 2022), PET is rather silent, when it
comes to the question of why some subsystems might gain in the advent of external shocks
while others lose. We argue that one potential explanation is that different areas are
functionally interlinked, i.e. they all must be addressed simultaneously to deal with a common
underlying problem. In this context, Fernandéz-í-Marin et al. (2020), for instance, highlight that
“that political systems experience pressure to redirect their focus of attention toward those
policy subsystems” and that the policy subsystems are “relevant for the design of policy
responses to the external shock experience a higher probability of receiving attention at the
system level” (p. 4-5). This insight might explain why – during our investigation and given the
impact of the COVID-19 health pandemic – health matters have coupled with otherwise rather
‘remote’ buddies such as “Education”, “Civil Rights”, and “International Affairs”.
5.4 Factors affecting the proximity between policy subsystems
The above discussion has shown that central theories of policy change so far lack analytical
sensors to capture interdependencies between policy subsystems. Yet, we have seen that
despite this analytical neglect, some expectations on the probability of such connections can
be derived by building on central elements emphasized in the different theoretical frameworks.
These theoretical insights, in turn, can be used to explain some (but definitely not all) of our
More specifically, we can expect that subsystem proximity generally increases with four
factors: (1) the presence of actors adhering to similar professional norms; (2) the presence of
actors with similar institutional or organizational affiliations; (3) the resonance of policy
problems with similar political cleavages; and (4) the functional need to solve an underlying
problem in different policy subsystems simultaneously. In short, subsystem proximity – at least
from the interpretation of existing theories of policy change – is a matter of factors rooted in
actor characteristics, institutions, politics, as well as policies.
6. Conclusion: Policy Proximity and the Study of Policy Change
The central aim of this article was to propose a novel measurement concept of policy proximity
that captures the ‘distance’ between different policy subsystems based on their joint
appearance in media reporting. By analysing 27.000 media reports aired by the German news
broadcast ‘Tagesschau’, we show that (1) policy subsystems differ in their connectivity to other
subsystems and (2) identify for each subsystem those subsystems to which it is most closely
connected. In addition, we reviewed the dominant theories on policy changes to identify
aspects that help to explain our empirical findings and to further theorize the level of
connectivity between different policy subsystems.
Despite these insights gained, however, this paper is only a first attempt in assessing
and discussing policy proximity empirically. Future analyses might check whether our findings
also hold in our contexts and media outlets such as newspaper articles. Moreover, it is worth
analysing whether the patterns observed vary across countries and over time based on,
amongst others, the (re)allocation of responsibilities within and across state departments
(Bertel & Schulze-Gabrechten, 2021) and the emergence of new societal cleavages (Ford &
Jennings, 2020). Moreover, it must be clear that studying policy proximity through media
reporting is a valid strategy to avoid endogeneity problems. However, the final analytical
interest must be on explaining and predicting actual policy changes across different policy
subsystems based on proximity measures. When do we observe major policy changes in
different policy subsystems at the same time? In this context, a promising starting point might
be to formulate expectations based on the insights gained in this paper and then to test whether
the presumed interconnections also hold and can be found in data on policy outputs (for the
German context see Breunig & Schnatterer, 2019).
Armborst, Andreas (2017). Thematic Proximity in Content Analysis. SAGE Open.
Bandelow, Nils C. (2015). Advocacy Coalition Framework. In Georg Wenzelburger & Reimut
Zohlnhöfer (Eds). Handbuch Policy-Forschung (pp. 305–324). Wiesbaden: Springer VS.
Baumgartner, Frank R. & Jones, Bryan D. (1993). Agendas and Instability in American Politics.
Chicago: University of Chicago Press.
Baumgartner, Frank. R., Breunig, Christian, Green-Pedersen, Christofferm Jones, Bryan D.,
Mortensen, Peter B., Nuytemans, Michiel & Walgrave, Stefaan (2009). Punctuated
Equilibrium in Comparative Perspective. American Journal of Political Science, 53, 603-
Bertels, Jana & Schulze‐Gabrechten, Lena (2021). Mapping the black box of intraministerial
organization: An analytical approach to explore structural diversity below the portfolio
level. Governance, 34(1), 171-189.
Bevan, Shaun (2019). Gone Fishing: The Creation of the Comparative Agendas Project Master
Codebook. In Frank Baumgartner, Christian Breunig, C. & Emiliano Grossman (eds.)
Comparative Policy Agendas: Theory, Tools, Data (pp. 305–324). Oxford: Oxford
Bos, Wilfried & Tarnai, Christian (1999). Content analysis in empirical social
research. International Journal of Educational Research, 31(8), 659-671.
Boscarino, Jessica E. (2009). Surfing for problems: Advocacy group strategy in US forestry
policy. Policy Studies Journal, 37(3), 415-434.
Breunig, Christian & Schnatterer, Tine (2020). Die politische Agenda Deutschlands. Politische
Vierteljahresschrift, 61, 131–149.
Burstein, Paul (1991). Policy Domains: Organization, Culture, and Policy Outcomes. Annual
Review of Sociology, 17: 327–350.
Capano, Giliberto (2009). Understanding Policy Change as an Epistemological and
Theoretical Problem. Journal of Comparative Policy Analysis: Research and
Practice, 11(1): 7-31
Cairney, Paul & Weible, Christopher M. (2015). Comparing and Contrasting Peter Hall’s
Paradigms and Ideas with the Advocacy Coalition Framework. In John Hogan & Michael
Howlett (Eds.). Policy Paradigms in Theory and Practice. Studies in the Political
Economy of Public Policy (pp. 83—99). London: Palgrave Macmillan.
Chui, Cheryl, Feng, Joyce Y. & Jordan, Lucy (2014). From good practice to policy formation—
The impact of third sector on disaster management in Taiwan. International Journal of
Disaster Risk Reduction, 10, 28-37.
Döring, Herbert (1995). Time as a scarce resource: government control of the
agenda. Parliaments and majority rule in Western Europe. In Herbert Döring (Eds).
Parliaments and Majority Rule in Western Europe (pp. 223-246). Frankfurt: Campus
Faling, Marijn, Biesbroek, Robert, Karlsson-Vinkhuyzen, Sylvia & Termeer, Katrien (2019).
Policy entrepreneurship across boundaries: A systematic literature review. Journal of
Public Policy, 39(2), 393-422.
Ford, Robert & Jennings, Will (2020). The changing cleavage politics of Western Europe.
Annual Review of Political Science, 23(1), 295—314.
Fernández‐i‐Marín, Xavier, Hurka, Steffen, Knill, Christoph & Steinebach, Yves (2019).
Systemic dynamics of policy change: Overcoming some blind spots of punctuated
equilibrium theory. Policy Studies Journal.
Goyal, Nihit & Howlett, Michael (2021) ’Measuring the Mix’ of Policy Responses to COVID-
19Comparative Policy Analysis Using Topic Modelling. Journal of Comparative Policy
Analysis: Research and Practice, 23(2), 250-261.
Herweg, Nicole (2015). Multiple Streams Ansatz. In Georg Wenzelburger & Reimut Zohlnhöfer
(Eds). Handbuch Policy-Forschung (pp. 325–355). Wiesbaden: Springer VS.
Herweg, Nicole (2016). Clarifying the Concept of Policy-Communities in the Multiple-Streams
Framework. In Reimut Zohlnhfer & Friedbert W. Rb (Eds). Decision-Making under
Ambiguity and Time Constraints Assessing the Multiple-Streams Framework (pp. 125–
146). Colchester: ECPR Press.
Hurka, Steffen (2017). Rampage shootings and gun control: Politicization and policy change
in Western Europe. London: Routledge.
Ingold, Karin & Varone, Frédéric (2012). Treating Policy Brokers Seriously: Evidence from the
Climate Policy. Journal of Public Administration Research and Theory, 22(2), 319–346.
Jochim, Ashley E., & May, Peter J. (2010). Beyond subsystems: Policy regimes and
governance. Policy Studies Journal, 38(2), 303-327.
Jones, Bryan D., & Baumgartner, Frank R. (2005). The politics of attention: How government
prioritizes problems. Chicago: University of Chicago Press.
Knill, Christoph & Steinebach, Yves (2022). What has happened and What has not happened
due to the COVID-19 Pandemic: A Systemic Perspective on Policy Change. Policy &
Society, 41(1), 25-39.
Kingdon, J. W. (2003). Agendas, Alternatives, and Public Policies. New York: Longman.
Krger, Udo M., & Zapf-Schramm, Thomas (2017). InfoMonitor 2016: Nachrichtenprofile
langfristig stabil. Analyse der Fernsehnachrichten von Das Erste, ZDF, RTL und Sat.1,
Media Perspektiven 2, 62–90.
Leifeld, Philip, Henrichsen, Tim, Buckton, Christina, Fergie, Gilian, & Hilton, Shona (2021).
Belief system alignment and cross-sectoral advocacy efforts in policy debates. Journal
of European Public Policy, 1-24.
May, Peter J., Jochim, Ashely E., & Pump, Barry (2013). Political Limits to the Processing of
Policy Problems. Politics and Governance, 1(2): 104-116.
May, Peter J., Jochim, Ashley E. & Sapotichne, Joshua (2011). Constructing Homeland
Security: An Anemic Policy Regime. Policy Studies Journal 39: 285–307.
McGee, Zachary A., & Jones, Bryan D. (2019). Reconceptualizing the policy subsystem:
Integration with complexity theory and social network analysis. Policy Studies Journal,
Nohrstedt, Daniel & Weible, Christopher (2010). The Logic of Policy Change After Crisis:
Proximity and Subsystem Interaction. Risk, Hazards & Crisis in Public Policy,1(2), 1-32.
Proksch, Sven‐O. & Slapin, Jonathan B. (2012). Institutional foundations of legislative
speech. American Journal of Political Science. 56(3), 520-537.
Sabatier, Paul A. (1993). Advocacy-Koalitionen, Policy-Wandel und Policy-Lernen: Eine
Alternative zur Phasenheuristik. In Paul A. Sabatier (Eds). Policy-Analyse (pp. 116-148).
Wiesbaden: VS Verlag für Sozialwissenschaften.
Sabatier, Paul A. & Weible, Christopher (2007). The Advocacy Coalition Framework.
Innovations and Clarifications. In: Paul A. Sabatier (Eds). Theories of the Policy
Process (pp. 189–220). Boulder, Co: Westview Press
Sichling, Florian (2020). ‘Mutatio Sub Pressura’: An Exploration of the Youth Policy Response
to the Syrian Refugee Crisis in Germany. Journal of Social Policy, 1-19.
Zahariadis, Nikolaos (2008). Ambiguity and choice in European public policy. Journal of
European Public Policy, 15(4), 514-530.