What’s in a name? The politics of name changes inside bureaucracy

In this article, we examine the effects of political change on name changes of units within central government ministries. We expect that changes regarding the policy position of a government will cause changes in the names of ministerial units. To this end we formulate hypotheses combining the politics of structural choice and theories of portfolio allocation to examine the effects of political changes at the cabi-net level on the names of intra-ministerial units. We constructed a dataset containing more than 17,000 observations on name changes of ministerial units between 1980 and 2013 from the central governments of Germany, the Netherlands, and France. We regress a series of generalized estimating equations (GEE) with population averaging models for binary outcomes. Finding variations across the three political-bureaucratic systems, we overall report positive effects of governmental change and ideological positions on name changes within ministries.

politics of structural choice and theories of portfolio allocation to examine the effects of political changes at the cabinet level on the names of intra-ministerial units.We constructed a dataset containing more than 17,000 observations on name changes of ministerial units between 1980 and 2013 from the central governments of Germany, the Netherlands, and France.We regress a series of generalized estimating equations (GEE) with population averaging models for binary outcomes.Finding variations across the three political-bureaucratic systems, we overall report positive effects of governmental change and ideological positions on name changes within ministries.

| INTRODUCTION
The design of public bureaucracies is far from neutral.The structure of bureaucracies is the result of political decisions and compromises over bureaucratic control, political accountability, and policy choices (Bertelli & Sinclair, 2018;Daniels, 1995;Hammond & Thomas, 1989;Hogwood & Peters, 1983;Lewis, 2003;Moe, 1995).
While the creation and termination of bureaucratic agencies may reflect radical policy changes, more intermediate but substantive changes find their expression in more subtle structural changes: the change of names of bureaucratic agencies and of the units within them.The names of such (sub)divisions are expressions of "enhanced political attention" by the political leadership to certain important policy issues (Mortensen & Green-Pedersen, 2014).Names and name changes reflect the (changing) policy priorities of an incoming government.The restructuring and naming of ministerial (sub)divisions can hence be seen as an important part of a government's agenda (Hammond, 1986(Hammond, , 1993)).So, when in 2010 in the Netherlands the incoming government changed the name of the "Ministry of Justice" into "Ministry of Security and Justice," the new cabinet signaled to citizens that it will take increasing threats of terrorism more serious (Hendrickx & Stoker, 2019).The inclusion of the German word "Heimat" in the new name of the Interior Ministry was criticized by certain political groups that this could be conceived of as a "dog whistle" to the far-right (Knight, 2018).Similarly, in 2007, newly elected president Sarkozy created a much debated "Ministry of Immigration, Integration, National Identity and Co-Development" as a signal sent to its right-leaning voters.
Despite the importance of name changes, studies of name changes have been underrepresented until now by students of bureaucratic politics.Current and recent studies predominantly focus on purely structural changes such as mergers, "demergers," splits, creations, and terminations of (semi-)independent agencies (Greasley & Hanretty, 2016;Lewis, 2003;Rolland & Roness, 2012;Yesilkagit, 2020) or of ministerial departments (Davis et al., 1999;Hogwood, 1997;Hood & Dunsire, 1981;Pollitt, 1984;White & Dunleavy, 2010).In these studies, bureaucratic agencies are "literally manufactured" by political actors in power "as a vehicle for advancing and protecting their own interests" (Moe, 1995, p. 145).While there exist studies that recognize the close relationship between public policy and bureaucratic structure (cf.Daniels, 1997;DeLeon, 1978;Hogwood & Peters, 1983), bureaucratic structure is primarily considered to be a function of the preferences of politicians regarding agency control, accountability, and the distribution of tasks and functions within a bureaucracy.Underestimated is the fact that bureaucratic reorganizations may be the means with which politicians signal or enact policy changes (Tosun, 2018), a function that was identified early on by organization theorist Meyer (1979) suggesting that "change of structure is a means through which organizations communicate policy and strategy both to their own members and to others." The connection between structure and policy is, by contrast, present in studies of coalition formation and portfolio allocation (Bäck et al., 2011;Bucur, 2018;Laver & Shepsle, 1996;Sieberer et al., 2019).In the early portfolio allocation theories, the distribution of ministries was considered to reflect the size of legislative seats shares, that is, the so called Gamson's law.With the rise of policy saliency theories, however, the allocation of ministerial portfolios came be to be understood as a function of parties' policy preferences.Policy saliency theories stress that political parties aiming at "specific themes corresponding to the policy remit of specific cabinet portfolios are more likely to receive those portfolios" (Bäck et al., 2011).Mortensen and Green-Pedersen (2014) have found that "changes in the ministerial structures are driven by changes in political attention in combination with the issue preferences of the government" (Mortensen & Green-Pedersen, 2014).According to Sieberer et al. (2019) "changes in portfolio design follow a political logic that is driven by preference alterations among the relevant political actors" (Sieberer et al., 2019).
The purpose of this study is to examine the relationship between political change and ministerial name changes within central government.We argue that name changes are an integral part of the politics of bureaucratic structure.
The aim of this study is to show that a study of bureaucratic name changes emphasizes the relationship between changing political preferences and the reflection of those substantive policy changes within the organizational schemes of government organizations.Starting from existing studies on bureaucratic design and portfolio allocation, we develop a number of expectations linking political turnover and name changes within the central government departments in three European states.To this end we constructed a unique dataset of 17,076 observations on structural changes, including name changes, within German, Dutch, and French ministerial departments between 1980 and 2013.We find that name changes are highly related with political turnover and structural features of bureaucracy.
The article is structured as follows.In Section 2, we discuss the merits and limitations of purely structural approaches to bureaucratic design.In Section 3, we develop and present our hypotheses.Section 4 describes the political systems of our chosen countries and the data and methods.In Section 5, the research design is explained, followed by the presentation of the results of our analysis in Section 6.The article ends with a discussion and conclusion.

| THE POLITICS OF STRUCTURAL CHOICE: MERITS AND LIMITATIONS
Studies on the politics of structural choice show that bureaucratic structure is the outcome of political decisionmaking processes (Horn, 1995;Lewis, 2003;Moe, 1995).Founded on the premise of the "rationality of structure" (Moe, 1984), the argument is that bureaucratic structure reflects political choices regarding the control of bureaucratic behavior, accountability, autonomy, and decision-making (Bawn, 1995;Bertelli, 2006;Elgie, 2006;McCubbins et al., 1989).Through structures, politicians (principals) "manufacture" efficient incentive structures within organizations, adjust monitoring costs and information asymmetries to satisfy their needs.Through particular vertical or horizontal recasting of organizational units and jurisdictions, structure may affect to a large extent the policy outcomes of bureaucratic decision-making (Hammond, 1986(Hammond, , 1993;;Hammond & Thomas, 1989;Moe, 1995).
Numerous studies have shown the existence of relationships between political change and bureaucratic structure (Carpenter & Lewis, 2004;Elgie & McMenamin, 2005;Epstein & O'Halloran, 1999;Gilardi, 2005;Huber & Shipan, 2002;Lewis, 2003) A general finding in these studies is that political turnover, that is, the replacement of an incumbent government by a government with substantially different policy preferences, causes changes in the structure of bureaucracy.In the United States, for example, structural change is almost imminent when the presidency changes from Democrats to Republicans, or vice versa.In parliamentary systems, the politics of bureaucratic structure takes a different shape as the constitutional relationship between the branches of government potentially cause less antagonism as in a separation of powers system (Moe & Caldwell, 1994).There, among the main concerns of political principals is the quality of the accountability relationship between government and bureaucracy (Bertelli, 2006;Bertelli & Sinclair, 2018) and the credibility of agencies with a regulatory function (Gilardi, 2002) that guide structural choice, albeit that institutional legacies may trump these motives (Yesilkagit & Christensen, 2010).
Within studies of politics of structural choice, however, there has been little attention to substantive, policydriven motives for structural change.The bulk of the literature assumes that the sole motivation of politicians to reorganize the bureaucracy once they come to power comes from their preferences for control, credible commitment, or accountability.However, politics is also about launching policy issues and changing existing policy programs.
"Structural choices have all sorts of important consequences for the content and direction of policy," writes Moe (1995, p. 127), and "because this is so, choices about structure are implicitly choices about policy."The policy termination literature, too, conceives laws, budgets, personnel, and organizations as different but inseparable aspects of public policy (Daniels, 1995;DeLeon, 1978;Hogwood & Peters, 1983).Although it is recognized that structural change is an indicator of policy change, the policy considerations behind structural change have, as Moe stated, been considered implicitly.
There is an emergent literature that explores the relationship between political change, issue dynamics, and bureaucratic structure (Hong & Park, 2019;Mortensen & Green-Pedersen, 2014;Sieberer et al., 2019).Focusing on the names of ministries Mortensen and Green-Pedersen (2014) found that "[G]overnments will organize the ministerial structure to fit the issues receiving the most political attention" (Mortensen & Green-Pedersen, 2014, p. 182).
They found a clear link between the policy concerns of politicians and the structure of central government.The act of "creating, eliminating, or restructuring ministries" is not only "a deliberate signal to the public of what designated policy area(s) is now given serious attention" (Hong & Park, 2019), but also determines the salience of policy issues that ministries under the incumbent government are engaged with.Similar conclusions are also drawn by Sieberer et al. (2019).They found that portfolio design and departmental structures experience change more frequently when a new coalition or a new prime minister takes office.In sum, these studies show that the likelihood of structural changes within central government does not only increase by administrative concerns but is also driven by the substantive policy preferences of incoming governments.

| EXPLAINING NAME CHANGES INSIDE BUREAUCRACY
The purpose of this article is to extend the theory of the politics of structural choice.We argue that the interests of politicians regarding the structure and organization of government are not driven mainly by preferences regarding political control over the bureaucracy or credible commitment concerns, as the standard theory of the politics of structural choice argues, but that they are also interested in the structure of public organizations to consolidate their substantive policy preferences, that is, by translating their agenda into an organizational form.The observable implication of our argument is that the names of organizational units inside the central government bureaucracy will change when a new political majority has formed a new cabinet and taken power.We developed three hypotheses to test the expectation that political majorities are policy-driven.
The first hypothesis tests the expectation that when the ideological position of a new cabinet differs substantially from its predecessor, the likelihood increases that the new ideological position will cause the names of administrative units to change too.Changing the names of bureaucratic agencies or ministries is hence perhaps one of "the most central aspect[s]" of politics of structural choice "as this is what the public is most likely to notice" (Mortensen & Green-Pedersen, 2014).In terms of Hannan and Freeman "[N]ame changes and publicity generated by the organization often signal important life events such as endings" (Hannan & Freeman, 1989).Bureaucratic reorganizations can ostentatiously serve the symbolic and political ends of reorganizations (March & Olsen, 1989;Stull et al., 1988).By deciding for new names, politicians can signal to the electorate and interest groups that they are sensitive to their cause and responsive toward their needs (Canes-Wrone, 2009;Hong & Park, 2019;Lichtmannegger, 2018;Tosun, 2018).Above else, changing names is a relatively cheap way to signal to the wider public that politicians "do something" about a problem.When a problem is too complex to solve in the short term, a name change may give the impression that politicians "supposedly address a pressing public concern and worry about the details of bureaucratic structure if [they] return to office" (Canes-Wrone, 2009, p. 33).The main lesson from these studies is that name changes can be considered as an incoming government's announcement of intentions to change the policy status quo, even when the change is meant to be symbolic.We may therefore expect that whenever a change of government takes place, the incoming government will change the names of the bureaucratic structures over which they gain control.This effect is present for agencies as well as for units within ministerial organizations, as they are crucial for governments in interacting with other actors such as the legislative, organized interests and stakeholders as well as other supranational bodies.
Hypothesis 1.A change in government increases the likelihood of name changes in ministries.
Another observable implication of our argument concerns the location where the name change takes place.Given the first-mover capability of organizational units at the lower levels of an organization for policy formulation, as Hammond (1986) argues, it can be expected that lower-level units are exposed to a higher risk of a name change than units at higher echelons of the organization.Hence, our expectation states that when a government's ideological position changes not just changes of organizational names will occur, but the changes will occur at a lower organizational level.Location is an important parameter of the structure of bureaucracy (Epstein & O'Halloran, 1999;Hammond, 1986).Location affects where which policy proposals are formulated, who decides over which policy alternatives, and how the flow of communication within the organization is structured.In hierarchies, the further away from the organizational apex a unit is located the more specialized the unit becomes and less politicized the process of policymaking is (Page, 1985).It is within units located at the lowest echelons where policy proposals are initiated and then transported upwards.Lower and middle levels bureaucrats are responsible for the crafting of the bulk of agreements between their political principals and the interest groups in the environment of their ministerial department.At the higher echelons near to the political principals, decisions become politicized as ministers and their top civil servants have to decide which of the proposed policy alternatives are going to be prioritized (Aberbach et al., 1981).In other words, multiple policy alternatives are simultaneously produced at the lower echelons, exported upwards through subsequent higher-level echelons for final decision-making at the top of the organization.This leads us to expect that name changes will occur more likely with units in the lower echelons of the organization than at echelons higher in the organization.This leads us to our second hypothesis: Hypothesis 2. The lower the administrative unit in the hierarchy of the organization, the more likely a unit will experience a name change when there is a change in government.
The final implication of our argument is that name changes will occur more likely with organizational units that are mandated with policymaking, that is, line units, rather than units tasked with specific advisory or support activities positioned outside the vertical line hierarchy and in direct relation to the cabinet minister, that is, staff units.In other words, we expect that units with staff positions are less likely to experience a name change when a new government enters office than line units.Staff units vary and house a diverse range of ministerial advisers, including personal assistants, media or policy advisers, consultants, legal advisers, and financial officers.Staff can consist of individual special advisers to the minister or organized ad hoc units (Shaw & Eichbaum, 2018).Since administrative traditions in various countries are different, the size and importance of ministerial staff units will be different.Of the countries in this study and during the period under study, for example, Dutch ministers have relatively modest staff organizations.In the Netherlands, individual cabinet ministers' dealings with their respective parties is organized at the level of the party leader in cabinet (the Prime Minister in case of the largest party in cabinet), the party leader in parliament, and the chairperson of the party (Andeweg & Irwin, 2009).By contrast, in France the political executive, including the Prime minister and its ministers, has been used to extensively create, in addition to ministerial cabinets and in the context of a semi-presidential regime, ad hoc staff units that functions as political institutions and counterbalances the power of the permanent civil servants in ministerial line units (Suleiman, 1974).In Germany, the earlier smaller number of staff units, mostly consisting of the ministers' private office and the liaison unit to parliament and cabinet, have developed into rather large structural leadership elements hosting several staff units (Hustedt, 2013).What they have in common is that staff units are located outside of the ministry's bureaucratic line hierarchy.The line units are located within the ministerial hierarchy and are typically organized into divisions, subdivisions, and bureaus.They become more specialized and focused on the lower levels of the organization as they address the more specific aspects of the larger policy portfolio of the ministry.In comparison with the staff units, line units are part of the policy machinery of the ministry.Therefore, we expect that it will be names of line units that are more likely to experience change than the names of staff units once the ideological orientation of the incoming government substantially differs from its predecessor.Hypothesis 3. Name changes are more likely to occur with line units within ministries than with staff units.

| NAME CHANGES IN GERMAN, DUTCH, AND FRENCH MINISTRIES
To study the effects of political change on bureaucratic name changes, we selected three countries (i.e., Germany, the Netherlands, and France) with different types of political systems (i.e., a federal, unitary, and semi-presidential system) but with similar types of central administrative systems.In all three countries, the central government is organized in ministerial departments that coincide with politically crafted portfolios.Decisions regarding the number of ministerial departments and the names of ministries are taken during the cabinet formation process.
Germany is a federal state with a bicameral parliamentary system.German governments are mostly based on minimum winning coalitions.After a federal election, parties engage in exploratory negotiation talks (Sondierungsgespräche) and subsequent coalition negotiations (Koalitionsverhandlungen) resulting in a coalition agreement (Bräuninger et al., 2019).The cabinet formation processes combine bargains over policies in the governmental program with negotiations over offices and the selection of cabinet ministers.Ministers are formally nominated by the chancellor and appointed by the federal president.Thus, neither a formal parliamentary vote of confidence to the government nor to individual cabinet ministers is necessary.
The Netherlands is a unitary state with a multiparty parliamentary system.The government is based on minimal winning coalitions.Cabinet formation can take a long time and concerns negotiations, first, on the content of the coalition agreement, and second, on the allocation of portfolios and appointment of ministers.Ministers execute the coalition agreement and organize their ministries such that the program within their jurisdiction can be implemented.
Once a government has been established, its survival depends formally upon the confidence of parliamentary majority, but in practice there has developed a monistic system whereby the government is dominant vis-à-vis parliament (Andeweg & Irwin, 2009).
Finally, France has a mixed system.The constitution of the Fifth Republic (1958present) combines a popularly elected president with extensive constitutional powers with a more or less conventional parliamentary system with a popularly elected legislature to which a prime minister and the members of his cabinet are responsible to (Guinaudeau & Persico, 2021;Thiébault, 2000).New governments are formed following general elections on the initiative of the president and bargaining duration is very short.The president nominates as prime minister the leader of the party that can reckon on the support of the parliamentary majority.After his nomination, the prime minister selects the members of his cabinet on line with pre-electoral alliances between parties more than actual electoral results but the president's role in cabinet formation remains important (Guinaudeau & Persico, 2021).At the end, the prime minister presents the list to the president for approval.Coalition formation then precedes the election of the prime and is quite straightforward.
In the three countries, central ministerial departments have a similar hierarchically layered and horizontally divided pyramidical structure.In this study we focus on name changes at the two top levels of the countries' central ministries.We refer to these levels as À1 and À2.These levels represent the two echelons directly below the top civil servant (permanent secretary) of a ministry.At level À1 we find the directorate-generals and at level À2 the directorates.Given the overall number of levels (up to three), we thereby analyze the top-level units formally closest to the political leadership as well as the middle-level units connecting this level with the lowest level.Table 1 presents the frequencies of name changes that have occurred within ministerial departments between 1980 and 2013.
Since name changes were not the only type of transitions that ministerial units can experience, the table presents name changes as a percentage of the total number of structural changes. 1   We find name changes to account for about one-sixth of all structural changes in this period.Compared across the countries, we find that name changes constitute only 11% of all structural changes within the French bureaucracy, but almost a quarter in the German and Dutch bureaucracies.Table 1 does lend furthermore some substance for our expectation that name changes will more likely occur at levels lower in the ministerial hierarchy.Comparing T A B L E 1 Share of name changes as share of structural changes per hierarchical level within ministries in Germany, the Netherlands, and France (1980-2013) the occurrence of names in terms of percentages we find that with 63% in Germany, 58% in the Netherlands, and 82% in France À2 level units experienced relatively more name changes than units closer to the political and administrative top.
Table 2 compares the occurrence of name changes in line and staff units across the three countries.It shows marked differences between the countries.Whereas in the German case line units have experienced relatively more name changes than staff units, in the Netherlands we find the reverse in relative terms.In France, line and staff units face equal likelihood of experiencing a name change.Hence our first impression on the basis of these descriptives is that we cannot assume for sure that line units have a higher likelihood of experiencing a name change than staff units.These differences are likely to be explained by the existence of distinct national traditions and legacies in the setup and use of staff units.They range from a longitudinal trend toward expanding leadership departments in Germany, ad hoc establishments of temporary nature in the Netherlands, and many policy units with a staff affiliation in France with consultative or advisory role.This national difference may explain our descriptive empirical pattern across countries.
A visual inspection of Figure 1 reveals there is no clear pattern of name changes throughout the term of a cabinet.Given the fact that ministers have broad, almost unchecked discretionary powers to (re)organize their ministerial departments and given the fact that changes at this level do not require parliamentary approval, we initially expected that name changes would follow suit after a new government assumed office.On average, the timing of the occurrence of name changes is dispersed throughout the cabinet terms.Unfortunately, our current dataset cannot tell whether the name changes that occur in year two or later during a cabinet term-for example, year 3 (1997) in Kohl IV, years 4 (1998) and 2 (2001) in Kok I and II, respectivelyare a government's reactions to scandals or crisis of some sort.
T A B L E 2 Share of name changes as share of structural changes in line and staff units within ministries in Germany, the Netherlands, and France (1980-2013)

| RESEARCH DESIGN
We regress a series of generalized estimating equations (GEE) with population averaging models for binary outcomes (Hilbe, 2009). 3We applied these models on a panel dataset where each ministerial unit at the levels À1 and À2 appears each year until it experiences a name change, is terminated, or leaves the dataset on 1 January 2013. 4As many units change names after only 1 year, we regressed our models with independent correlation structures instead of the usual autocorrelational or Markov structures for clustered longitudinal data.A comparison of the available correlation structures with the quasi-likelihood information criterion supports our choice for an independence model, that is, a model in which no dependence is assumed between each individual observation within a cluster.
Loss of accuracy is limited as independence models prove to produce consistent estimators even when the data are in fact correlated (Hilbe, 2009, p. 448).
The dependent variable of this study is binary.We count a name of a ministerial unit as changed if one or more nouns in the name of a unit has changed in comparison to the name of that unit a year before.Hence, a name change marks a change by which the unit is maintained, and only words within its name are dropped, added or altered.
Moreover, we excluded any name changes that occurred in the aftermath of a change at their superior level, for example, if a ministry's renaming or a ministry merger results in a larger number of name changes of subordinate ministerial units.With the data at hand, we were not able to distinguish between symbolic and substantive name changes.Whether name changes are merely symbolic or substantive is difficult to infer from the name change itself.
To infer that a name change is more substantive, one ought to investigate changes in personnel and budgets as well.Our first independent variable is political turnover.Political turnover indicates the degree of substantive ideological change that takes place when an incumbent government is replaced by a new one.It is measured as the absolute distance between the mean ideological positions of two subsequent governments.The mean left-right position of each individual cabinet is calculated as the mean of the ideological position of all political parties' manifesto programs that are represented in the government, taken from the Comparative Manifesto Project (Carroll et al., 2020). 5Our second independent variable is the level or echelon at which the unit is located.To determine the level of a unit we take the position of the permanent secretary as reference point and code until two levels below that position.In a typical ministry at level À1, we find (two to four) directorates general and at level À2 (two to four) directorates.Our third independent variable is a dummy variable that distinguishes line units from staff units.We coded as line units, units that reside under a hierarchical line of a "directoraat-generaal" (the Netherlands), "directorate generale" (France), and "Abteilung" (Germany) and were tasked with policy mandates.Staff units are units outside the vertical hierarchy, under the direct authority of ministers, chiefs of directorates or secretariat of the minister, entrusted with various heterogeneous tasks, from ad hoc advisory activities, communication, and media to support functions like legal services, finance or personnel.
In the Online appendix to this article, an organizational chart is provided that distinguishes between staff and line units.
The main source of our coding of the line and staff units are the civil service yearbooks.
We employ several control variables.As the likelihood of names being changed increases with the longevity of a cabinet, we control for (the natural log of) cabinet duration (in days).The gross domestic growth rate at the beginning of each cabinet period is an indicator of the percentage growth or decline of the economy.The effect can assume both directions as both economic growth and decline can urge incoming government to implement policy changes through name changes.We included this measure to control for possible significant economic effects, if any.The cabinet duration data as well as the gross domestic product (GDP) data are taken from the European Representative Democracy Dataset (Andersson et al., 2014).The variable age controls for the number of years that a unit exists unaltered with the same name.Organizational theorists consider high organizational age as a proxy for organizational resilience, in particular the strength of organizational leaders, staff or culture to resist external interventions or as the relative success of organizational leaders to secure support and resources for organizational survival (Boin & Christensen, 2008;Stinchcombe, 1965).Age will be interacted with political turnover and line unit to assess the effects of age on the effects of both explanatory variables.We also include two political control variables.
Assuming that the likelihood of a name change decreases with increasing levels of policy conflicts among coalition parties, we control for the level of polarization within cabinet (Tsebelis, 2002).We measure polarization as the absolute distance between the ideological positions of the two ideologically most outer parties in the government.Likewise, expecting that an increasing number of parties in cabinet deem agreement on policy change less likely, we control for the number of political parties that make up the coalition. 6 Finally, to control for the overall lower likelihood of name changes in France, we include intercepts for Germany and The Netherlands (dummies) with France as the reference country.Also, with France accounting for the majority of all name changes in the dataset, we may still expect that the results for the explanatory variables will be driven by the French cases.In order to alleviate this problem, we run models per country and interact the explanatory variables in the full models with all countries.The descriptive statistics for the dependent and independent variables are offered in Table 3.

| RESULTS
Table 4 presents the results of the tests of the first and third hypotheses.We first regressed the covariates of political turnover and line units separately (models 1 and 2) and combined them in two additional models with and without controls for age of line units (models 3 and 4).The results confirm our first hypotheses but reject the third.
Whereas political turnover increases the likelihood of a name change, the difference between line and staff units does not turn out as important in the politics of name change.The table shows that the longer a unit exists, the less likely it will experience a name change.This finding nevertheless fits nicely with the relationship between old age and survival found in other studies.For political turnover we find that for each standard deviation increase in absolute ideological distance the incoming government differs from the predecessor the odds of a name change increases with 12%.The variable maintains its significance and direction in the other models, which lends confidence to our finding.In terms of our data, this means that for the likelihood of a ministerial unit to experience a name change, the ideological mean of the incoming cabinet change needs to shift with 14 "Manifesto" points in comparison with the outgoing cabinet.Given the fact that the total range of the political turnover variable is 45 points, our finding implies Plot of predicted probability of name change for political turnover (with 95% confidence interval) The effects of political turnover on name changes at levels À1 and À2  To test the effects of political turnover on the likelihood of name changes at different echelons within the ministry, Table 5 presents the results of our regressions for the two hierarchical levels separately.The results confirm our expectation that units lower down the hierarchy have a higher probability of experiencing a name change than units higher up in the hierarchy.Whereas in terms of significance levels ministerial units at the level directly below the political leadership display a weak sensitivity to political changes, ministerial units two levels below and thus inbetween the highest and the lowest level inside the ministry have substantially higher odds of experiencing a name change.To have a more meaningful presentation of the effects of the hierarchical level of the units, Figure 3 draws the predicted probability plot with confidence intervals for level À2.

| DISCUSSION AND CONCLUSION
This study has shown, first, that politicians have a keen interest in the manner in which their substantive policy interests are reflected in the structure of bureaucracy.We found a consistent and significant relationship between changing ideological positions of government coalitions and changes in the names of ministerial units.This finding attributes to the existing studies of the politics of structural choice the insight that politicians are not only interested in matters of political control and accountability, as is with the traditional principal-agent theories of bureaucratic structure.Political control of the bureaucracy surely remains a major point of attention to politicians.We believe that our findings will be replicated in political systems other than parliamentary ones, as organizational names have generic capacities to shape and design organizational agendas (Hammond, 1986(Hammond, , 1993)).However, we underscore with our study, more research must start from the policy-driven motives of politicians instead of control and accountability-related motives.Bureaucratic structure is an important carrier of substantive policy ideas.Politicians aim for power not only for the sake of controlling bureaucrats.Ultimately, the control and steering of bureaucrats serves the goal of launching and implementing the policies that governments prefer.With its focus on names of ministerial units this study aims to the policy saliency underlying the structure of bureaucracy.
The second finding of this study is that politicians are interested in location-related aspects of bureaucratic structure.As changes at specific locations within the bureaucracy can influence the flow of information and the process of decision-making, politicians will have a close eye on the internal structure of the bureaucratic organizations F I G U R E 3 Plot of predicted probability of name change for political turnover for level À2 units (with 95% confidence interval) they assume power over.Our findings suggest that changes to the structure of bureaucracy occur not randomly but are targeted at those echelons and units where politicians can expect the highest degree of leverage in terms of influencing the outcomes of internal policy processes.Unit names structure the agenda of the ministry.The subordination of names in the hierarchy of the organization distinguishes between main and subtopics of which an issue consists.Names of units lower down the hierarchy designate locations where policy specialists are integrated around a specific issue.Juxtaposing issues in separate ministerial units is the manner how bureaucratic organizations can distinguish between different detached, complementary, or even competing issues.To change information flows in order to influence the generation of policy alternatives within an organization, policy reformers target units located the lower within the hierarchy.
An important role needs to be played by top civil servants within the ministry.The average minister is hardly interested in the internal organization of her ministry.In parliamentary systems, ministers become appointed after the leaders of the coalition parties have agreed upon a program and have decided upon the allocation of portfolios (Andeweg, 2000).Ministers are on average merely agents of the political leadership of the coalition.The most important decisions on the structure of the government are related to the designation of portfolios and the resulting number and names of ministerial departments.These decisions belong to the prerogative of the coalition parties' leaders.
And these decisions can result into the merger or splitting off of existing ministries or the creation of entirely new ones.Individual ministerial units such as directorates general and subdirectorates can experience reshuffles from one ministry to another.The fitting of these units and the crafting of a new ministerial department also address the realm of top civil servants.In our study, we have found a strong positive relationship between political changes and name changes, but it should be kept in mind that the changing of unit names so as to fit the new structure is mostly executed by the top civil servants.This means that in order to better understand the mechanisms of name changes inside ministries, we need additional studies of the political-administrative management of organizational changes.
This study has taken its cues also from portfolio allocation studies.Students of cabinet formation and portfolio allocation have been the first to study the relationship between political change and policy change within central government.Studies of portfolio allocation, delegation of policy tasks to ministerial departments and the appointment of ministers share common ground with studies on the politics of structural choice and bureaucratic design.
We urge for more studies that examine the relationship between politics and administration in parliamentary systems.The study of ministries has been beyond the scope of students of structural choice in parliamentary systems.
Like their US counterparts, students of bureaucracy in parliamentary systems have employed the theory of the politics of structural choice mostly on (semi-)independent agencies.Central government in general and ministries in particular have been the realm of students of portfolio allocation.We hope to stimulate the further integration of various aspects of bureaucratic organization in the study of the institutional design within parliamentary systems.

F
I G U R E 1 Distribution of name changes per cabinet period in Germany, the Netherlands, and France (1980-2013) A substantive name change would, expectedly, be accompanied by a change in personnel and budget of the unit experiencing change.Budgetary cuts or increases, or the appointment of personnel are indicators necessary to judge the weight of the name change.However, these data are hard to come by.
that a relatively substantive shift to the right or the left is required for a name change to occur.Figure2presents the predicted probability plot for the standardized value of political turnover for each standard deviation change holding other variables constant.
T A B L E 3 Descriptive statistics Binomial logit odds ratio coefficients (standardized scores)