Policy preferences and ministerial structures: Immigrant integration governance in Western Europe, 1997 – 2017

This article tests an often-stated assumption in the “ machinery of government ” literature: that government parties ’ issue and policy preferences affect the ministerial structure. Using a new Dataset on Immi-grant Integration Governance (DIIG) that documents the ministerial structure of immigrant integration in 16 Western European countries during 1997 – 2017, the analysis finds that divergence and change characterize the ministerial structure. The analysis finds that right-oriented governments are more inclined to couple integration with immigration and justice than are left-oriented governments, indicating different policy preferences. However, it does not find that governments with radical parties are more inclined to signal “ integration ” in ministerial titles. The study's comparative approach challenges the generalizability of existing knowledge on drivers of ministerial changes. Additionally, it fills a gap in current immigrant integration literature: the study of the horizontal governance structure.


| INTRODUCTION
The horizontal organization and reorganization of an issue matter; different organizational structures produce different policy outcomes.The horizontal structure-how tasks are distributed among the same hierarchical level, and how different issues are linked or decoupled from each other (Egeberg, 2012)-broadly defines interests and goals to be pursued, forms criteria on which to base comparisons of alternative polices, and influences which conflicts rise to the top level in the policymaking process (Egeberg, 2012;Hammond, 1986).Additionally, the horizontal structure is an opportunity for the government to express priorities, (re)frame an issue, and meet policy challenges and external pressure (Christensen, Egeberg, Laegreid, & Aars, 2014;Davis, Weller, Eggins, & Craswell, 1999;White & Dunleavy, 2010).
The ministerial structure encompasses the complexity and dilemmas of horizontal governance, and has been addressed by the literature on "machinery of government changes" (Mortensen & Green-Pedersen, 2014).This literature identifies several drivers that affect both the number of ministries and the number of changes in the ministries' issue content, for example, administrative challenges and external pressure (White & Dunleavy, 2010), developments in the policy agenda (Mortensen & Green-Pedersen, 2014), and internal party politics (Davis et al., 1999;Heppell, 2011;White & Dunleavy, 2010).Mortensen and Green-Pedersen (2014, p. 166) argue that although the literature on portfolio allocation has shown that the allocation of ministries is important to government parties, the political dynamic behind what determines the issue content of ministries deserves more theoretical elaboration and empirical investigation-not necessarily by identifying more explanations, but by unfolding the political logic behind explanations already identified.Issue preference has been important in previous studies of changes in ministries' issue content (Mortensen & Green-Pedersen, 2014;White & Dunleavy, 2010), and stems from the idea of issue ownership, implying that political parties prefer high attention to some issues rather than others (Karlsen & Aardal, 2016, p. 262).However, that the government's policy preferences, that is, the actual policies that should be pursued on a given issue, might affect the ministerial structure is a constantly repeated, but yet unexplored, assumption (Davis et al., 1999;Heppell, 2011;Mortensen & Green-Pedersen, 2014).This article addresses this gap by analyzing both elements, reflecting two dilemmas a government faces when designing the ministerial structure.First, whether an issue should be signaled or not in the ministerial titles, reflecting the government's issue preferences.Second, how the issue should be coupled with or decoupled from other related issues, reflecting the government's policy preferences.To exemplify using the integration of immigrants (hereafter immigrant integration 1 ) as an issue: A government might express its issue preference for "integration" by either including it or excluding it from the ministerial titles, depending on whether the government wants to draw attention to or deflect attention from the issue.The government's policy preferences-whether it prefers restrictive or permissive integration policies-might affect what other issues the government wants to couple integration with in the ministerial structure, for example, immigration and justice (restrictive) versus social policy and inclusion (permissive).
To analyze whether and how issue and policy preferences affect the ministerial structure, it is necessary to move beyond the traditional approach in the machinery of government literature, that is, longitudinal case studies concentrating on changes in the ministerial structure as a whole in one or a few countries.This article studies how one single issue is structured at the ministerial level across countries over time.Davis et al. (1999, p. 13) argue that the literature needs comparative analyses to test if drivers of machinery changes are place-specific in addition to time-specific.This study will do so by analyzing the ministerial structure of immigrant integration, using a new data set compiled by the author, the Dataset on Immigrant Integration Governance (DIIG) that documents the national structure of the immigrant integration field in 16 Western European countries from 1997 to 2017 (Hernes, 2020).The article asks two questions: First, how has immigrant integration been coupled with other issues and been signaled in the ministerial structure over time, and are there cross-national differences?The article focuses specifically on whether the integration issue is coupled with the immigration and justice field, as a contrast to other issues such as social affairs.Second, the article asks: Do government parties' issue and policy preferences explain cross-and within-country developments?
Immigrant integration-a politically salient and "wicked" issue infused with symbolic politics-constitutes a "most likely" case for observing issue and policy preferences manifested in the ministerial structure.First, it is more likely that political parties actively use the ministerial structure as a tool to draw attention to or deflect attention from a politically salient issue than they do for less salient issues.Second, it is more likely to identify policy preferences in the ministerial structure for "wicked issues.""Wicked issues" is a term used to describe complex social problems that have no agreed upon problem definition or solution, are transboundary, ambiguous, and uncertain, and cross-traditional sectorial lines (Rittel & Webber, 1973).Thus, wicked issues have the potential to be coupled with multiple related issues, opening the possibility for multiple alternative structures.Third, immigrant integration policies often are infused with symbolic politics (Borevi, Jensen, & Mouritsen, 2017;Joppke, 2017), thus it is also plausible to assume that political parties would be more inclined to make symbolic changes in the ministerial structure for this particular policy issue.
The empirical focus on immigrant integration and the DIIG data set also contribute to fill a gap in the immigrant integration literature.Although research on immigrant integration has become one of the fastest growing areas in political science (Goodman, 2015(Goodman, , p. 1906)), empirical studies have mostly focused on policy content (Goodman, 2012(Goodman, , 2015;;Hernes, 2018;Joppke, 2007Joppke, , 2017) ) or on vertical multilevel governance (Dekker, Emilsson, Krieger, & Scholten, 2015;Hernes, 2017;Scholten, Collett, & Petrovic, 2017).The horizontal governance structure, however, has gained less attention.While some qualitative studies mention the horizontal dimension when describing the governance structure of integration (Desiderio & Weinar, 2014;Scholten et al., 2017), this literature lacks spatial and longitudinal analyses that systematically explore the horizontal structure as the dependent variable.
The article starts with a description of the theoretical assumptions for how issue and policy preferences could affect the ministerial structure, and how these preferences could be reflected in four stylized horizontal structures.Then follows a presentation of hypotheses predicting which types of ministerial structures different governments would prefer for the integration issue.Using the DIIG data set, the descriptive analysis shows how divergence and change characterize the ministerial structures in Western Europe the last 20 years.The explanatory analysis finds that right-oriented governments are more inclined to couple integration with immigration and/or justice than left-oriented governments are.However, it does not find that governments that include a radical party are more inclined to signal integration in ministerial titles.The discussion elaborates on the study's limitations and advantages by applying a broader comparative approach than earlier studies of machinery changes and argues that the findings challenge the generalizability of existing knowledge on drivers of ministerial changes.Lastly, the conclusion presents ways to exploit the new DIIG data set in future analyses to provide new insight not only for the literature on immigrant integration, but also for the public administration literature more generally.

| COUPLING AND SIGNALING IN THE MINISTERIAL STRUCTURE
Earlier studies show that the ministerial structure actually is rather stable over time (Mortensen & Green-Pedersen, 2014).National traditions for how to organize the ministerial structure limit the possibility for radical change because relevant actors will adapt to the existing practice (Thelen, 1999).Additionally, ministerial changes have substantial costs, both financially and administratively (White & Dunleavy, 2010, p. 7).However, earlier studies show that while the ministerial structure is relatively stable for some issues, others experience instability (Davis et al., 1999, pp. 30-31;White & Dunleavy, 2010, p. 26).In Western European countries, the authority to decide the ministerial structure lies with the government parties, and is usually the prerogative of the prime minister or the party leaders in a coalition government. 2As a contrast, in the United States, the control over the federal administration is divided between the Presidency and Congress, and all U.S. ministerial reorganizations must be approved by Congress (White & Dunleavy, 2010, p. 23).When deciding on the ministerial structure of specific issues, the governing parties must consider two elements.One is how the issue should be coupled with or decoupled from other related issues, reflecting their policy preferences.The other is whether the issue should be signaled in the ministerial titles, reflecting their issue preferences.

| Policy preference: The coupling or decoupling of issues
Coupling as an analytical concept has often been used to study the vertical governance dimension, but Philipp Trein (2016, p. 420) argues that this concept has unexploited potential in the study of horizontal governance.The horizontal structure might express a government's alternative policy preferences by how an issue is coupled with or decoupled from other related issues.Coupling is particularly relevant for wicked issues that often cross traditional sectorial boundaries and have the potential to be coupled with multiple issues.Issues within one organizational unit are more likely to be coordinated than are those belonging to different units (Egeberg, 2012, p. 159).An official in a particular unit is expected to consider only certain alternatives and their consequences, while leaving it to other units to consider their respective alternatives and consequences (Egeberg, Gornitzka, & Trondal, 2016, p. 33).Thus, the horizontal structure broadly defines the interests and goals to be pursued, and the considerations and alternatives that should be treated as relevant (Egeberg, 2012, p. 158).A reorganization of the ministerial structure by changing which issues are coupled or decoupled could portray the government's (new) framing of an issue (Christensen et al., 2014, p. 43), reflecting their policy preferences.
The coupling of policy issues might be a product of different types of structural changes (White & Dunleavy, 2010, p. 14): mergers of previously separate ministries; demergers by moving sets of functions previously carried out as a subcomponent in a broader ministry into a separate ministry; transfers of functions, where two or more ministries remain in being, but a set of responsibilities is transferred from one to another; and terminations, where a ministry is closed and its remaining activities are transferred to other ministries.

| Issue preference: Signaling in the ministerial structure
The government can use the ministerial structure to enhance or signal political priorities or to give extra weight to an issue (Heppell, 2011, p. 426).A government that wants to draw attention to an issue might include it in the ministerial titles, and a government that wants to deflect attention from a political issue might exclude it (Mortensen & Green-Pedersen, 2014, pp. 168, 172).Including an issue in the ministerial titles could reflect a real investment in the given issue, for example increased funding.Even without increased funding, it could also be a symbolic change intended to signal to the public that the government is tackling an issue.Either way, including an issue in ministerial titles is a tool for the government to express their prioritized issues to the public, irrespective of whether they actually follow up with increased funding.Not every issue can have its own ministry or minister; thus, the government must prioritize.Signaling an issue in ministerial titles is expected to be preferred by government parties that have a particular ownership of the issue and/or that would benefit from drawing a high level of attention to it (Mortensen & Green-Pedersen, 2014, pp. 168, 172;Karlsen & Aardal, 2016).
Changes concerning the inclusion or exclusion of an issue in the ministerial titles could be made without relocation in the ministerial structure: the issue can remain within the same ministry, while the name is included or excluded from the title or a specialized ministerial position is established or abolished within a more generalized ministry.Intuitively, changing the ministry title seems like a more substantial change than creating a specialized minister within a generalized ministry.However, a change in the ministry's title could merely be a symbolic change (i.e., one involving no additional funding).If a specialized minister for integration is established within a generalized ministry, even though the new post involves no increased funding, it still has gained (one would assume) a dedicated high-ranking politician who will work on this issue.Additionally, ministers are mostly described in the media by their ministerial title, and not by the ministry's title.Thus, as a tool to signal to the public the government's prioritized issues, it is not clear whether a change in the ministry's title is more profound than establishing a specialized minister would be.Consequently, the analysis does not make separate theoretical expectations for one or the other.

| Four stylized horizontal structures
The two dilemmas that politicians face concerning if and how to signal an issue and couple an issue in the ministerial structure create four stylized horizontal structures, presented in Table 1.Table 1 distinguishes whether the main issue under study-Policy A-is signaled or not in the organizational title, and whether Policy A is coupled with or decoupled from (an)other issue(s)-Policy B. For integration, the coupling with or decoupling from immigration or social affairs is often of particular interest.For the energy field, it may be climate or transport.For disabilities, it may be health or employment.Which issues are substantially relevant to focus on will depend on the research question.Classification of organizations within-and movement between-these categories capture changes both with and without relocations in the ministerial structure.Whereas vertical movement implies changes in how the government signals an issue, horizontal and diagonal movement also implies changes regarding how the issue is coupled with other issues.

| Parties' position on integration policies and hypothesized structure preferences
If political parties that have a particular ownership of an issue would be expected to signal the issue in the ministerial structure, and to structurally couple the issue with other issues that support their policy preferences, changes in the composition of government parties could lead to changes in the ministerial structure if the new government parties have different issue and policy preferences than their predecessors did.Studies documenting political parties' issue ownership and positions on immigrant integration policies make it possible to derive hypotheses on how different (coalition) governments might manifest these positions in the ministerial structure.In a study of political parties' policy homogeneity across 17 European countries, Laurenz Ennser (2012, p. 162) found that political parties classified within the same party family have relatively similar preferences on integration policies (with the "liberal" party family as an exception).The parties' positions follow the classic left-right economic dimension: Radical-right parties are the most inclined to support restrictive integration policies; radical-left parties are the least inclined, and the other parties have positions that are more moderate and are distributed in-between.Additionally, several studies have shown that the government composition of political parties affects the direction of integration policies on the permissive-restrictive scale (Akkerman, 2012;Borevi et al., 2017;Goodman, 2012).Thus, that parties within the same party family have relatively similar issue and policy preferences makes it possible to examine if governments that include the same types of political parties also have similar preferences for how to organize integration in the ministerial structure.However, for the integration issue, what structural logic might be derived from different issue and policy preferences?
First, turning to the question of coupling integration with or decoupling it from other issues: As a wicked issue, integration encompasses multiple other issues: immigration, health, education, employment, housing, crime, discrimination, and so forth.Thus, the issues it could potentially be coupled with are numerous.In the integration literature on public policies, the coupling of integration with immigration has received particular attention, because studies have shown that integration policies are increasingly used as means to limit immigration (Baldi & Goodman, 2015;Hernes, 2018).Although immigration and integration policies focus on the same target group, the policies focus on different areas.The focus of immigration policies is mostly on questions of access to the country, for example, migration, visas, asylum, and the process until an applicant is granted a legal residence permit.The focus of integration policies is often on the process after a residence permit is granted (e.g., the cultural, social, and economic integration into society).However, parties that want to use integration policies as means to restrict further immigration could be more inclined to couple integration with topics such as immigration, national security, crime, and terrorism in the ministerial structure-often responsibilities under a ministry of immigration and/or justice.Alternatively, integration might be coupled with welfare issues such as social affairs, employment and inclusion, or other issues that do not link integration directly with immigration and/or justice.The coupling of integration with or the decoupling of it from either of these two groups of issues could be related to the governments' general positions on the restrictive-permissive scale of integration policies.Right-oriented parties generally support more restrictive integration policies than left-oriented parties do (Ennser, 2012, p. 162;Akkerman, 2012), and thus, could be expected to couple integration with immigration and/or justice.Left-oriented governments could be expected to couple integration with issues they have particular ownership of, for example, social affairs and employment policies, as opposed to immigration and justice-issues typically "owned" by the right (Petrocik, 1996, pp. 831-32).Thus, the first hypothesis states that right-oriented governments will be more inclined to couple integration with immigration and/or justice in the ministerial structure than left-oriented governments will be.
Second, integration is a politically salient issue, and because of their opposing political positions, political parties could have different incentives to keep this issue on the political agenda.Both radical-left and radical-right parties generally have more radical positions than do parties closer to the center concerning immigrant integration (Ennser, 2012, p. 162); consequently, these parties might have a stronger incentive to express these positions in the ministry's and/or the minister's title than moderate parties do.The integration issue has great potential for both radical-left and radical-right parties to mark their ground with, for example, permissive or restrictive policies regulating immigrants' access to general welfare benefits, liberal or restrictive requirements to obtain citizenship, and so forth.This is not only the case for radical-right parties where immigration and integration most often is the issue of greatest concern to their voters (Akkerman, 2012, p. 511), but also increasingly so for radical-left parties that can brand themselves as an opposition to the restrictive policies of the right (Bale, Green-Pedersen, Krouwel, Luther, & Sitter, 2010).Following this logic, the hypothesis would be that governments that include a radical party will be more inclined to signal integration in the ministerial titles than governments without such parties will be.However, an opposite hypothesis might also be presented.Although radical parties have more radical positions on integration, these positions may not lead to preferences to signal the issue in the ministerial titles.For radical-left parties an alternative logic supporting permissive policy preferences could be at play: Signaling integration in the ministerial titles implies singling out immigrants as a target group as opposed to the rest of the population (Schneider & Ingram, 1993).Not signaling integration in the ministerial titles is more in line with a mainstreaming approach to organizing the integration field (Scholten et al., 2017), and avoids constructing immigrants as a group in need of targeted policies, dividing them from the majority population.Thus, this alternative logic would imply that radical-left parties would not signal integration in the ministerial titles.For radical-right parties, an alternative logic could be that because radical-right parties oppose immigration, they are less likely to signal integration because they do not want immigrants to come to the country in the first place, and do not want the state to use time and resources to integrate immigrants.Consequently, radical-right parties may be more inclined to signal "immigration" or other issues that signal their restrictive immigration policies, and not "integration."

| DATASET ON IMMIGRANT INTEGRATION GOVERNANCE (DIIG)
This study presents a unique comparative data set, DIIG, mapping the national structure of the immigrant integration field in 16 Western European countries from 1997 to 2017 (Hernes, 2020).The following variables from the data set have been used to conduct the analysis: country, year, descriptive account of structural change, title of ministry responsible for integration, title of minister responsible for integration, title of ministry responsible for immigration, type of change, classification of coupled issues (seven categories), classification of integration as signaled/not signaled, and classification of integration as coupled/decoupled from immigration and/or justice.
The main source to document historical developments in the ministerial structure has been annual country reports and cross-country analyses from the European Migration Network and the INTERACT project.In these annual reports, country experts and government authorities have reported systematically on the governmental bodies responsible for integration and immigration.Not all the countries have annual reports for the entire period of analysis; thus, findings from the reports are supplemented and validated by the following sources: (a) European Website on Integration and the EU Immigration Portal, (b) governmental webpages, (c) official historical online archives of ministries and ministers in government cabinets, (d) national databases on state administration (for Norway and Ireland), (e) policy documents, and (f) academic books and articles with detailed case studies of the selected countries.For a detailed list of references for each country, see the online list of references (Hernes, 2020).
Two researchers compiled the data set.The first researcher documented the governance structure with specific references to the original sources, and identified observations that were unclear; for example, for a few countries, the sources on the historical structure differed concerning the exact years for ministerial changes or the exact titles of ministers.The second researcher surveyed all original references to validate the data set, and thereafter, uncertainties on how to classify and code countries were discussed and decisions were jointly arrived at to ensure similar classifications across countries.For a detailed description of the variables' operationalization, see the online codebook (Hernes, 2020).

| Data on government parties: ParlGov
The data set also includes variables on national elections, governments, and government parties and their party family affiliation [according to their position in an economic (state/market) and a cultural (liberty/authority) left/right dimension], extracted from the ParlGov database (for information on the database, see Döring & Manow, 2016). 3The hypotheses distinguish between four types of governments, according to whether the government (a) is right-or left-oriented and (b) includes a radical party or not.In his analysis of the party families' policy positions on immigration and integration, Ennser (2012) finds that parties in the center of the political spectrum also have a centered position on immigration and integration policies.Consequently, the inclusion of a centrist party in a government is not expected to affect the government's preferences in a particular direction.Thus, the eight party families identified by Döring and Manow (2016) will be further classified as described in Table 2.
The four types of governments could be either with or without a center party (C): • Right-oriented governments with a radical-right party: R + RR(+ C) • Right-oriented governments without a radical-right party: R(+ C) • Left-oriented governments with a radical-left party: L+ RL(+ C) • Left-oriented governments without a radical-left party: L(+ C) Seventy-five percent of the observations in the data set (counting each year from 1997 to 2017) could be classified within these four types of governments.Centrist governments constitute the largest government type outside these four groups, including both a Social-Democratic party and a Conservative or Christian-Democratic parties (16%).Additionally, the remaining 9% represent other government types: caretaker cabinets, "rainbow governments," and Center-Right-Green governments.The article does not analyze and discuss these latter types of governments.

| Methodological challenges with the comparative perspective
The data set has some methodological challenges that are important to specify.First, administrative traditions for political appointees in government differ across Europe (Dahlström, 2009), particularly at lower hierarchical levels under the respective ministers.The data set is limited to documenting the top level, thus it includes only political appointees who were members of the government cabinet.Additionally, the countries use different titles to address the top political level (ministry, department, cabinet offices, etc.) and political appointees (e.g., minister, secretary general, secretary of state).This study applies "ministry" and "minister" as common terms for these top political levels.
Second, the focus of national integration policies may differ, both cross-nationally and over time within the same country.In relation to the nexus between integration and immigration policies described in the theoretical section, which subfields "belong" to one or the other may differ cross-nationally.For example, Hernes (2018) shows that while family reunification was framed as an integration issue in Norway during the 2015 refugee crisis, in Sweden this policy area was explicitly restricted to deter further immigration.Detailed analyses of where different subfields in the heart of this nexus are placed-within the body responsible for integration or immigration-have not been conducted and compared cross-nationally.Thus, there are some differences regarding which subfields the countries delegate to the respective fields of immigration and integration.Because the data set largely builds on official primary sources and secondary sources from country experts, these sources' descriptions of which bodies are responsible for integration and immigration form the basis for the classification.
Another challenge is that the focus of the integration policies may differ cross-nationally and over time, depending on the circumstances in each country (e.g., the number of refugees from one year to another) and the political and public debates.However, irrespective of the exact content of the integration policies, the hypothesis tested in this article is at a generic level, hypothesizing which ministerial structure political parties would prefer based on their respective positions on integration policies as restrictive or permissive.These positions could be argued to be quite similar irrespective of the exact content of the integration issues; for example, a radical-right party is expected to advocate for restrictive policies both concerning refugee integration policies and concerning civic integration policies.Thus, although the content of integration policies may differ, such variation is not very problematic for the theoretical hypothesis tested, which focuses on whether issue and policy preferences affect the ministerial structure.

| INCREASINGLY DIVERGENT MINISTERIAL STRUCTURES
How has immigrant integration been coupled with other issues and signaled in the ministerial structure over time, and are there cross-national differences?Table 3 describes which type(s) of ministerial structure(s) the countries had from 1997 to 2017, and shows that 75% of the countries changed their type of structure at least once.Thirty-one percent (Italy, Greece, Finland, Germany, and Luxemburg) had two types of ministerial structures, another 44% (Austria, Denmark, France, Ireland, Norway, the Netherlands, and Sweden) tested three types.Several countries also moved back and forth between different categories, both horizontally, vertically, and diagonally.Thus, only four of 16 countries had the same stylized ministerial structure for the 21 years analyzed: Belgium, the United Kingdom, Spain, and Portugal.Belgium is a federal state and decentralized the responsibility for integration the entire period.The other three countries made minor changes in the organizational structure of the integration issue, but none that involved changes concerning if integration was signaled or not in the titles or concerning the coupling with or decoupling from immigration and/or justice.Thus, a stable structure is the exception rather than the rule.

| A diverse structure of coupled policy issues
Western European governments have coupled integration with numerous different issues.During the last 20 years, integration has been coupled with 41 policy fields when counting unique words included in ministry or ministers' titles, including asylum, public order, employment, social affairs, equality, culture, international cooperation, and democracy.Coupling integration with or decoupling it from justice and immigration is of particular interest for analyzing government parties' policy preferences.So, has integration been coupled with or decoupled from immigration and justice the last 20 years? Figure 1 describes the number of countries that coupled integration with immigration and/or justice in the same ministry.From 1997, about half the countries had these two areas coupled, and the number steadily rose until 2005-06, followed by a fluctuating and decreasing trend until 2017.

| Increased signaling of integration in the ministerial titles
Ten of 16 countries signaled integration in the ministerial titles during the period of analysis: five in the ministry title (Austria, Denmark, France, Luxemburg, and Sweden), and an additional five had a minister responsible for integration that was affiliated with a more generalized ministry (Germany, Ireland, Italy, Norway, and the Netherlands).
Figure 2 portrays a clear cross-national trend of increased signaling of integration in the ministerial titles the last 20 years.In 1997, only Sweden included integration in a minister's title; however, from 2007, around 40% signaled integration in the ministerial structure.This number dropped in between 2012 and 2014, but rose again after the refugee crisis in 2015.In the three countries that introduced "integration" in the ministerial titles after the refugee crisis (Norway, Sweden, and Ireland), all established a specialized minister within a more generalized ministry.

| Increasingly divergent cross-national structures
Figure 3 illustrates the distribution of observations when combining the two components of signaling and coupling into four stylized ministerial structures (see Table 1).
Figure 3 shows that ministerial structures where integration is signaled in the title and coupled with immigration and/or justice did not exist until 2001, but by 2005, nearly 25% had this structure.However, after remaining relatively stable until 2011, it decreased the following years until 2014, but regained popularity after the 2015 refugee crisis.Structures that signaled integration in the title and decoupled integration from immigration and/or justice were less common the first years of analysis and have fluctuated during the entire period; however, from 2007, about 20% of the countries have had this type of structure.Turning to the observations where integration was not signaled in the titles, about 45% started in 1997 with a structure where integration was not signaled and coupled with immigration and/or justice, and this number remained stable until 2006, and then followed a 10-15% decline.The observations where integration was not signaled in the title and decoupled from immigration and/or justice have been the most volatile.In the late 1990s, around 50% had this structure.The percentage steadily decreased to about 20% in 2007; however, from 2011, the trend shifted, and approximately 35% of the countries have had this structure in recent years.
Summarized, the ministerial structure of the integration issue in Western European countries has become increasingly diverse the last 10 years, because the countries spread themselves across the four stylized ministerial structures to a larger degree than they did in the first 10 years of analysis.

| PARTIES' ISSUE AND POLICY PREFERENCES: REFLECTED IN THE MINISTERIAL STRUCTURE?
The next question is if the government parties' issue and policy preferences help explain these general traits of change and divergence.Figure 2 shows that it was not common to include integration in the ministerial titles the first 10 years of analysis, that there was a steady growth of countries that signaled integration from 1997 to 2007.The rise of integration on the political agenda could be caused by several factors; increased saliency of the integration field, the introduction of integration policies at the EU level from 1999 with the Treaty of Amsterdam, and so forth.Because the objective of the analysis is to investigate if certain types of governments correlate with the different types of ministerial structures, it is necessary to focus on a period in which all four structures had become "normal" alternatives.Thus, the subsequent analysis narrows the period to 2007-2017.Each observation (N = 108) refers to the ministerial structure for a given year.The first hypothesis predicts that right-oriented governments will be more inclined to couple integration with immigration and/or justice than left-oriented governments will be, and arises from the assumption that parties with more restrictive positions on immigration and integration would prefer to couple these issues.
Table 4 shows that right-oriented governments more often coupled integration with immigration and/or justice than left-oriented governments did and the observed difference is significant (p > .01), 4 thus confirming the hypothesis. 5Denmark is a prime example of this pattern, where the coupling and decoupling of integration and immigration has shifted back and forth depending on the political orientation of the government.The results are further supported by an analysis of the particular cases driving the results.Approximately half of the governments did inherit the hypothesized structure from the previous government (a decoupled structure for left-oriented governments and a coupled structure for right-oriented governments).Governments that inherit the structure are naturally not expected to change this structure, and there is only one case where a government inherited the expected structure and changed it (in 2007, a Swedish right-oriented government moved integration out of the Ministry of Justice and established a Ministry of Integration and Equality).For governments that did not inherit the hypothesized structure from the previous government, however, six governments made changes in the hypothesized direction, including left-oriented governments that changed from a coupled to a decoupled structure (Norway in 2009, Denmark in 2011, and the Netherlands in 2012), and right-oriented governments that changed from a decoupled to a coupled structure (France 2007, Norway 2015, and the Netherlands 2010).This implies that only 36% of the governments that inherited a structure that contradicts the hypothesis chose to keep this structure, which corresponds well with the results presented in Table 4.

| Signaling and governments with radical parties
The theoretical section presents two competing hypotheses based on different logics for why governments that include a radical party will be more inclined or not to mention integration in the ministerial titles than governments without such parties will be.
Table 5 shows that for right-oriented governments, there is no clear pattern: The discrepancy between right-oriented governments with or without radical-right parties is minor and the difference is not significant (p = .08). 6 However, for left-oriented governments, Table 5 shows that left-oriented governments with a radical-left party are more inclined to signal integration than governments without such a party, and this difference is significant (p > .01).Nevertheless, when examining the specific cases that drive the results for left-oriented governments with a radical party, the support for the hypothesis is weakened.Governments with radical-left parties either (a) inherited the hypothesized structure from the previous government (Denmark and Luxembourg), or (b) made changes during their governing period as a response to the refugee crisis in 2015 (Sweden).Luxembourg has had a signaled structure since a center government introduced a Minister of Family Affairs and Integration already in 2004, thus the 2013 left-oriented government that included the Socialist Workers' Party and the Greens inherited this structure when they took office.Similarly, Denmark has had a signaled structure since a right-oriented government introduced "integration" in the Ministry title in 2001, and the left-oriented government including the Socialist People's Party that took office in 2011 merely continued this tradition.The last case driving the results is Sweden.The 2014 left-oriented government including the Greens, actually removed "integration" from the title when they took office in 2014, but reintroduced it as a response to the refugee crisis in 2016.Consequently, combining these insights, the qualitative analysis weakens the empirical support for the hypothesis expecting that governments with radical parties would be more inclined to signal integration, even for governments with radical-left parties.
It is important to emphasize that governments with radical parties constitute a limited number of observations, particularly those including radical-right parties, which brings high uncertainty to the results.Nevertheless, with this precaution in mind, the lack of evidence to support the hypothesis that radical parties would be more inclined to signal integration does not necessarily imply that the these parties do not have such preferences for signaling.One possible explanation for why preferences for signaling "integration" is not manifested in the ministerial structure, may be that radical parties often are junior partners in coalition governments, and that they merely lost the battle when the ministerial structure was negotiated.In an analysis of radical-right parties' direct influence on the restrictiveness of immigration and integration policies, Akkerman (2012) found that the inclusion of a radical party in a right-oriented government does not lead to policies that are more restrictive.He points to possible explanations for the radical-right parties' lack of influence, for example, that the bargaining position of radicalright parties in coalition governments has been relatively weak because they have mostly been junior partners, and because of their organizational weakness (Akkerman, 2012, pp. 512, 523).Additionally, the radical parties' radical position could have made other coalition parties hesitant to agree on a structure that would increase the attention to such a salient issue where especially the radical-right parties have such a strong issue ownership, providing them with a platform to broadcast their political views.
If the lack of empirical support for the hypothesis predicting that radical parties would be more inclined to signal integration is caused by the radical parties' position as junior partners, or that these parties merely prefer not to signal integration as stated in the alternate hypothesis, could be the subject of future in-depth analyses of each process for the respective countries.

| NEW INSIGHTS AND LIMITATIONS OF THE COMPARATIVE APPROACH
Demonstrating cross-national traits of how issue and policy preferences are manifested in the ministerial structure would not be possible without the study's comparative approach.The broader comparative approach applied in this analysis than has been applied in previous studies of machinery changes made it possible to focus on a new political context, where the majority of countries have coalition governments instead of single-party governments.Although the literature on machinery of government changes has identified several drivers of change, internal party politics has often been highlighted as one of the most important drivers-if not the most important-for ministerial changes (Davis et al., 1999;Heppell, 2011;White & Dunleavy, 2010).These studies mostly use empirical analyses of single-party governments; however, most Western European countries have multiparty systems, where coalition governments are the rule rather than the exception (in the DIIG data set, approximately 80% of the governments were coalition governments).The political dynamic of machinery changes could be expected to work differently for single-party and coalition governments.In single-party governments, the ministerial structure is ultimately the prime minister's prerogative, and ministries could be established, abolished, and altered for internal political considerations.For coalition government, this dynamic might be quite different because the allocation of ministries among the parties is subject to thorough negotiations.Although the party leaders of each coalition party can alter the personnel responsible for their party's ministries, it is less likely that internal party politics plays such a substantial role in altering the ministerial structure, because such changes need the acceptance of all coalition parties.Issue and policy preferences, however, might play a larger part for coalition governments, because parties could be expected to fight to ensure that the ministerial structure reflects their issue and policy preferences.
The broad comparative approach also challenges a new statement in the immigrant integration literature, namely that there is trend toward "mainstreaming" in immigrant integration governance in Western-Europa-implying that the responsibility for integration policy is distributed across various units, contrary to having one specialized unit for integration (Scholten et al., 2017).Figure 2, however, shows an increasing trend of signaling integration in the ministerial titles, implying a more specialized focus on integration in the horizontal governance structure.
Still, the comparative approach applied in this study has its limitations.Which stylized structure political parties in each country prefer for particular issues, and how decisive this factor is when determining the ministerial structure as a whole, cannot be revealed by quantitative analyses, but must ultimately be answered by in-depth analyses of each process leading to ministerial changes and stability.Additionally, by expanding the comparative focus, the analysis was limited to investigating a single issue.Immigrant integration was chosen as a "most likely" case for detecting government parties' issue and policy preferences, because of its characteristics as politically salient and wicked.Consequently, these characteristics affect how the main findings could be generalized to issues of less salience and wickedness.Although the analysis found that only a few countries had stable structures for the integration issue, stronger traits of stability could be expected for less salient and wicked issues, because government parties might be less inclined to use the ministerial structure-the top level of bureaucracy-as a way to signal and prioritize issue and policy preferences.
The methodological approach in the analysis, dividing signaling and coupling into four stylized organizational structures (see Table 1), proved useful.The novelty of the approach was not that the question of policy preferences and coupling had not been touched upon in the existing literature, but the advantage is that it enables a systematic analysis of how policy preferences might affect how issues are coupled or decoupled, and how policy preferences might ultimately affect the overall ministerial structure.Additionally, the generic operationalization of the organizational structure into four stylized categories makes it transferable not only to other policy fields (e.g., energy, environment, disabilities), but also vertically to other governmental levels (international, regional, and local) in analyses where the organizational structure is the dependent variable.

| CONCLUSION
That not only government parties' issue preferences but also policy preferences could affect the horizontal ministerial structure has been an often-stated, but yet unexplored, assumption in the literature on "machinery of government changes."This study investigated this assumption by analyzing whether and how issue and policy preferences might explain longitudinal and crossnational traits in the ministerial structure of immigrant integration in 16 Western European countries.
The descriptive analysis found that stable ministerial structures were the exception rather than the rule.Divergence and change characterize the ministerial structure of the integration issue, driven by an increasing trend of signaling integration in the ministerial titles and a fluctuating trend concerning the coupling of integration with immigration and/or justice.The first hypothesis gained empirical support: The analysis showed that right-oriented governments were more inclined to couple integration with immigration and/or justice than left-oriented governments were, reflecting the right's policy preferences for restrictive integration policies.However, governments that included a radical party were not more inclined to signal integration in the ministerial titles.While internal party politics has been highlighted in previous studies as one of the most important drivers for ministerial changes (Davis et al., 1999;Heppell, 2011;White & Dunleavy, 2010), the article argues that this mechanism might be less relevant for coalition governments (that constitute the majority of Western European governments), and that issue and policy preferences might play a larger role when coalition governments determine the ministerial structure.
Why does the organizational coupling and decoupling of issues matter?The ministerial structure is a product of the government parties' evaluation of which structure will be the best overall solution to achieve the government's (often conflicting) goals.Still, current literature has focused more on the political aspect rather than on the actual policy implications of these choices.Although the manifestation of an issue in the ministry or ministers' titles might be important both symbolically and politically, the implications of how an issue is coupled with other issues might be just as-or even more-influential for the policy outputs.As Vestlund (2015) concluded in her analysis of an EU pharmaceutical unit that shifted its affiliation from being coupled with "enterprise and industry" to "health and consumer": "the horizontal specialization systematically tips the scales in the direction of certain actors, solutions, interests and concerns in decision processes, eventually resulting in a change of policy focus."Still, the relative significance of the horizontal structure on policy outputs is debated in the public administration literature (Vestlund, 2015), and quantitative analyses with the horizontal structure as an independent variable are rare.Although this study concentrated on the horizontal structure as a dependent variable, the presented DIIG data set could also be applied to examine the horizontal structure as an independent variable in future studies.The literature on immigrant integration has produced several comprehensive data sets on immigration and integration policy developments, for example, data sets such as MIPEX (2015), CIVIX (Goodman, 2012), and IMPIC (Helbling, Bjerre, Römer, & Zobel, 2017).These data sets could be combined with the DIIG data set to examine if and how different stylized structures have influenced the content of national integration policies, providing new insight not only for the literature on immigrant integration, but also for the public administration literature more generally.
Last but not least, knowledge on the policy implications of organizational changes is essential, because these types of changes consume public resources.White and Dunleavy (2010, p. 7) demonstrate that ministerial changes have substantial costs, both financially and administratively, and that procedures of reorganizing the ministerial structure do little to minimize these costs.Thus, although it is the government's prerogative to design the ministerial structure, these costs make it legitimate to question the purpose of such changes.If the government uses the ministerial structure to solve internal party conflicts or to communicate the government's issue and policy preferences as part of a political battle with the opposition to win the electorate, constant changes that consume public resources might be more criticized than if the changes were made to support the administration and implementation of actual policies.
immigrants.European countries-and most scholars-often use the term "immigrant integration" policies (or just "integration policies"), while American scholars refer to either alienage, immigrant, integration, or incorporation policies (Filindra & Goodman, 2019).As this study applies the Dataset on Immigrant Integration Governance (DIIG), which documents the West-European ministerial structure, it uses the term "immigrant integration." 2 In some countries, certain ministries have set titles (e.g., the Home Office), and cannot include "integration" in the ministry title.This restricts the government parties' prerogative to choose ministerial titles freely.Nevertheless, it is a political choice to place a specific issue within a ministry subject to such restrictions.If the government wanted to signal "integration" in ministerial titles, it could place the integration field within a ministry that was not subject to such restrictions. 3The operationalization of government parties in the ParlGov data set does not include parties supporting minority governments.However, because the prerogative to alter the ministerial structure ultimately lies with the prime minister or leaders of the coalition parties in Western Europe, supporting parties are less relevant when the ministerial structure is determined. 4For t test, see online appendix: https://doi.org/10.18710/VH95QC.analyses controlling for immigration and asylum rates show the same main findings as the correlation analyses (Tables 4 and 5) and the t tests. 6For more information on the t test, see online appendix: https://doi.org/10.18710/VH95QC.
Four stylized horizontal structures Policy A coupled with policy B Policy A decoupled from policy B Policy A signaled Policy A in title and coupled with policy B Policy A in title and decoupled from policy B Policy A not signaled Policy A not in title and coupled with policy B Policy A not in title and decoupled from policy B

F
I G U R E 1 Number of countries that coupled integration with immigration and/or justice (N = 16).The countries that are not classified as coupled are either: (a) decoupled horizontally within different ministries, or (b) decoupled vertically between the federal and state levels (Austria before 2011 and Belgium)

F
I G U R E 3 Distribution of countries across four stylized ministerial structures (N = 16) 5.1 | The decoupled left and coupled right Types of governments and a coupled/decoupled ministerial structure, 2007-2017 Types of government and ministerial structures with and without integration signaled in the title,