Journal of Public Administration Research And Theory, 2018, 239–253

doi:10.1093/jopart/mux045
Article
Advance Access publication January 24, 2018

Article

The Bureaucracy and the Policy Agenda

Martin Baekgaard,* Peter B. Mortensen,* Henrik Bech Seeberg*
*Aarhus University

Downloaded from https://academic.oup.com/jpart/article/28/2/239/4823058 by guest on 02 May 2024

Address correspondence to the author at [email protected]

Abstract
The public administration literature has been dominated by questions about how politicians can
control the bureaucracy’s application and implementation of laws at the back end of the policy pro-
cess. Much less scholarly attention is devoted to the influence of the bureaucracy on the content
and composition of the policy agenda at the front end of the process. Agenda setting is a funda-
mental aspect of politics, and this article examines the influence of the bureaucracy on the policy
agenda and the conditions for this influence. The core proposition is that the policy agenda is
larger and more diverse in political systems in which administrative professionals take up a larger
share of the bureaucracy. This effect is expected to be mitigated by the involvement of elected
representatives in the policymaking process. The empirical analysis supports these expectations.
The findings are based on a time-series cross-section dataset from 98 Danish municipalities over
7 years containing a detailed coding of local council agendas and rich register data.

Since the US community studies of the 1950s and 1960s and Jones 2014; Green-Pedersen and Walgrave 2014;
(e.g., Dahl 1956; 1961; Hunter 1953; Mills 1956), the Zahariadis 2016).
importance of agenda setting (i.e., the ongoing selec- Traditionally, the literature on policy agenda set-
tion of a limited number of policy problems1 for poli- ting has explained the policy agenda using variables
ticians to handle) has been widely acknowledged in relating to party politics (Novotny et al. 2016), interest
political science. Schattschneider is often credited as groups (Bonafont 2016), public opinion, and/or prob-
the founding father of the field with his observation lem indicators, such as economic conditions, unem-
that “[s]ome issues are organized into politics while ployment, or crime rates (e.g., Baumgartner et al. 2011;
others are organized out” (Schattschneider 1960, 69). Mortensen and Seeberg 2016). Despite two recent
This identification of a “conflict of conflicts” was path- exceptions (Baumgartner and Jones 2015; Workman
breaking, as it emphasized how political conflict is not 2015), however, the bureaucracy has been left out of
only about decisions on issues that are already on the the equation when it comes to the question about what
political agenda but also about which issues make it or who influences the policy agenda.
onto the political agenda in the first place (for recent This ignorance is surprising given the longstanding
reviews, see Baumgartner et al. 2017; Eissler, Russell, scholarly interest in the relationship between adminis-
tration and politics (e.g., Jacobsen 2006; Peters 2010;
Earlier versions of this article have been presented at The 9th Annual
Svara 2001; Weber 1922; Wilson 1887). It is further
Conference of the Comparative Agendas Project in Geneva, and at surprising given the importance and resources of
the Public Management Research Conference in Aarhus. Thanks to bureaucracies in modern political systems. As argued
the other participants for valuable comments. Also thanks to Jørgen by Meier (2007), any theory of politics must also be
Grønnegaard Christensen, Carsten Jensen, Matt W. Loftis, and the a theory of public administration. The bureaucracy
three anonymous reviewers for constructive and helpful comments
and suggestions. Finally, thanks to Kurt Houlberg and FLS for generous
does not merely implement policies at the back end
assistance with data. of the policy process. With their expertise and special-
1
We use “policy problems” and “issues” interchangeably. ized knowledge on various issue domains, bureaucrats

© The Author(s) 2018. Published by Oxford University Press on behalf of the Public Management Research Association. 239
All rights reserved. For permissions, please e-mail: [email protected].
240 Journal of Public Administration Research and Theory, 2018, Vol. 28, No. 2

filter information and generate alternatives to political Examining these questions empirically requires rich
decision-makers at the front end of the policy process and valid data on the policy agenda, the composition
(Baumgartner and Jones 2015; Workman 2015). In of the public bureaucracy, and the involvement of poli-
other words, the bureaucracy is in a core position in ticians. Being able to distinguish the effect of bureau-
the process of transforming conditions into political cracy also requires a research setting in which we can
problems and solutions that reach the policy agenda. control for the potentially large range of other policy
This point is just as important for research in public agenda determinants. To meet these requirements, we
administration as to agenda setting research. Even if base our empirical analysis on time series data (2007–
the relationship between the political level and bureau- 2013) from all 98 Danish municipalities. As detailed
cracy has featured prominently in the public adminis- below, the Danish municipalities are potent political
tration literature, the debate has primarily centered on units responsible for providing about half of all pub-
the back end of the policy process: How do popularly lic service deliveries and accounting for around half of
elected politicians get a bureaucracy to turn laws into all public spending in Denmark. Danish municipalities

Downloaded from https://academic.oup.com/jpart/article/28/2/239/4823058 by guest on 02 May 2024

rules in line with the politicians’ intentions (e.g., Acs employ some 500,000 persons (more than 60% of all
2015; West and Raso 2013)? Although this is a natural public employees). For each municipality, we conduct
and still highly relevant point of departure to under- a systematic content coding of the local council meet-
stand the influence of bureaucracy on the output of the ing agendas, which we analyze against data on central
political system, it neglects the role of the bureaucracy characteristics of the local bureaucracies. By studying
in the earlier stage of the policy process, where pol- the large number of Danish municipalities, we can
icy problems are prioritized for action at the political statistically control for a range of potentially relevant
level. In fairness, scholars of bureaucratic politics have variables, including population size, the economic con-
emphasized how bureaucrats can advance policy pro- ditions of the municipalities, and measures of the party
posals and ideas within their areas of expertise (e.g., political composition of the local councils. The article
Carpenter 2001; Krause 1999), but they have not been concludes with a discussion of the generalizability of
focusing on how characteristics of the bureaucracy the empirical findings.
may influence the size and issue composition of the
broader policy agenda as such. Since studies in policy The Bureaucracy and the Policy Agenda
agenda setting document how critical this stage is for Much policy agenda research has focused on describ-
the final output of the political system, this neglect is ing how policy agendas develop over time with a spe-
undesirable if we are to better appreciate the signifi- cial focus on the stability and change of policy agendas
cance of the bureaucracy in the political system. (e.g., Jennings, Bevan, and John. 2011; Jennings and
To stimulate more systematic research on the John 2011). Some studies have focused on explaining
agenda-setting effect of public bureaucracies, this arti- the changes in policy agendas investigating, for instance,
cle examines how the bureaucracy matters to the pol- how the size and composition of the policy agenda is
icy agenda and under what conditions it is most likely shaped by party politics (Bevan, John, and Jennings.
to have an effect. Following the Weberian tradition, we 2011; Mortensen et al. 2011), economic conditions
focus on administrative professionals who constitute (Jones and Baumgartner 2005, chapter 8), changes in
a special part of the bureaucracy and typically have a real-world problems (Baumgartner and Jones 2002),
university degree in administrative sciences, econom- or changes in the media and public agenda (Jennings
ics, or law (Bhatti, Olsen, and Pedersen 2009; Bhatti, and John 2009; Soroka 2002). Meanwhile, scholars
Olsen, and Pedersen 2011). In particular, we argue have recently started studying how bureaucracy influ-
that the share of administrative professionals influ- ences what is coined “the rulemaking agenda” (Acs
ences the size and issue diversity of the policy agenda 2015; West and Raso 2013). Conceptually, however,
in the sense that when there are more administrative this agenda is different from the policy agenda as it
professionals in the administration, more issues and describes the decisions by bureaucrats on which legis-
broader sets of issues are elevated onto the political lative statutes to put into rules and when as opposed to
agenda. Furthermore, applying a contingency perspec- which legislative statutes to consider in the first place
tive (Jacobsen 2006) to the relationship between the (Boushey and McGrath 2017).
bureaucracy and the policy agenda, we argue that the To understand the influence of bureaucracy on the
policy agenda effect of the bureaucracy is contingent policy agenda, the starting point of this article is that
upon the politicians’ involvement in the policymaking policymaking requires information and that one of
process. Although there might be other relevant con- the most important functions of the bureaucracy is to
tingencies, given the focus of this article, we begin by structure information about problems and alternatives
exploring the direct involvement of politicians in the and to channel it into the system. In a policy world
policymaking process. in which bounded rationality (Simon 1947) prevents
Journal of Public Administration Research and Theory, 2018, Vol. 28, No. 2 241

politicians from attending to inputs proportionately Whereas health or education professionals in the or-
(Jones and Baumgartner 2005), the processing of ganization can be expected to try to gain professional
information from the bureaucracy to the elected politi- autonomy from political interference in their area, ad-
cians is critical to the policy agenda. Information and ministrative professionals may be more politically ori-
attention are closely related, and information makes ented (Smith and Christensen 2016). These skills allow
politicians aware of new problems or facets of prob- administrative professionals—and their professional
lems possibly requiring attention (Baumgartner and interest encourages them—to identify, connect, and
Jones 2015). Workman (2015) places similar emphasis process a wide range of potentially sensitive political
on the relevance of the bureaucracy for policy agenda problems. In contrast to bureaucrats with a more sec-
setting. According to Workman (2015), in addition to tor-specific educational background, they may also be
policy implementation, decision-makers also rely on able to communicate the identification of such prob-
the bureaucracy for policy feedback, problem detec- lems to politicians more credibly.2
tion, and the design of new policies. Workman (2015) Providing relevant input to the policy agenda puts

Downloaded from https://academic.oup.com/jpart/article/28/2/239/4823058 by guest on 02 May 2024

thereby advances some classic perspectives on bureau- great demands on the bureaucracy. Analytical exper-
cracy and public administration (Lindblom 1959; tise is necessary in order to cut through large amounts
March and Olsen 1989; Simon 1947; Wildavsky 1964; of information, eliminate noise, and channel issues to
Wilson 1989). the political level. Moreover, input to the policy agenda
Despite the significance of the recent contribu- should be legally sound, economically possible, and
tions by Workman (2015) and Baumgartner and Jones politically feasible (Workman 2015, 42). The compe-
(2015), they also leave several important questions tencies to meet these demands vary among the differ-
open about the agenda-setting influence of the bur- ent professions employed in the public administration
eaucracy. First and foremost, it is important to begin (Egeberg and Trondal 2009). Bureaucrats who are
exploring whether and under what conditions bureau- highly skilled in political, legal, and economic matters,
cracy matters to the policy agenda. Every bureaucracy such as administrative professionals, can be expected
of a certain size consists of different types of employees to be better equipped to lift and fit issues to the polit-
with different skills and responsibilities (e.g., Bhatti, ical level than those with more limited training in these
Olsen, and Pedersen 2009; 2011; Wilson 1989). The areas. For instance, even if chemists in the sewage sec-
next section focuses on a group of bureaucrats, the ad- tion of a body of public administration might be the
ministrative professionals, who are likely to play a spe- most capable of detecting a growing risk of water pol-
cial role for the policy agendas. lution, administrative professionals are best equipped
to pick up on this potential problem and process the
Administrative Professionals and the Policy Agenda information in a manner that elevates it to the political
Bureaucracies typically consist of agents with differ- level.3
ent educational backgrounds and responsibilities. We Another contrast between administrative profes-
argue that administrative professionals constitute a sionals and other public employees (e.g., engineers,
part of the bureaucracy that is particularly likely to school principals) is that the latter groups are often
affect the number of policy issues up for active con- concerned with implementation and service deliv-
sideration on the agendas of the elected politicians. ery, whereas administrative professionals are skilled
“Administrative professionals” refers to administra- in planning and evaluation and often more focused
tive personnel with a professional, administrative on legislative proposals, future problems, and pol-
background for whom managing and developing the icy change and development. For instance, research
public sector and advising politicians is a main task addressing local government innovation has shown
(e.g., Bhatti, Olsen, and Pedersen 2009). Typically, they that the increased presence of administrative profes-
have a university degree in administrative sciences, sionals in local government administration enhances
economics, or law. Still, they might not be considered a
profession in a strict sense, as defined by, for instance, 2
Hird (2005) shows how professionals in US nonpartisan, publicly funded
Roberts and Dietrich (1999), because they might not
policy research organizations working on the borders of the US public
necessarily share norms about appropriate behavior. administration serve a similar role as information processors. Hence,
Due to their respective educational backgrounds, how- he also identifies the need to nuance the view of bureaucrats and their
ever, they do possess strong analytical skills and theor- role in the policy process.
etical knowledge distinguishing them from bureaucrats
3
An alternative mechanism is that the chemist and the administrative
professional forward equal amounts of issues for the political agenda,
with other educational backgrounds (Bhatti, Olsen,
but only the latter’s proposal is picked up by the politicians due to
and Pedersen. 2009; 2011). reputation. Based on previous literature (Bhatti, Olsen, and Pedersen
Moreover, administrative professionals are educated 2011; Hird 2005), the mechanism we put forward appears more plausible
and hired to be generalists rather than area specialists. than this alternative, even if both are at play.
242 Journal of Public Administration Research and Theory, 2018, Vol. 28, No. 2

the likelihood of adopting new, untried public services professionals. We focus on two aspects of the involve-
(Bhatti, Olsen, and Pedersen 2011). This is argued to ment of politicians in the policymaking process.
be due to the importance of administrative profession- First, the attention and involvement of politicians
als to innovation because of their analytical skills and possibly depend on the organization of the policy-
theoretical knowledge. They are entrepreneurs, able to making process. The committee structure of the pol-
take risks and to work outside of the standard oper- itical system has been studied extensively in research
ating procedures and routines within an organization on the US Congress (Groseclose 1994; Shepsle and
(Bhatti, Olsen, and Pedersen 2011, 581–2). Hence, they Weingast 1985; Sprague 2008) and political systems
are qualitatively different from public sector employ- from other countries, including Danish municipali-
ees who are specialized in casework and in implement- ties (Bækgaard 2010; 2011). Important agenda setting
ing political decisions. studies indicate that the committee structure shapes
All else equal, then, what difference would it make political attention (Sheingate 2006; Talbert, Jones, and
to the policy agenda to staff a bureaucracy with a rela- Baumgartner 1995). Within its jurisdiction, a commit-

Downloaded from https://academic.oup.com/jpart/article/28/2/239/4823058 by guest on 02 May 2024

tively larger share of administrative professionals? tee may initiate hearings and investigations to follow
When combining the above characterization of admin- an issue area closely. Thus, like the bureaucracy, com-
istrative professionals with the work of Baumgartner mittees consisting of elected representatives also pro-
and Jones (2015) and Workman (2015), who have cess information and generate input to the political
shown that a larger bureaucracy means a larger policy agenda.
agenda, we argue that this effect of the bureaucracy is The effect of committees, however, may depend on
especially strong in more professionalized bureaucra- their organization (Sheingate 2006; Talbert, Jones, and
cies. In other words, we expect the size and diversity of Baumgartner 1995). We argue that the number of com-
the policy agenda to increase with more administrative mittees with elected officials in general will dampen the
professionals. Here, the sheer number of administra- policy agenda effect of administrative professionals. If
tive professionals is not what matters—perhaps except a large, specialized committee structure is in place, the
in very small bureaucracies—but rather the relative political level will already be familiar with many dif-
share. The question is not if there are 50 or 60 admin- ferent issues and closely involved in building the policy
istrative professionals in the organization but if the hir- agenda. Thus, the information processing of the bur-
ing of administrative professionals replaces other types eaucracy will make less of a difference because elected
of bureaucrats. The latter may indicate that the bur- officials will be both broadly and deeply involved in
eaucracy prioritizes the processing of information to the policymaking process.4 Conversely, if there are few
the political level, which increases the likelihood that committees, the direct and broad involvement of the
the bureaucracy influences the general policy agenda. political level may be limited and the bureaucracy can
This leads to our first hypothesis: have a greater impact on the policy agenda. In other
words, we expect that a professionalized bureaucracy
Hypothesis 1: The larger the share of administrative may be an important substitute to a well-developed
professionals in the bureaucracy, the and highly specialized committee system, assuring that
greater the number of issues on the otherwise neglected issues actually enter the decision
policy agenda and the greater the issue agenda of the political assembly.
diversity of the policy agenda. The second aspect of political involvement that we
In a recent study of bureaucratic influence in US state leg- examine is the professionalization of elected politi-
islatures, Boushey and McGrath (2017) forcefully argue cians. In political systems with higher compensation
and show that the bureaucratic influence on administra- for elected representatives, politicians can often spend
tive rulemaking is conditioned on the relative degrees of more time on politics and possess more expertize on
professionalization of the state legislative and executive policy matters. The policy agenda effect of admin-
branches. This argument is consistent with the founda- istrative professionals might therefore decrease,
tional agenda-setting assumption of policymakers hav- because the politicians will be less dependent on the
ing limited capacities and attention spans (Jones 1994; skills of the administrative professionals. This is con-
2001). However, although the cognitive limitations of sistent with the aforementioned US literature on state
policymakers are assumed to be invariant in this body regulation, which shows that “… the eroding policy
of work, Boushey and McGrath (2017) point out that expertise of state legislators has resulted in increased
there may be variation in the time and attention politi- bureaucratic participation in the policy process,
cians devote to politics. Here, we extend their point and
argue that the involvement of politicians not only mod- 4
In our analysis, we control for the size of the council to take into account
erates bureaucratic influence on administrative rulemak- that many committees in a small council may not allow for this
ing but also the policy agenda influence of administrative specialization and capacity at the political level.
Journal of Public Administration Research and Theory, 2018, Vol. 28, No. 2 243

as amateur politicians rely more heavily on profes- Table 1. Key Facts About Danish Local Governments
sionalized executive agencies to define problems and
Key Facts Meana
develop solutions” (Boushey and McGrath 2017,
85). Our reasoning is that such an effect of political Council members 25
involvement on regulation must go through an effect Parties in council 6b
on the policy agenda. As discussed below, we utilize a Turnout 71.9%
measure of councilor remuneration as the indicator of Council members average working 18.3
how professionalized the elected politicians are. This hours/week
Pct. of revenue from local taxes 75%
corresponds directly to the measure used by Boushey
Share of public employees at local 65% (500,000
and McGrath (2017) in their study of US states. level employees)
Theoretically, both the committee structure and the Share of GDP spent at local level 20%
councilor remuneration are seen as two different indi-
cators of the degree of politicians’ involvement in the Note: aAverage across municipalities at/around the 2013 election.

Downloaded from https://academic.oup.com/jpart/article/28/2/239/4823058 by guest on 02 May 2024

policymaking process: b
Apart from the parties also represented in the national parlia-
ment, representatives from “local party lists” are typically elected
Hypothesis 2: 
The policy agenda influence of admin- together with independents.
istrative professionals decreases as the
number of standing committees increases. systems. Danish municipalities have both breadth and
depth in their responsibilities to deliver public services
Hypothesis 3:The higher the remuneration of the poli- on a wide number of key policy areas, including public
ticians, the less policy agenda influence schools, job training, daycare, eldercare, road and park
of administrative professionals. maintenance, disaster relief, and environmental control.
Thus, the municipalities can be characterized as multi-
purpose political units. Apart from block grants from
DATA AND RESEARCH DESIGN the central government, a main source of local govern-
The Danish Municipalities
ment revenue comes from income taxes set by the local
councils and which range between 22.5% and 27.8%
Most of the advances in research on agenda setting
across municipalities (2017 data). In competitive elec-
builds on studies of one national, political system over
tions with high voter turnout, the local representatives
time. The work by Baumgartner and Jones (2015) and
are elected every 4 years (Bækgaard and Jensen 2011)
Workman (2015) offers some of the most recent con-
and form durable coalitions in the council through their
tributions. A key limitation of this analytical approach
membership of competitive political parties (Bækgaard
is the difficulty in estimating the effect of slow-moving
et al. 2014; Serritzlew, Blom-Hansen, and Skjæveland
characteristics of the political system, which might pos-
2010). A committee form of government is used in the
sibly lead to, for example, the underestimation of the
municipalities, where standing committees composed
impact of differences in the bureaucracy on the con-
of city council members oversee the daily administra-
tent and structure of the policy agenda. Cross-national
tion of their jurisdiction (Bækgaard 2011). At public
studies are also constrained by limited access to com-
council meetings, typically held once or twice monthly,
parable data on the bureaucracy as well as limited
the city council makes final decisions.
statistical control possibilities given the low number of
observations.
To overcome these challenges, we utilize the 98 Measuring the Local Policy Agenda
municipalities in Denmark to examine the hypotheses Policy agendas have been operationalized and meas-
derived above. A major advantage of this choice of ured in several ways. At the national level of policy-
research unit is the high comparability of units across making, prime ministers’ speeches, congressional
time and space, which enables the control for a range of hearings, parliamentary debates, legislative activity,
potentially relevant alternative explanations. Although and budgets have all been used as indicators of the pol-
Denmark can be characterized as a unitary parliamen- icy agenda. A distinction is normally drawn between
tary state, there is substantially more local autonomy the systemic and governmental agendas (Cobb and
than is usually found in such states (Boadway and Shah Elder 1983). The issues that are commonly perceived
2009, 5). As illustrated by the key facts of the Danish by members of the political community as meriting
municipalities summarized in table 1, they are potent public attention are usually considered part of the
political units with directly elected politicians and con- systemic agenda, whereas the governmental agenda is
siderable autonomy from the national level of govern- more narrowly defined as the set of items explicitly up
ment, which renders them well-suited for generating for the active and serious consideration of the decision
insight into the workings of political decision-making makers (Cobb and Elder 1983, 86).
244 Journal of Public Administration Research and Theory, 2018, Vol. 28, No. 2

The council agendas utilized in this study come n

very close to what could be termed a local govern- ∑ i

sum of those products: − (p(x )) ⋅ ln p(x ) , where xi
i =1
i
mental agenda. This includes the issues that are up
represents an item, p(xi) is the proportion of the total
for active and serious consideration at local council
attention the item receives, and ln(xi) is the natural
meetings. Furthermore, no nationally defined institu-
log of the proportion of attention the item receives.
tional rules set the local council agendas; each local
The entropy score increases as the spread of atten-
council puts together its own.5 From the homepages of
tion across all items becomes more equal (Boydstun,
each municipality, we have collected and counted the
Bevan, and Thomas 2014). Across the council agendas
number of points on the docket for local council meet-
of the 98 municipalities, the entropy score varies from
ings and coded the content of each agenda point. We
2.9 to 4.2 with a mean of 3.8. This slightly exceeds
have done so for all of the council meetings for each
the entropy scores reported by Jennings et al. (2011)
municipality for the years 2007 through 2013.6 For the
and Mortensen et al. (2011) for the policy agendas of
content coding, we build on the issue coding scheme
Western European countries, which range from about

Downloaded from https://academic.oup.com/jpart/article/28/2/239/4823058 by guest on 02 May 2024

of the comparative agendas project (www.compara-
1.2 to 2.7 with an average of 2.1–2.4. As a compara-
tiveagendas.info), which identifies the substance of an
ble codebook is applied in our study and the national
item on the agenda. Using this scheme, each item on
study, these entropy scores corroborate that the Danish
the council agendas has been coded into one of 189
local councils generally deal with many different issues.
different subcategories. The topic codes are found in
Supplementary Table S1. Trained student coders have
Administrative Professionals and Standing
been used to code the material in combination with
Committees
machine coding.7
To measure the topical diversity of the policy After a local government reform in 2007 merged many
agenda, we count the number of categories used every of the smaller municipalities, reducing the number of
year to code the council agendas of a municipality. On municipalities from 271 to 98,8 the number of admin-
average, 68 of the 189 available issue categories are istrative professionals in each municipality has been
used. There is substantial variation in the diversity of tallied annually along with the total number of admin-
the council agendas. In the municipalities with the least istrative personnel. In combination with the detailed
diverse agendas, we find years where only 36 issue cat- content coding of the issues on the council agendas,
egories are employed, whereas up to 110 issue catego- this large-n research design with comparable yet not
ries are used in the coding of other council agendas. identical units of analysis offers an opportunity to con-
The diversity of an agenda depends on the number duct a fine-grained investigation of the impact of the
of different issues as well as the variation in the space bureaucracy on the local policy agenda.
allocated to each issue. Two municipalities might have The administrative professionals are not politically
the same number of issues on the agenda, but while the appointed but hired based on a classic merit system.
first one might spend most of its time on only a few Ultimately, the bureaucracy serves the council and
such issues, attention is more evenly distributed in the whereas the top city manager is typically hired on a
second. In order to capture these differences, we use the short-term contract (Christensen et al. 2014), the
entropy score commonly used in agenda setting studies administrative professionals are generally more shel-
(Baumgartner and Jones 2015; Jennings et al. 2011). tered from direct political pressure through the Danish
For the entropy score, we follow the recommenda- public law of civil servants and a strong code of conduct.
tions of Boydstun, Bevan, and Thomas (2014) and use These professionals typically hold university degrees
Shannon’s H, which is calculated by multiplying the in law, economics, or public administration, hold gener-
proportion of the agenda that each item receives by the alist position in the administration, often as a manager,
natural log of that proportion, then taking the negative and are members of DJØF, the Danish professional
association for public managers, lawyers, and econo-
mists. However, it can be hard to capture the population
5
National level regulation may require that some issues—such as an
annual local school performance report—are placed on the council
of administrative professionals based on only infor-
agendas, but this is the exception and it affects all municipalities mation about their educational background, employ-
equally, meaning that it cannot account for the agenda variation ment position, or union membership. Relying only on
observed below. educational information might include civil servants
6
The comprehensive reform of local government in 2007 involving large- that despite a relevant education to be administrative
scale municipal mergers prevents an extension of the time series
further back in time. Furthermore, given the massive task of content
professionals do not work as such. Looking only at
coding these ~200,000 agenda dockets, the time series ends in 2013, union-membership might under- or overestimate the
where the data collection process was initiated.
7
See Loftis and Mortensen (2017) for a description of the coding strategy, 8
239 municipalities were merged to form 66 new municipalities, whereas
including intercoder-reliability tests. 32 municipalities remained unchanged.
Journal of Public Administration Research and Theory, 2018, Vol. 28, No. 2 245

true number of administrative professionals due to self- Particularly, we measure our main independent vari-
selection bias. Hence, to arrive at the most valid and able as the number of administrative professionals (our
reliable indicator, we combine three indicators into an index) per 100 administrative employees in the muni-
index. Our first indicator is taken from previous work cipality. Along with this measure, we control for the
(Bhatti, Olsen, and Pedersen 2009, 2011), and from total number of administrative employees in the mu-
each municipality and each year we gather from the nicipality (i.e., not service providers such as teachers
DJØF member database information about the number and daycare workers). As explained above, we use this
of municipal employees who are also DJØF members. measure to study the implications of administrative
Many of the other administrative employees, typically professionals as part of the composition of the full ad-
caseworkers and secretaries, are members of HK, the ministration. If administrative professionals make up a
“white-collar office workers’ union.” Moreover, from large part of the bureaucracy, we expect the processing
the Danish Statistical Bureau (Statistics Denmark), information to the political agenda to be one of its key
we extract information on the number of employ- tasks. Only a relative measure reveals the commitment

Downloaded from https://academic.oup.com/jpart/article/28/2/239/4823058 by guest on 02 May 2024

ees in each municipality and for each year who hold of the bureaucracy to prioritize this task. We control
university degrees at the master level or higher in law, for the total number of administrative employees to
economics, or public administration. Finally, as an indi- take into account that the ratio of administrative pro-
cator of employment position, we extract the number fessionals may change not because of a change in the
of municipal employees who are covered by a collective number of administrative professionals but because of
agreement negotiated by DJØF from the Danish Public general changes in the administration unrelated to the
Salary Office (www.fldnet.dk). commitment to hire administrative professionals. Had
With factor loadings between 0.85 and 0.91, our we instead relied on an absolute measure of admin-
three variables are highly correlated despite the fact istrative professionals, our measure would not have
that they come from independent sources (the eigen- told if an increase in their number truly reflected an
value of the first factor in the principal factor analysis increased role for administrative professionals in the
is 2.35, and the Cronbach’s alpha is 0.9; see Table A3). bureaucracy or if such increase was only part of a
Hence, we are confident that our index is a useful broader expansion of the entire administration poten-
proxy for the number of administrative professionals, tially leaving the administrative professionals less im-
and we construct the index by adding the three vari- portant in the administration.
ables and dividing by three. Comfortingly, our index To examine Hypothesis 2, we count the number of
approximates a normal distribution (Figure A1). permanent standing political committees from official
Most Danish municipalities are broadly organized municipal documents. These are committees that each
according to policy domains corresponding to the new council agree on forming after the election for the
main tasks of local governance, the most common coming term and do not include the mandatory commit-
being schools, water supply, sewage, roads, elderly tee for economic affairs. Given that a local election took
care, culture, and leisure. Aggregate data from the place at the end of 2009, for observations 2007–2009
Danish Public Salary Office (KRL 2017) suggest that and 2010–2013, respectively, the same score is used for
administrative professionals are not concentrated in each municipality. With respect to Hypothesis 3, we
any one issue domain but are quite evenly scattered use councilor remuneration collected from the Danish
among them. More importantly, the vast majority Statistical Bureau (Statistics Denmark 2017). We have
of administrative professionals (roughly 75%) are extracted the total sum of payment (real prices) for
located in the central administration at the city/town each council in each year and divided it by the number
hall, which coordinates across domains and prepares of council members. We follow Boushey and McGrath
bills to the council floor. Hence, we should not expect (2017) and take remuneration as a valid proxy for poli-
issue-specific effects in which, say, primary educa- ticians’ professionalization.9 To the extent that salary
tion makes it to the political agenda more often in is based on the hours put into the job as councilors,
a bureaucracy with a high share of administrative it reflects professionalization in the sense that it takes
professionals. Consistent with the theoretical argu- time to excel in a job. Councilors who invest more time
ment, this instead points to an influence on the size
and diversity of the local government agenda as such.
Moreover, DJØF (2017) membership data reveals The formal rules for financial remuneration are regulated by
9

that a high proportion of the administrative pro- the departmental order on remuneration for local politicians
(Vederlagsbekendtgørelsen 2016). The amount contains a fixed element
fessionals are in central management and advisory which is increasing with municipality size. The municipalities have
positions. This underlines how administrative pro- wide discretion, however, with respect to whether and to what extent
fessionals engage with the political level to develop this fixed element is supplemented by additional salaries for committee
public policies. appointments and other posts.
246 Journal of Public Administration Research and Theory, 2018, Vol. 28, No. 2

can be expected to deliver more and to rely less on number of committees but possibly also reflects the
administrative professionals to form the policy agenda. much higher share of administrative professionals in
Moreover, a higher salary might attract more dedicated Albertslund compared to Herlev. A similar comparison
and competent councilors to public service. Hence, sal- could be made for a set of municipalities that are al-
ary is a central indicator of the involvement and profes- most opposite on the background variables compared
sionalization of the local councilors. to Albertslund and Herlev but with a similar agenda
Detailed summary statistics of the main variables in diversity. Both located in rural Denmark, far from
the analysis are reported in the appendix ( Table A4). the capital and governed by liberal-conservative may-
Important for our analysis, the statistics reveal sub- ors and with low socio-economic pressure, one might
stantial variation, both over time and across units. expect Kerteminde and Rebild to have rather similar
The number of permanent political committees var- council agendas, but the composition of the respective
ies considerably (from 2 to more than 10). Moreover, council agendas proves highly unequal. Rebild, with
the standard deviation of the number of administra- the most diverse council agenda, has almost twice the

Downloaded from https://academic.oup.com/jpart/article/28/2/239/4823058 by guest on 02 May 2024

tive professionals per 100 administrative employees number of administrative professionals per 100 ad-
reflects how the ratio of administrative professionals ministrative employees. These are the sorts of puzzles
in the municipalities has generally grown each year. that will be subjected to multivariate statistical testing
The same is true for our diversity measure: more issues in the analysis reported below.
have come to the policy agenda in our period of analy-
sis. This might obviously suggest that if our empirical Controls
analysis shows a relationship between administrative One advantage of studying the Danish municipali-
professionals and the policy agenda, it may simply be ties is that we have access to a number of potentially
an artifact of the reorganization in 2007. To control important control variables, including the number of
for this we add a dummy variable to the analysis sepa- residents, the degree of urbanization, and the socio-
rating merged from non-merged municipalities. economic pressures measured using an index com-
The concentration of administrative profession- piled by the Danish Ministry of the Interior and Social
als is partly—but far from fully—correlated with the Policy, which includes indicators of the percentage of
municipal population size, the size of the public admin- residents without employment, with limited education,
istration at large, and the number of standing commit- and with low income. Moreover, it may be the strength
tees (correlation coefficients are shown in Table A2). of the political level vis-à-vis the bureaucracy—where
Importantly, we are therefore able to disentangle the the political level is unable to resist the pressure from
impact of administrative professionals from that of the bureaucracy—and not the professionalization of
other determinants of the city council agenda, including the politicians that restrain the influence of adminis-
other characteristics of the local administration. trative professionals. Directly measuring the relation-
To give a better sense of the variation underlying ship between the bureaucracy and the political level is
the quantitative results reported below, table 2 pro- difficult, but we believe that several of our controls are
vides information about four of our 98 municipalities. relevant. Strong and durable political coalitions might
Consider for example Albertslund and Herlev, two be harder to form in a context with many political par-
municipalities in the metropolitan area of the Danish ties, and the absence of such coalitions might offer the
capital of Copenhagen, both of which have a popula- bureaucracy greater opportunity to influence the polit-
tion with relatively weak socio-economic backgrounds ical agenda. We control for this possibility by including
and both of which have an average number of council the number of effective parties (Laakso and Taagepera
committees and socialist mayors. Nevertheless, the 1979) in the local council. Based on a similar argu-
council agenda covers many more issues in the former ment, we also control for the number of council mem-
than the latter municipality. This might reflect the bers. Furthermore, we control for the party color of the
slight difference in the socio-economic index or the mayor, as a socialist majority might be more receptive

Table 2. Summary Statistics for Four of the 98 Danish Municipalities (Averages 2007–2013)

Socio- Number of Administrative

economic Issues on the Professionals
Municipality Residents Index Committees Agenda (Percentages) Mayor

Albertslund 27,733 1.68 6 79.3 9.2 Socialist

Herlev 26,666 1.32 5 45.9 6.3 Socialist
Rebild 28,836 0.67 5 75.4 7.8 Liberal/conservative
Kerteminde 23,715 0.82 5 59.3 3.8 Liberal/conservative
Journal of Public Administration Research and Theory, 2018, Vol. 28, No. 2 247

Table 3. The Effect of Administrative Professionals on Agenda Size and Agenda Diversity in Danish Municipalities
2008–2013

Number of Issues Entropy

(1) (2) (3) (4)

Administrative professionalst−1 1.32** (0.61) 1.23* (0.62) 0.03** (0.01) 0.03* (0.01)
Size of the public administrationt−1 −0.01 (0.01) −0.00 (0.00)
Mayor (left = 1) −0.58 (1.13) −0.60 (1.11) −0.02 (0.02) −0.02 (0.02)
Effective parties 0.49 (1.11) 0.41 (1.09) −0.01 (0.02) −0.01 (0.02)
Election year (2009 = 1) 3.43*** (1.11) 3.32*** (1.09) 0.04 (0.03) 0.04 (0.03)
Size of council −0.01 (0.30) 0.00 (0.29) −0.01 (0.01) −0.01 (0.01)
Standing committees 0.58 (0.46) 0.56 (0.46) −0.00 (0.01) −0.00 (0.01)
Resource pressure −0.09 (0.14) −0.10 (0.14) −0.01 (0.00) −0.01 (0.00)
Socio-economic problems −4.87 (9.52) −4.36 (9.55) −0.26 (0.19) −0.26 (0.19)

Downloaded from https://academic.oup.com/jpart/article/28/2/239/4823058 by guest on 02 May 2024

Residents −0.00 (0.00) −0.00 (0.00) −0.00 (0.00) −0.00 (0.00)
Urbanization 1.40** (0.61) 1.43** (0.61) 0.01 (0.01) 0.02 (0.01)
Yt−1 −0.01 (0.04) −0.01 (0.04) 0.02 (0.05) 0.02 (0.05)
Constant −44.81 (56.69) −47.01 (56.87) 3.40*** (1.03) 3.36*** (1.04)
Observations 466 466 466 466

Note: Robust standard errors in parentheses. Year dummies not reported in the table.
*p < .10, **p < .05, ***p < .01 (two-sided tests).

to input from the bureaucracy that invites politicians The percentage of administrative professionals var-
to consider a new problem and potentially take on a ies from 2% of the total number of administrators in
new task. Summary statistics and source descriptions some municipalities to more than 15% in others. Thus,
of the control variables used in the analysis are pre- the span in the diversity of the council agendas can be
sented in Table A1. quite marked between local councils. In the municipal-
To estimate the model, we use fixed effects with ity with only two administrative professionals per 100
municipalities as our panels and robust standard administrative employees, the council agenda is pre-
errors. This choice of estimation is corroborated by dicted to contain 63 issue categories. This figure is 79 in
a Hausman test, which indicates that fixed effects are the municipality with 15 administrative professionals
preferable to random effects. The models also include per 100 administrative employees. Thus, the difference
year dummies as well as a lagged dependent variable to is quite substantial, which supports Hypothesis 1.
account for the autocorrelation revealed by diagnostic The results also show that the administrative pro-
tests.10 To ensure a correct order of time, the main in- fessionals make a difference to the council agenda, not
dependent variables enter the model with a 1-year lag. the size of the administration as such. The effect of the
We explore the direction of causality in greater detail total number of administrators in the municipality is
after the main findings have been reported. statistically and substantially insignificant according to
the results in table 3. This is in accordance with our
Findings theoretical argument. Issues are not brought into the
Our analysis of the 98 Danish municipalities shows policy agenda by ordinary administrators but rather by
that the percentage of administrative professionals skilled administrative professionals typically employed
matters to the local council agendas. Taking our con- in central management positions from which they can
trol variables into account, the diversity of the council advise the politicians and shape the development of the
agenda systematically increases as the share of admin- council agenda.
istrative professionals in the bureaucracy increases. Additional analyses show that the effect also holds
According to the coefficients in table 3, a municipality up when controlling for the sheer size of the council
that has one more administrative professional per 100 agendas measured simply as the number of items on
administrative employees typically also has a little more the dockets. This implies that increasing the number
than one more issue area on its agenda, and its entropy of administrative professionals in the bureaucracy does
score is 0.03 points higher (ranging from 2.9 to 4.2). not just lead to more of the same, but rather to a more
diverse policy agenda, as hypothesized.11 Furthermore,
Additional analyses show that including a count variable does not
10 we have explored the effects of leaving out the measure
change the results, which further corroborates that the statistical
relationship found between our dependent and independent variables
is not merely a product of a common time trend. These results are shown in Supplementary Table S2.
11
248 Journal of Public Administration Research and Theory, 2018, Vol. 28, No. 2

Table 4. The Effect of Agenda Size and Agenda Diversity on the Number of Administrative Professionals in Danish
Municipalities 2008–2013

Administrative Professionals
(1) (2) (3) (4) (5) (6)

Number of issuest−1 −0.01* (0.00) −0.01* (0.00)

Number of issuest−2 −0.00 (0.00) −0.00 (0.00)
Attention entropyt−1 −0.56** (0.23) −0.56** (0.23)
Attention entropyt−2 −0.20 (0.20) −0.18 (0.19)
Size of the public 0.00 (0.00) 0.00 (0.00) 0.00 (0.00) 0.00 (0.00) 0.00 (0.00) 0.00 (0.00)
administrationt−1
Mayor (left = 1) −0.09 (0.08) −0.09 (0.08) −0.09 (0.08) −0.09 (0.08) −0.09 (0.08) −0.10 (0.08)
Effective parties 0.02 (0.09) 0.01 (0.09) 0.02 (0.09) 0.02 (0.09) 0.02 (0.09) 0.02 (0.09)
Election year (2009 = 1) −0.42*** (0.10) −0.42*** (0.10) −0.42*** (0.10) −0.45*** (0.10) −0.43*** (0.10) −0.46*** (0.10)

Downloaded from https://academic.oup.com/jpart/article/28/2/239/4823058 by guest on 02 May 2024

Size of council 0.04 (0.03) 0.04 (0.03) 0.04 (0.03) 0.03 (0.03) 0.04 (0.03) 0.03 (0.03)
Standing committees −0.01 (0.03) −0.01 (0.04) −0.01 (0.03) −0.00 (0.03) −0.01 (0.04) −0.00 (0.03)
Resource pressure 0.01 (0.02) 0.01 (0.02) 0.01 (0.02) 0.01 (0.01) 0.01 (0.02) 0.01 (0.01)
Socio-economic problems 0.92 (0.84) 0.89 (0.86) 0.93 (0.85) 0.91 (0.83) 0.94 (0.86) 0.97 (0.84)
Residents 0.00** (0.00) 0.00** (0.00) 0.00** (0.00) 0.00** (0.00) 0.00** (0.00) 0.00** (0.00)
Urbanization 0.07* (0.04) 0.08* (0.04) 0.07 (0.05) 0.08* (0.05) 0.07 (0.05) 0.07 (0.05)
Yt−1 0.50*** (0.06) 0.49*** (0.06) 0.50*** (0.06) 0.49*** (0.06) 0.49*** (0.06) 0.49*** (0.06)
Constant −6.67 (4.71) −7.04 (4.63) −6.36 (4.74) −4.99 (4.89) −6.18 (4.96) −4.01 (5.22)
Observations 465 465 465 465 465 465

Note: Robust standard errors in parentheses. Year dummies not reported in the table.
*p < .10, **p < .05, ***p < .01 (two-sided tests).

of administrative professionals in order to examine confounding factors, this provides evidence in support
whether it is a mediator of some of the other factors of the claim that administrative professionals draw the
included in the models. This does not turn out to be the attention of politicians to more issues.
case.12 None of the other variables changes direction Based on Hypotheses 2 and 3, we conclude the ana-
or statistical significance when leaving out the admin- lysis by investigating if the effect of administrative
istrative professionals. This offers further indication professionals systematically diminishes with the in-
that the measure of administrative professionals brings volvement of the politicians in the policymaking pro-
added explanatory value to the table. cess. In accordance with Hypothesis 2, the influence
Reverse causality is a concern when studying the of administrative professionals on the council agenda
relationship between the bureaucracy and the council decreases as the number of committees increases. This
agendas. Analytically, a very diverse council agenda suggests that the effect of administrative profession-
might well lead to demands for more administrative als is higher in municipalities where a limited number
professionals to implement decisions from the council of standing committees channels information and
meetings. First, however, this concern is mitigated by the demands into the political system. This result applies
use of lagged measures of the bureaucracy variable in to the number of issue categories on the council agenda
table 3. Second, we have estimated alternative models, as well as the entropy score and is documented by the
where the proportion of administrative professionals is negative sign of the A × B interaction terms in table 5.
estimated as a function of the diversity of the council It is worth noticing that the number of committees had
agendas (lagged 1 or 2 years). Table 4 reports the results a statistically insignificant effect in the noninteractive
of this analysis, which clearly support the interpretation analysis in table 3, which suggests that the influence of
that the administrative professionals positively influence administrative professionals and the number of com-
the council agendas rather than the other way around. mittees on the policy agenda is truly conditional rather
A 1-year lagged agenda diversity does systematically than merely additive.
influence the number of administrative professionals in In accordance with Hypothesis 3, we find a dimin-
the bureaucracy, but it does so negatively. Although this ishing effect of administrative professionals at higher
mechanism may call for further exploration, the results levels of counselor remuneration. To illustrate, aver-
in table 4 strengthen the confidence in the interpreta- age remuneration ranges from DKK 113,000–374,000
tion of the results from table 3. In combination with across municipalities, and if it changes only by one
the statistical control for a range of other potentially standard deviation (DKK 35,000), the marginal effect
of an additional administrative professional on the
These results are shown in Supplementary Table S3.
12
number of issues on the council agenda decreases ten
Journal of Public Administration Research and Theory, 2018, Vol. 28, No. 2 249

Table 5. The Effect of Administrative Professionals on Agenda Size and Agenda Diversity at Increasing Numbers
of Committees and Increasing Average Councilor Remuneration in Danish Municipalities 2008–2013

Number of Issues Number of Issues Entropy Entropy

(A) Administrative professionalst−1 3.58*** (0.96) 3.67*** (1.09) 0.06*** (0.02) 0.05*** (0.02)
(B) Standing committees 2.87** (1.13) 0.03* (0.02)
(C) Councilor remuneration 0.10*** (0.04) 0.00* (0.00)
A×B −0.45*** (0.16) −0.01** (0.00)
A×C −0.01** (0.01) −0.0001* (0.00)
Public administrationt−1 −0.01 (0.01) −0.01* (0.01) −0.00 (0.00) −0.00 (0.00)
Mayor (left = 1) −1.05 (1.18) −0.51 (1.15) −0.03 (0.02) −0.02 (0.02)
Effective parties 0.08 (1.08) 0.54 (1.11) −0.01 (0.02) −0.01 (0.02)
Election year (2009 = 1) 2.88*** (1.07) 3.38*** (1.12) 0.03 (0.02) 0.04 (0.03)
Size of council 0.04 (0.29) 0.18 (0.29) −0.01 (0.01) −0.00 (0.01)

Downloaded from https://academic.oup.com/jpart/article/28/2/239/4823058 by guest on 02 May 2024

Resource pressure −0.11 (0.14) −0.06 (0.13) −0.01 (0.00) −0.01 (0.00)
Socio-economic problems −4.44 (9.54) −4.39 (9.45) −0.26 (0.19) −0.25 (0.19)
Residents 0.00 (0.00) 0.00 (0.00) −0.00 (0.00) 0.00 (0.00)
Urbanization 1.25** (0.54) 1.49** (0.61) 0.01 (0.01) 0.02 (0.01)
Yt−1 −0.01 (0.05) −0.01 (0.04) 0.03 (0.05) 0.02 (0.05)
Constant −41.78 (52.98) −76.29 (58.43) 3.43*** (0.98) 2.87*** (1.05)
Observations 466 466 466 466

Note: Robust standard errors in parentheses. Year dummies not reported. Councilor remuneration is measured in DKK 1000 (Danish
Kroner).
*p < .10, **p < .05, ***p < .01 (two-sided tests).

percent. Hence, even if a very professional political or looked at how bureaucratic interests may shape the
level (i.e., with high remuneration) does not remove policy within particular policy domains. Furthermore,
the effect of administrative professionals, the diminish- the bureaucracy has been ignored in most previous
ing effect is quite substantial. In line with the estimated policy agenda setting studies. This article has shown
effect of committees, this also suggests that administra- that the composition of the bureaucracy matters in the
agenda-setting phase of the policymaking process. In
tive professionals are more important to the council
particular, the share of administrative professionals
agenda when the political level is unable to feed and in the bureaucracy has been shown to have a positive
form the agenda on its own due to a lack of resources effect on the number of issues and composition of the
and time. This is indicated by the negative sign of the local council agendas. This is an important message to
A × C interaction term in table 5. policy agenda setting scholars who have traditionally
To further spell out these conditional effects, we have paid very little attention to the agenda-setting effects
visualized the marginal effect of administrative profes- of the bureaucracy.
sionals at increasing number of committees in figure 1. A particular strength of these results is the fact that
The slope of the line in figure 1 shows how substantial the large-n study allows for the statistical control of
the conditional effect of committees is. With few com- a range of possible confounders. The large number
mittees, an extra administrative professional per 100 of municipalities also allows us to begin examining
administrative employees brings, on average, almost conditional effects. Particularly, the article focused
three additional issue areas to the council agenda. This on various indicators of the involvement of politi-
is more than twice the average effect in table 3. At about cians in the policymaking process and revealed an im-
nine committees, the other end of our measure of com- portant interaction effect between two indicators of
mittees, the number of professionals no longer makes the involvement of politicians, namely the committee
a difference to the council agenda. This result may be system and the counselors’ remuneration in relation
interpreted as evidence that the bureaucracy takes over to the effect of administrative professionals. A plaus-
and helps supplement the policy agenda by focusing on ible interpretation of this effect is that administrative
issues that are ignored in narrow committee systems. professionals take over and help to qualify and sup-
plement the council agenda by focusing on issues that
tend to be ignored in narrow committee systems or
CONCLUSION when council members do not have the time to pro-
Bureaucracy matters. We already knew that, but dispro- cess problems themselves.
portionally many studies have examined the effect of the The political systems at the local and national levels
bureaucracy from an implementation perspective and/ are obviously not identical. In terms of generalizability,
250 Journal of Public Administration Research and Theory, 2018, Vol. 28, No. 2

Downloaded from https://academic.oup.com/jpart/article/28/2/239/4823058 by guest on 02 May 2024

Figure 1. The Average Marginal Effect of Administrative Professionals on Agenda Diversity at Increasing Numbers of Committees and at
Increasing Councilor Remuneration. Note: The Figure is Based on the Results in table 5. Dashed Lines Represent 95% Confidence Intervals.

however, we should expect our results to apply beyond Or is it an unintended side effect? In pursuing these
Danish municipalities. The professionalization of the new research questions, much can probably be learned
bureaucracy is a trend observed in most Western politi- by further investigating the local government level of
cal systems, and there is reason to expect to see the policymaking.
effects on the policy agenda identified in our analysis,
depending on the institutional setup of the political
system, such as the committee structure. That said, and Supplementary Material
given our finding that bureaucratic influence dimin- Supplementary material is available at the Journal of
ishes with the professionalization of politicians, our Public Administration Research and Theory online
results may travel best at the sub-national level (local (www.jpart.oxfordjournals.org).
and regional), where the professionalization of poli-
ticians can vary considerably and sometimes be very
References
low. This is rarely the case at the national level. The
Acs, Alex. 2015. Which statute to implement? Journal of Public Administration
evidence from the studies of US state legislatures by
Research and Theory 26:493–506.
Boushey and McGrath (2017) and Hird (2005) hint Baumgartner, Frank, and Bryan Jones, eds. 2002. Policy dynamics. Chicago,
at the potential to generalize our results to the subna- IL: Univ. of Chicago Press.
tional level in other countries. Baumgartner, Frank, Sylvain Brouard, Christoffer Green-Pedersen,
Both the general and conditional effects obviously Bryan Jones, and Stefaan Walgrave. 2011. The dynamics of policy change
in comparative perspective. Special Issue in Comparative Political Studies
raise new questions about the mechanisms behind these
44.
effects as well as the scope conditions of bureaucratic Baumgartner, Frank, and Bryan Jones. 2015. The politics of information.
influence. First and foremost, the quantitative results Chicago, IL: Univ. of Chicago Press.
invite further investigation of why we observe these Baumgartner, Frank, Bryan Jones, and John Wilkerson. 2011. Comparative
effects. What is it about administrative profession- studies of policy dynamics. Comparative Political Studies 44:947–1119.
Baumgartner, Frank, Bryan Jones, and Peter B. Mortensen. 2017. Punctuated
als that makes them important to the policy agenda?
equilibrium theory: Explaining stability and change in public policymak-
And how do they exert their influence? An avenue for ing. In Theories of the policy process, 4th edition, ed. Paul Sabatier and
comparative research could also be to compare the Christopher Weible, 55–101. Boulder: Westview Press.
effects of administrative professionals with other types Bevan, Shaun, Peter John, and Will Jennings. 2011. Keeping party programmes
of bureaucrats, such as politically appointed advisors. on track: The transmission of the policy agendas of executive speeches
to legislative outputs in the United Kingdom. European Political Science
Kingdon (1995), for instance, argued that the latter are
Review 3:395–417.
more powerful agenda setters than civil servants, but Bhatti, Yosef, Asmus Olsen, and Lene Pedersen. 2009. The effects of adminis-
this proposition has yet to be systematically examined. trative professionals on contracting out. Governance 22:121–37.
Furthermore, given the fact that this article shows that ———. 2011. Administrative professionals and the diffusion of innovations:
administrative professionals make a difference to the the case of citizen service centres. Public Administration 89:577–94.
Boadway, Robin, and Anwar Shah. 2009. Fiscal federalism: Principles and
policy agenda also calls for further investigation into
practices of multiorder governance. New York: Cambridge Univ. Press.
the decision to hire few or many administrative pro- Bonafont, Laura Chaqués. 2016. Interest groups and agenda setting. In
fessionals. Is it an agenda-setting effect that is calcu- Handbook of public policy agenda setting, ed. N. Zahariadis, 200–16.
lated and anticipated by the local deciding coalition? Cheltenham, UK: Edward Elgar.
Journal of Public Administration Research and Theory, 2018, Vol. 28, No. 2 251

Boushey, Graeme and Robert McGrath. 2017. Experts, amateurs, and bureau- Kingdon, John. 1995. Agendas, alternatives, and public policies. New York:
cratic influence in the American states. Journal of Public Administration Harper Collings.
Research and Theory 27:85–103. Kommunernes og Regionernes Løndatakontor (KRL). 2017. Data on admin-
Boydstun, Amber, Shaun Bevan, and Herschel Thomas. 2014. The impor- istrative professionals’ employment in Danish municipalities 2007–2013.
tance of attention diversity and how to measure it. Policy Studies Journal Data send to the authors. Data available upon request.
42:173–96. Krause, George A. 1999. A two way street: the institutional dynamics of the
Bækgaard, Martin. 2010. Self-selection or socialization? A dynamic analysis of modern administrative state. Pittsburgh, PA: Univ. of Pittsburgh Press.
committee member preferences. Legislative Studies Quarterly 35:337–59. Laakso, Markku, and Rein Taagepera. 1979. The “effective” number of par-
———. 2011. Committee bias in legislatures with a high degree of party cohe- ties: a measure with application to West Europe. Comparative Political
sion: Evidence from Danish municipalities. European Journal of Political Studies 12:3–27.
Research 50:315–35. Lindblom, Charles. 1959. The science of muddling through. Public
Bækgaard, Martin, and Carsten Jensen. 2011. The dynamics of competitor Administration Review 19:79–88.
party behaviour. Political Studies 60:65–80. Loftis, Matthew, and Peter B. Mortensen. 2017. Collaborating with the
Bækgaard, Martin, Carsten Jensen, Peter Mortensen, and Søren Serritzlew. Machines: a hybrid method for classifying policy documents. Policy
2014. Local news media and voter turnout. Local Government Studies Studies Journal (forthcoming).
40:518–32. March, James, and Johan Olsen. 1989. Rediscovering institutions. New York:

Downloaded from https://academic.oup.com/jpart/article/28/2/239/4823058 by guest on 02 May 2024

Carpenter, Daniel. 2001. The forging of bureaucratic autonomy: reputa- The Free Press.
tions, networks, and policy innovation in executive agencies, 1862–1928. Meier, Kenneth J. 2007. The public administration of politics, or what political
Princeton, NJ: Princeton Univ. Press. science could learn from public administration. PS: Political Science and
Christensen, Jørgen, Opstrup, Niels, and Robert Klemmensen. 2014. Politics 40:3–9.
Politicization and the replacement of top civil servants in Denmark. Mills, C. Wright. 1956. The power elite. Oxford: Oxford Univ. Press.
Governance 27:215–41. Mortensen, Peter B., Christoffer Green-Pedersen, Gerard Breeman,
Cobb, Roger W., and Charles D. Elder. 1983. Participation in American Laura Bonafont, Will Jennings, Peter John, Arco Timmermans, and
politics: The dynamics of agenda-building. Baltimore, MD: The Johns Anna Palau. 2011. Comparing government agendas: executive speeches
Hopkins Univ. Press. in the Netherlands, United Kingdom, and Denmark. Comparative Political
Dahl, Robert. 1956. A preface to democracy. Chicago, IL: Univ. of Chicago Studies 44:973–1000.
Press. Mortensen, Peter B., and Henrik Seeberg. 2016. Why are some policy agendas
——. 1961. Who governs? Democracy and power in an American city. New larger than others? Policy Studies Journal 44:156–75.
Haven, CT: Yale Univ. Press. Novotny, Vilém, Michel Perottino, and Martin Polásek. 2016. Political parties
DJØF. 2017. DJØF gets member number 90,000, DJØF Newsletter, 4 and the policy agenda. In Handbook of Public Policy Agenda Setting, ed.
November 2016. Nikolaos Zahariadis, 217–35. Cheltenham, UK: Edward Elgar.
Egeberg, Morten, and Jarle Trondal. 2009. Political leadership and bureau- Peters, Guy. 2010. Politics of bureaucracy, 6th ed. Abingdon, UK: Routledge.
cratic autonomy: Effects of agencification. Governance 22:673–88. Roberts, Jennifer, and Michael Dietrich. 1999. Conceptualizing professional-
Eissler, Rebecca, Annelise Russell, and Bryan Jones. 2014. New avenues for the ism. American Journal of Economics and Sociology 58:977–98.
study of agenda setting. Policy Studies Journal 42:71–86.
Green-Pedersen, Christoffer, and Stefaan Walgrave. 2014. Agenda setting, Schattschneider, Elmer. 1960. The semisovereign people. New York: Hold,
policies, and political systems: a comparative approach. Chicago, IL: Univ. Rinehart, and Winston.
of Chicago Press. Serritzlew, Søren, Jens Blom-Hansen, and Asbjørn Skjæveland. 2010. Portfolio
Groseclose, Tim. 1994. The committee outlier debate: a review and a reexami- allocation or policy horizons? Determinants of coalition formation in
nation of some of the evidence. Public Choice 80:265–73. Danish local government. Local Government Studies 36:843–66.
Hird, John. 2005. Policy analysis for what? The effectiveness of nonpartisan Sheingate, Adam. 2006. Structure and opportunity: Committee jurisdiction
policy research organizations. Policy Studies Journal 33:83–105. and issue attention in Congress. American Journal of Political Science
Hunter, Floyd. 1953. Community power structure: a study of decision makers. 50:844–59.
Chapel Hill, NC: Univ. of North Carolina Press. Shepsle, Kenneth, and Barry Weingast. 1985. Policy consequences of govern-
Jacobsen, Dag. 2006. The relationship between politics and administration: ment by Congressional subcommittees. Proceedings of the Academy of
The importance of contingency factors, formal structure, demography, and Political Science 35 (4):114–31.
time. Governance 19:303–23. Simon, Herbert. 1947. Administrative behavior. New York: The Free Press.
Jennings, Will, Shaun Bevan, Arco Timmermans, Gerard Breeman, Sylvain Brouard, Smith, Bo, and Jørgen Grønnegaard Christensen. 2016. Embedsmænd som
Laura Bonafont, Christoffer Green-Pedersen, Peter John, Peter Mortensen, dydens vogtere eller demokratiets tjenere. Politica 48:373–91.
and Anna Palau. 2011. The effects of core functions of government on the Soroka, Stuart. 2002. Agenda-setting dynamics in Canada. Vancouver,
diversity of executive agendas. Comparative Political Studies 44:1001–30. Canada: UBC Press.
Jennings, Will, and Peter John. 2009. The dynamics of political attention: Sprague, Mary. 2008. The effects of measurement and methods decisions
Public opinion and the Queen’s speech in the United Kingdom. American on committee preference outlier results. Political Research Quarterly
Journal of Political Science 53:838–54. 61:309–18.
———. 2011. Punctuations and turning points in British politics: The policy Statistics Denmark. 2017. Data on councilor remuneration in Danish
agenda of the Queen’s speech, 1940–2005. British Journal of Political Municipalities. www.statistikbanken.dk (accessed December 27, 2017).
Science 40:561–86. Svara, James H. 2001. The myth of the dichotomy: complementarity of politics
Jennings, Will, Shaun Bevan, and Peter John. 2011. The British Government’s and administration in the past and future of public administration. Public
political agenda: The speech from the Throne, 1911–2008. Political Administration Review 61:176–83.
Studies 59:74–98. Talbert, Jeffery, Bryan Jones, and Frank Baumgartner. 1995. Nonlegislative hear-
Jones, Bryan. 1994. Reconceiving decision-making in democratic politics. ings and policy change in congress. American Journal of Political Science
Chicago, IL: Univ. of Chicago Press 39:383–405.
Jones, Bryan. 2001. Politics and the architecture of choice: bounded rationality UgebrevetA4. 2015. LO: In 15 years, only half will be a union-member (25
and governance. Chicago, IL: Univ. of Chicago Press. March 2015).
Jones, Bryan, and Frank Baumgartner. 2005. The politics of attention: how Vederlagsbekendtgørelsen. 2016. Departmental order no. 1530, dated
government prioritizes problems. Chicago, IL: Univ. of Chicago Press. December 9 2016.
252 Journal of Public Administration Research and Theory, 2018, Vol. 28, No. 2

Weber, Max. 1922. Makt og byråkrati. Oslo, Norway: Gyldendal Norsk Wilson, James. 1989. Bureaucracy. New York: Basic Books.
Forlag. Wilson, Woodrow. 1887. The study of public administration. Political Science
West, William F. and Connor Raso. 2013. Who shapes the rulemaking Quarterly 56:197–222.
agenda? Implications for Bureaucratic Responsiveness and Bureaucratic Workman, Samuel. 2015. The dynamics of bureaucracy in the US Government.
Control. Journal of Public Administration Research and Theory 23:495–519. Cambridge, MA: Cambridge Univ. Press.
Wildavsky, Aron. 1964. The politics of the budgetary process. Boston, MA: Zahariadis, Nikolaos, ed. 2016. Handbook of public policy agenda setting.
Little, Brown, and Company. Cheltenham, UK: Edward Elgar.

Appendix

Table A1. Summary Statistics and Data Sources for Control Variables

Variable Mean Std. Min Max Data Source

Mayor (left-wing = 1) 0.5 0.5 0 1 Ministry of the Interior and Social Affairs (www.

Downloaded from https://academic.oup.com/jpart/article/28/2/239/4823058 by guest on 02 May 2024

noegletal.dk)
Effective number of 3.7 0.8 1.9 6.8 Statistics Denmark (www.statistikbanken.dk) and
parties www.kmdvalg.dk
Size of council (# of 26.0 5.4 15 55 Statistics Denmark (www.statistikbanken.dk)
members)
Resource pressure 100.6 4.7 85.9 109.9 KORA and Statistics Denmark
Socio-economic problems 0.95 0.2 0.5 1.8 Ministry of the Interior and Social Affairs (www.
noegletal.dk)
Residents 58,798 63,730 12,399 559,440 Ministry of the Interior and Social Affairs (www.
noegletal.dk)
Urbanization 83.8 11.8 57.7 100 Ministry of the Interior and Social Affairs (www.
noegletal.dk)

Note: N = 94 (municipalities) × 6 (years) = 564 observations on all variables.

Table A2. Correlation Matrix

Administrative Professionals Size of Public Administration Standing Committees

Size of public administration 0.41

Standing committees −0.01 0.17
Residents 0.40 0.99 0.18

Table A3. Factor Analysis of Three Indicators of Administrative Professionals

Administrative Professionals Based on… Factor Loadings

…union membership 0.91

…collective agreement 0.85
…education 0.89

Note: Principal factor analysis. Eigenvalue factor 1 = 2.35; Cronbach’s alpha = 0.90.

Figure A1. Distribution of Administrative Professionals Across Categories.

Journal of Public Administration Research and Theory, 2018, Vol. 28, No. 2 253

Table A4. Summary Statistics, Danish Municipalities 2008–2013

Variable N Mean Std. Min Max

Administrative professionals per 100 public administrators Overall 560 6.0 2.2 2.2 15.5
Between 94 2.5 2.5 14.1
Within 6 0.66 3.9 8.2
Number of employees in the public administration Overall 563 797.9 959.3 130.0 8412.0
Between 94 961.5 140.0 7968.3
Within 6 49.0 258.2 1249.6
Standing committees Overall 563 5.9 1.6 2 14
Between 94 1.4 2 10.7
Within 6 0.7 3.9 9.3
Overall 563 192.3 39.0 113.1 436.0
Councilor remuneration (DDK 1000) Between 94 33.1 124.5 325.9

Downloaded from https://academic.oup.com/jpart/article/28/2/239/4823058 by guest on 02 May 2024

Within 6 21.0 135.6 306.2
Residents (1,000s) Overall 563 58.8 63.7 12.4 559.4
Between 94 63.9 13.4 534.1
Within 6 20.3 34.6 841.3
Number of issues on council agendas Overall 563 68.0 10.7 36 110
Between 94 9.5 43 100.8
Within 6 5.0 50.6 86.7
Entropy of council agendas Overall 563 3.8 0.2 2.9 4.2
Between 94 0.2 3.2 4.1
Within 6 0.1 3.4 4.1

Note: N = 94 (municipalities) × 6 (years) = 563 observations. Four municipalities are left out due to missing observations on one or more
variables. The between-estimate gives the average across municipalities of the average estimate over time for each municipality. The within-
estimate gives the average across municipalities of the average change within each municipality over time.