Who’s fit for the job? Allocating ministerial portfolios to outsiders and experts

Abstract Why do parties appoint outsiders and experts to ministerial positions? Extant research offers explanations based on institutional arrangements and external shocks (e.g. political or economic crises). We go beyond such system-level variables to argue that the characteristics of ministerial appointees are a function of the portfolio they are being appointed to. Drawing on theories of political delegation, we argue that outsider and expert appointments to ministerial office are affected by a portfolio’s policy jurisdiction, its financial resources and appointment powers, and the partisan leanings of the ministerial bureaucracy. We test these arguments on all appointments of senior and junior ministers in Austria between 1945 and 2020. The analysis shows that outsiders are more likely to be appointed to ministries with greater party support in the bureaucracy, while experts are more likely appointed to portfolios dealing with high-salience issues.


Introduction
How are members of government selected? The extant literature locates the answer to this question on a single dimension between the poles of party government and technocratic government.
Under party government, ideologically distinct parties compete in elections, the winning parties place party members in government offices, and these partisan ministers then implement the party agenda (or, in coalition governments, some compromise between different party agendas). The popular mandate confers legitimacy on governments (Rose, 1974;Katz, 1986;Blondel and Cotta, 2000;Amorim Neto and Strøm, 2006;Mair, 2008).
By contrast, under technocratic government, expertise provides an alternative source of legitimacy. Government ministers are selected based on their subject-matter knowledge and enact 'rational' solutions, unconstrained by short-term popular approval or electoral considerations (Habermas, 1973;Majone, 1994;Centeno, 1993;Fischer, 2009;Caramani, 2017;Sánchez-Cuenca, 2017;Costa Pinto et al., 2018;Caramani, 2020). Conceptualizing party government and technocratic government as polar opposites on one underlying dimension (McDonnell and Valbruzzi, 2014;Cotta, 2018;Costa Pinto et al., 2018;Bertsou and Caramani 2020b), the empirical literature has so far either studied appointments of outsiders (people with no or weak party affiliation) or technocrats, lumping 'outsiderness' and expertise together. 1 We argue that, while correlated, 'outsiderness' and expertise should be conceived as two distinct properties of government ministers and that party leaders have different incentives to select outsiders and experts, depending on characteristics of the specific government post they need to fill.
Drawing on theories of political delegation, we present arguments about how portfolio characteristics shape who becomes a minister. More specifically, we argue that issue salience, a ministry's budgetary and patronage resources, as well as the ideological leanings of the ministerial bureaucracy should affect the probability of outsiders and experts being appointed. We test these expectations with data on post-war ministerial appointments in Austria. Our paper thus contributes novel perspectives on the process of ministerial selection with implications for governments' legitimacy and effectiveness in parliamentary democracies.

Conceptual premises
Why do parties appoint ministers with varying degrees of 'outsiderness' and expertise? A number of studies have examined this question, using system-level explanations such as economic crises (Wratil and Pastorella, 2018;Alexiadou and Gunaydin, 2019) or institutional features of the polity (e.g. semi-presidentialism) (Amorim Neto and Strøm, 2006;Schleiter and Morgan-Jones, 2009). Our paper provides a novel perspective by focusing not just on the number of outsiders and experts appointed, but on the specific role they are given. We therefore theorize how the characteristics of ministerial portfolios shape the appointment of party insiders vs. outsiders and experts vs. non-experts. More specifically, we examine the role of issue salience, financial resources, patronage powers, and bureaucratic support.
Before we delve into the theoretical discussion, three important premises need to be clarified. First, we assume in our discussion that outsiderness and expertise are two independent dimensions. While the literature on technocrats often combines these two dimensions and defines technocrats as experts without political or partisan affiliation (McDonnell and Valbruzzi, 2014), we argue that outsiders are not necessarily experts (Marsh et al., 2010;Street, 2012) and insiders are not necessarily non-experts (Lavezzolo et al., 2021).
Second, we conceptualize outsiderness in relation to a specific party, not to politics in general, which is in line with the delegational logic of party-based representation (Müller 2000a;Strøm, 2000). While most political insiders are also party insiders, there are cases of political insiders without (strong) attachment to any party. For example, senior bureaucrats or diplomats are highly involved and experienced in the practice of politics (and should therefore count as political insiders), but may be unaffiliated to a party. Another group to which these criteria may apply is party switchers: individuals who used to be party (and thus political) insiders, but whose transition to another party has turned them into party outsidersat least temporarily.
Third, we do not expect party leaders to have carte blanche when selecting government personnel, but to be constrained in various ways. Besides formal selection rules set out by the constitution (e.g. recruiting ministers from the legislature in Westminster systems or formal appointment powers of the head of state), party leaders need to account for the representativeness of their government team (Rose, 1974;Andeweg 2000b) in terms of gender (Goddard, 2019), region (Ennser-Jedenastik 2013b), party faction or tendency (Mershon, 2001;Ceron, 2014). Naturally, several other characteristics such as the candidates' age, their scandal-proneness, their policy preferences and their competence need to be considered as well in the selection process (Rose, 1974;Kam et al., 2010;Bäck et al., 2016).

Theoretical framework
We develop our argument from theories of political delegation. Delegation typically means that a principal asks an agent to perform a task that the principal cannot do himself or herself and that the agent may be better equipped to do. The appointment of ministers is a crucial step in the chain of delegation that runs from voters to the bureaucracy (Strøm, 2000). Formally, ministers thus act as agents of the cabinet, yet their more important principal is typically the cabinet party that appoints them (Müller 2000a). In addition, ministers themselves act as principals by delegating tasks to the ministerial bureaucracy whose expertise they require to achieve their policy goals (Huber, 2000;Huber and Shipan, 2006). In normative terms, both these steps in the delegation process are essential for parties to make the chain of delegation from voters to the governing apparatus work in parliamentary democracies (Müller 2000a).
Cabinet party # Minister # Bureaucracy All delegation potentially produces agency loss, as agents may deviate from the preferences of their principals (Bendor and Meirowitz, 2004). Parties typically anticipate the potential for agency loss and thus chose ministerial appointees in ways that mitigate delegation problems (Müller and Meyer, 2010). Our core theoretical contribution is to argue that both, the political damage from agency loss and the overall risk of agency loss to happen in the first place vary systematically with portfolio characteristics. Some portfolios are substantively or strategically more important to a party , which will increase the costs of agency loss. Some are more likely to cause delegation problems, either in general terms or for specific types of appointments. Portfolio characteristics should therefore influence the calculus by which parties appoint outsider and expert ministers.

Insiders vs. outsiders 2
Choosing insiders over outsiders for government office allows party leaders to minimize problems in the delegation from party to minister. For one, uncertainty with regard to relevant selection criteria (Andeweg 2000b;Müller 2000a;Strøm, 2000) is lower for insiders than for outsiders. In particular, ministerial candidates' previous party careers will be informative in terms of their competence in various fields, their loyalty to the party and their policy preferences. Secondly, insiders have had years of socialization within the party and have thus internalized the party's norms, values, and policy commitments, ensuring at least a basic level of ideological congruence with the party agenda (Van Haute and Carty, 2012).
In contrast, appointing outsiders to ministerial office is usually a riskier choice. All else equal, information on crucial characteristics of outsider candidates is not as easily accessible and outsider ministers are less socialized into party norms and less steeped in party ideology. Consequently, outsiders are more likely to pursue different issue priorities, to hold deviant policy positions, and should therefore act as less faithful agents of the party in government than partisan appointees would. The appointment of outsiders thus increases the risk of agency loss (Amorim Neto and Strøm, 2006;Schleiter and Morgan-Jones, 2009). While ministers in real-world politics may not act as the policy dictators that, for example, Laver and Shepsle's portfolio allocation model envisages (Laver and Shepsle, 1990;Laver and Shepsle, 1996), they still have a substantial impact on 2 Given the two dimensions of outsiderness and expertise, it would, in theory, be possible to build our argument with respect to four types of ministers: expert and non-expert insiders, expert and non-expert outsiders. Yet, this would require theorizing the effects of the independent variables on the prevalence of each of the four ministerial types and thus make our argument unwieldy. For the benefit of theoretical clarity, we thus put forward our hypotheses with respect to the two dimensions independently of each other. policy outcomes (Alexiadou, 2016;Bäck et al., 2022). Appointing outsiders will therefore produce outcomes further away from a party's ideal point (and that of its voters).
Despite this obvious disadvantage of outsider appointments for governing parties, they are an empirical fact and the literature has pointed to various reasons why party leaders recruit outsiders for ministerial office. For one, there are instances when party leaders view outsiders' distance to party policy as an assetespecially, when they seek to shift the party's position in a policy area (e.g. for electoral reasons or to send signals to investors and creditors in the context of economic crises) (Alexiadou and Gunaydin, 2019). Second, outsiders are, by definition, characterized by a weak or even non-existent power base within the party that appoints them, which allows party leaders to keep them on a shorter leash than insiders, who will often have intra-party networks and factional support. Finally, outsiders may be appointed because of their celebrity status or media savvy (Marsh et al., 2010;Street, 2012), thus creating potential electoral benefits.
Notwithstanding these (and potentially other) incentives to choose outsiders over insiders for ministerial office, we argue that party leaders will manage delegation problems associated with outsider appointmentsin terms of both delegation steps, from cabinets/parties to ministers and from ministers to the bureaucracyby selectively placing them in portfolios with specific characteristics.

Experts vs. non-experts
Unlike outsiderness, expertise is always a desirable trait in ministerial candidates. Expert ministers are likely to govern more effectively than non-experts (Alexiadou, 2016) and voters prefer to be governed by people who know what they are doing (Lavezzolo et al., 2021). Expertise among government personnel also functions as an additional source of political legitimacy, independent of the party label.
Contrary to the insider-outsider dimension, there is thus no trade-off along the expertise dimension. Rather, constraints on expert appointments arise from the limited availability of experts (with ministerial ambitions as well as sufficient compatibility with the party line) or the fact that party leaders may prioritize expertise lower than other requirements such as descriptive representation or maintaining an intra-party power balance. In addition, expertise in ministerial appointees becomes more desirable when bureaucratic expertise is less accessible (e.g. due to the partisan leanings of bureaucrats).
Parts of the literature suggest that the objective demand for experts in ministerial office has increased over the course of the 20th centuryin terms of government effectiveness and legitimacy. These studies argue that effective governance requires more and more technical knowledge of systemic characteristics and dynamics, as well as legal frameworks due to the expansion of state regulation into various sectors of economic and social life and the increasing interdependence of economic systems (Meynaud, 1964;Fischer, 2009;Bertsou and Caramani 2020b). Likewise, as declining trust in and identification with political parties threatens to undermine party-based representation in parliamentary democracies (Müller 2000b;Brunclk and Parzek, 2019), ministers' individual expertise might serve to compensate the legitimacy deficit of political parties (Habermas, 1973;Costa Pinto et al., 2018;Caramani, 2020).
While appointing experts (from within or from outside the party organization) will thus undoubtedly be beneficial for party leaders in most instances, we argue that expert appointments, too, are governed by portfolio-specific considerations of agency loss in the delegation from cabinets/parties to ministers and from ministers to the bureaucracy.

Damage from agency loss
We now turn to generating specific expectations about how portfolio characteristics relate to outsider and expert appointments. As outlined earlier, we expect the potential damage from agency loss in a ministerial portfolio to affect party leaders' choices in ministerial selection. The more valuable a portfolio is to the party, the greater these potential costs, and thus the less willing party leaders should be to take the risk of outsider appointments. As expertise, per se, is not associated with a higher propensity to cause delegation problems, expert appointments should not be affected by damage control considerations. These expectations pertain to the delegation step from cabinets/parties to ministers.

Issue salience
For one, the salience of policy issues within a portfolio's jurisdiction should affect how party leaders perceive the value of a ministerial portfolio and thus the potential costs associated with agency loss in that ministry. Specifically, the damage from agency loss should be smaller in portfolios that fall outside of a party's core issue commitments . Most real-world parties vary considerably in how much attention they devote to certain issues (Budge and Farlie, 1983;Green-Pedersen, 2007) and how competent they are viewed by voters on these issues (Walgrave et al., 2009;Lefevere et al., 2015).
For governing parties, failure to deliver on their core issues is likely to be electorally harmful (Schumacher et al., 2013). What is more, constituencies inside the party (e.g. trade unionists in social democratic parties, farmers' organizations in agrarian or Christian democratic parties, or business representatives in conservative and liberal parties) (Allern et al., 2021) will react particularly negatively to malperformance on core issues, and might eventually punish party leaders by withholding their support in subsequent electoral campaigns or in party leadership elections. As party leaders will seek to minimize electoral and intra-party damage, we expect outsider appointments to be less common in portfolios that cover a party's core issues.
Hypothesis 1 Outsider appointments are less common in portfolios that deal with issues of high salience for the appointing party.

Resources
In addition to their policy value for specific parties, ministerial portfolios differ in terms of their access to financial resources (Bojar, 2019). While correlated to some extent with governing parties' policy prioritiesas more money will be budgeted for more important ministriesdifferent government activities also objectively require different sets of resources. For instance, a comparatively modest budget will often suffice to cover the state's tasks in the field of cultural affairs. In contrast, public expenditures on welfare state programs or education cover a sizeable chunk of most industrialized countries' economies (Garritzmann and Seng, 2016;Savage, 2019). Naturally, these resources are instrumental for governing parties to act on their policy goals (e.g. to implement certain redistributive policies), but they also have the intrinsic value of allowing parties to engage in clientelistic behavior (Batista, 2017). Governing parties may pursue the latter in various ways, ranging from typical pork barrel politics to handing out government contracts or licenses to private businesses that are close to the party (Müller, 2000). Access to financial resources is thus crucial in determining a portfolio's value for political parties. As ministers have substantial discretion with regard to how money is spent within their jurisdiction, the impact of agency loss for the party in terms of its policy goals and patronage distribution is greater in portfolios with more extensive access to these resources. Hence, we hypothesize that outsiders are less likely appointed to ministries with larger budgetary resources.
Hypothesis 2 Outsider appointments are less common in portfolios with larger budgetary resources.

Appointment powers
Ministers do not only make policy and spend money, they also make appointments. Most obviously, many civil servants are appointed by ministers (or their subordinates), and these appointment powers are among the most important tools that ministers have to influence the implementation of policy in their jurisdictions (Huber, 2000;Huber and Shipan, 2006). But appointment powers go far beyond the core civil service to include, among others, regulatory agencies (Gilardi, 2005), policy and advisory commissions (Hesstvedt and Christiansen, 2022), and the boards of state-owned enterprises (Ennser-Jedenastik, 2014). While such appointments can be instrumental in that they may advance a party's or minister's policy goals (as the old adage goes: 'personnel is policy', see Lewis, 2008, 27), they also have an intrinsic value as private (patronage) goods. They can thus serve a 'control' as well as a 'reward' purpose (Kopecký et al., 2012). Ministerial portfolios differ widely in how much appointment powers they provide. This is largely a function of the number of staff that is directly or indirectly appointed by the minister, but it also depends on the established rules and norms regulating the (non-)politicization of bureaucratic appointments. To be sure, the most valuable positions in terms of controlling policy are top-level administrators (Page and Wright, 1999).
No matter whether appointments are viewed as instrumentally ('control') or intrinsically ('reward') relevant, the value of a ministerial portfolio is clearly affected by its appointment powers. Greater appointment powers may mean greater policy influence and/or a larger pool of jobs to hand out to party loyalists (Meyer-Sahling, 2008). In both cases, agency loss becomes more damaging to the party, as the party either suffers in terms of policy implementation or patronage goods (or both). Therefore, we assume that appointment powers are negatively correlated with outsider appointments.
Hypothesis 3 Outsider appointments are less common in portfolios with greater appointment powers.

Risk of agency loss
Notwithstanding how parties manage varying levels of potential damage from agency loss, portfolio characteristics should also affect how party leaders evaluate the risk associated with expert appointments in a particular ministry (delegation from cabinet/parties to ministers) and a portfolio's propensity to cause problems in the delegation from the minister to the bureaucracy. We expect such portfolio-specific risk evaluations to condition party leaders' choices in terms of outsider and expert appointments.

Issue salience
Besides the implications of issue salience for outsider appointments outlined earlier, the relative importance of a ministry's policy jurisdiction for a governing party will determine the risk associated with expert appointments. Specifically, expertise is more likely to be present in or available to a party on issues of high salience. Parties have stronger links to interest groups that represent core voter constituencies and central issue demands (Otjes and Rasmussen, 2017;Allern et al., 2021). Also, parties' issue attention is linked to their ties with interest groups (Otjes and Green-Pedersen, 2021). The supply of politically compatible experts will therefore be greater in policy areas that are central to the party. As parties will be able to recruit from a pool of experts that are politically and ideologically aligned, expert appointments will involve a lower risk of agency loss in portfolios that cover core policy issues, all else equal.
Hypothesis 4 Expert appointments are more common in portfolios that cover issues of high salience for the appointing party.

Bureaucratic support
As argued above, ministers depend on bureaucrats to be effective (Huber, 2000). In particular, ministers facing a politically hostile ministerial bureaucracy run a higher risk of having their policy agenda sabotaged than ministers sharing common goals with their civil service. This phenomenon has been termed the 'ally principle': politicians are more likely to delegate tasks and grant discretion to bureaucratic agents who share their preferences (Bendor and Meirowitz, 2004;Huber and Shipan, 2006). After all, policy outcomes are assumed to be a function of politicians' and bureaucrats' ideal points. Party leaders thus have incentives to act strategically and take bureaucratic preferences into account when recruiting individuals for ministerial office (Bertelli and Feldmann, 2007). In particular, they will seek to contain the risk of agency loss by appointing partisan insiders to ministries with a more hostile bureaucracy, whereas ministries with a high proportion of co-partisan bureaucrats may see more outsider appointments.
Hypothesis 5 Outsider appointments are more common in portfolios with high party support in the bureaucracy.
Likewise, we expect bureaucratic support in a ministry to affect expert appointments. After all, the central purpose of delegating tasks to the bureaucracy is to tap into its expert knowledge (Huber, 2000). Loyal bureaucrats will happily assist ministers with their know-how and will have little reason to exploit their informational advantage over the minister to obstruct party policy (Huber and Shipan, 2006). Ministers facing a hostile bureaucracy, however, cannot expect the same level of cooperation, which will constrain their access to the bureaucracy's expertise (Müller, 2007). In the latter scenario, expertise on part of the minister should be an important asset for parties to counteract potential sabotage by the ministerial bureaucracy. Party leaders will thus have stronger incentives to appoint experts to portfolios where their ministers face a potentially non-cooperative civil service than in ministries where bureaucrats share the party's policy goals. Following this line of reasoning, we expect expert appointments to be more likely in portfolios with low bureaucratic support for the nominating party. Table 1 provides a summary of the hypotheses, expected effects of the independent variables and the respective theoretical reasoning.
Hypothesis 6 Expert appointments are more common in portfolios with low party support in the bureaucracy.

Case selection and data
To answer our research question we study the appointments of Austrian post-war ministers and junior ministers . Austria is a typical party democracy, resembling many other parliamentary democracies regarding key characteristics of the political system. In terms of outsider appointments, it is an unlikely case compared to other countries, given Austrian parties' organizational strength and traditionally strong roots in society (Müller and Philipp, 1987;Andeweg 2000a;Scarrow et al., 2017). Findings on outsider appointments should thus be likely to generalize to other European parliamentary democracies. Notwithstanding change over time due to the state of modernization and integration in supranational organizations, we expect the overall demand for expertise in ministerial office to be similar across contemporary democracies. However, Austria should be a critical case in terms of the relationship between outsiderness and expertise. Given political parties' dominant role in society and stateand their linkages to corporatist interest organizations in particular (Ennser-Jedenastik, 2017) -Austrian parties should have larger recruitment pools for experts within the party (that is, insiders who are also experts) than governing parties in other countries. As a consequence, the correlation between outsiderness and expertise should be modest compared to other countries, where expertise may be more exclusively confined to outsider appointees. The Austrian case therefore provides a particularly suitable empirical basis to disentangle the drivers of outsider and expert appointments. We analyze original data on the career trajectories of cabinet members, collected from three online sources: the official website of the Austrian Parliament, Wikipedia, and the Munzinger biographical data base.
In line with most of the extant literature, ministers and junior ministers without elected public office or party affiliation prior to their appointment are classified as outsiders (McDonnell and Valbruzzi, 2014;Cotta, 2018;Bertsou and Caramani 2020a). Furthermore, we also code as outsiders those ministers who held public office for or were affiliated to any party other than the nominating party before their appointment (e.g. party switchers).
We then define occupational areas where expertise can be acquired for each ministerial portfolio and code ministers as experts when their area of expertise matches the portfolio they are appointed to. For instanceamongst other thingsa prior position in the central bank qualifies a minister as an expert finance minister, the education portfolio requires a career in the educational system, lawyers and judges are experts for the justice portfolio, diplomats are considered foreign policy experts and a career in business (e.g. management positions), business associations or trade unions qualifies a minister as an expert for the economic affairs and employment portfolios. We also coded individuals with jobs in the state bureaucracy that relate to their ministry, with sub-national executive offices matching their portfolio's policy jurisdictions, with leading functions in related parliamentary committees and with academic careers in corresponding fields as expert ministers. 3 Using these detailed portfolio-specific measures for expertise, we seek to extend the scope of previous empirical studies on the topic (McDonnell and Valbruzzi, 2014;Alexiadou, 2016;Costa Pinto et al., 2018;Bertsou and Caramani 2020b) and to anticipate the biases towards economic training and the natural sciences that exist in the extant literature (Bertsou and Caramani 2020a).
To test Hypothesis 1 and Hypothesis 4 we operationalize parties' core issues based on issue salience in election manifestos. Specifically, we build on the Austrian National Election Study's coding of electoral manifestos (Müller et al., 2012;Dolezal et al., 2016), assigning issue-areas to every portfolio and recording the proportion dedicated to these areas in each manifesto Ecker et al., 2015). In cases where two or more policy portfolios are combined in one ministry, we use the sum of the salience scores of all portfolios bundled in a ministry to accurately reflect the post's value to the party. We utilize individual ministries' shares of the overall budget (across all portfolios within a cabinet) to measure portfolios' budgetary resources (Hypothesis 2). Given that there are only minor variations across the annual budgets of individual ministries within the same cabinet, we use the financial resources budgeted for each ministry in the second year of the government's tenure. This information was extracted from annual reports (Bundesrechnungsabschluss) published by the Austrian Court of Audit, which are available for the years 1961-2021. 4 Due to skewness of the issue salience and budgetary resources measures, we use the logged versions of both variables in our regression models. To operationalize ministers' appointment powers (Hypothesis 3) we use two indicators. First, we count the number of departments (Sektionen) 5 per ministry as listed in the Amtskalender, the official registry of state institutions and public administration. 6 Second, we code a dummy variable that equals one for the few ministries that control the bulk of appointments to state-owned enterprises: finance, economic 3 For additional information on our expertise measure, please consult the online appendix, Table A8. 4 These reports can easily be obtained from the Austrian parliament's website: https://www.parlament.gv.at/PAKT/RGES/. 5 Ministries are divided into departments (Sektionen), subdepartments (Gruppen), divisions (Abteilungen), and subdivisions (Referate) (Liegl and Müller, 1999). affairs, state industries, and infrastructure (Ennser-Jedenastik 2013a). Parties' support in the bureaucracies of each ministry (Hypothesis 5, Hypothesis 6) is measured based on the results of staff representatives' elections which are available for the years 1967-2020. In these elections, all public employees at the federal level elect their representatives from party lists.
In addition to the main explanatory variables, we include various controls in our statistical models. To account for residual differences in the 'objective' value of ministerial portfolios (e.g. stemming from the prestige associated with specific portfolios), we use portfolio salience ratings by Druckman and Warwick (2005). 7 Moreover, we include a dummy variable for ministries that are the traditional domains of the Austrian social partnership: social affairs, economic affairs, health, and agriculture. Due to the close linkages between corporatist interest organizations (unions, chambers) and the two (erstwhile) major parties, SPÖ and ÖVP (Ennser-Jedenastik, 2017), there is a large pool of partisans with high levels of expertise in these domains. Hence, we expect these ministries to have fewer outsider and more expert appointments.
As the party leader of a governing partywho typically also receives the most valuable government post available to the partyplays a crucial role in nominating ministerial candidates (Andeweg 2000b;Strøm et al., 2010), we account for party leaders' career trajectories in the statistical models. We expect outsider appointments to be more likely when the party leader did not take the traditional party route to office as such leaders should have smaller personal networks within the party organization than traditional party leaders would. We therefore include a dummy variable indicating whether the party leader was a member of the party executive before appointment to government office.
Anticipating likely differences in ministerial selection dynamics between first-round appointments (usually following elections and coalition negotiations) and appointments in cabinet reshuffles, we incorporate a dummy for cabinet reshuffles in our regression models. We also use control variables for appointments to two types of junior minister posts to account for different job-requirements in terms of insiderness and expertise that the roles senior minister, co-partisan junior minister and watchdog junior minister might entail. For instance, while insiders should be a better fit for the watchdog role (e.g. monitoring a coalition partner in the best interest of the party) than outsiders, experts might be a good choice for junior ministers supporting a co-partisan senior minister.
Finally, we include fixed effects for decades and parties in all statistical models to account for variation due to party-specific factors and time trends.

Analysis
We first present some descriptive information. Figure 1 displays the percentage of four ministerial types over time: non-expert insiders, expert insiders, expert outsiders, and non-expert outsiders. Outsider appointments were relatively common in the immediate post-war period (around 20%), as new elites that were untarnished by the preceding twelve years of authoritarian rule were sought after. Once the political system had stabilized, outsider appointments became a more marginal phenomenon. Only from around 2000 has there been a significant increase to, again, one in five appointments in the recent decade.
Expert appointments (which include insiders) are more common in general. They experience a first peak during the social democratic single-party cabinets of the 1970s, then again during the grand coalitions of the 1990s, and reach an all-time high in the 2010s. This increase holds when 7 In cases where two or more portfolios are bundled in one ministry, we use the salience rating of the most important portfolio. This is because some relatively unimportant portfolioswhich are frequently bundled with portfolios of higher valueare not included in the Druckman and Warwick (2005) expert survey. Using the sum of the salience ratings for bundled portfolios would thus introduce bias due to missing values for such 'secondary' portfolios. Note, however, that maximum portfolio salience and total portfolio salience are closely correlated (r=0.79). excluding the non-partisan cabinet Bierlein that was installed following the 2019 'Ibiza affair' and the subsequent vote of no confidence in the first cabinet of Sebastian Kurz.
Looking across portfolios 8 there is a significant gradient between those that do not typically feature expert ministers (Interior, Finance, Defense) and those that do (Justice, Social Affairs, Agriculture). Given the significant prosecutorial powers of the minister of justice (he or she can give instructions to all public prosecutors in individual cases), this position has often been filled with non-partisans (and thus outsiders). Outsiders were also common in the finance ministry during the early post-war period.
To illustrate the relationship between outsider and expert status, Table 2 shows a crosstabulation of the two variables. There is a moderately strong and statistically significant  . 8 We present numbers for those nine portfolios that remain relatively stable over longer periods of time and thus provide at least N > 30 in terms of appointments. correlation between the two (Cramèr's V is 0.34, p < 0:001). This largely reflects the fact that there are hardly any non-expert outsiders 9 : While 26% of expert appointments are also outsider appointments, the same is true for only 3% of non-expert appointments. Overall, the largest group by far are non-expert insiders who make up more than half of all appointments.
Next, we present binary logistic regressions with outsider and expert status as the dependent variables (Table 3). 10 To account for heterogeneity between and dependence within governments, we specify cabinet-level random effects and cabinet-clustered standard errors. Models I and III use all available data points, whereas Models II and IV exclude appointments to junior minister positions (JMs) from the analysis.
Note that the variable 'party support in bureaucracy' is only available from 1967 (when elections for employee representatives were introduced). In addition, the variable has some missing data points in later years, for example, when a ministry was newly created (often through splits and mergers). In such cases, the variable only has values after the next round of elections which take place every four (up until 1999) or five (after 1999) years.
Let us first examine the results pertaining to outsider appointments. In Models I and II, we find no statistically significant effects for issue salience and thus no support for Hypothesis 1. Outsiders are no more likely to be appointed to portfolios dealing with high-or low-salience issues. Likewise, results are inconclusive with regard to our expectation that ministries controlling greater budgets see fewer outsider appointments Hypothesis 2. While coefficients are negative in Models I and II, as expected, they do not reach conventional levels of statistical significance. 11 We also find no significant effects for appointment power, at least in the full model. Neither the number of departments nor the indicator for portfolios controlling the bulk of potential patronage appointments yield significant coefficients in Model I. Only when excluding junior ministers (who have no appointment powers) do we find a positive effect of the number of departments (counter to Hypothesis 3). While patronage powers sustain a negative impact on outsider There are idiosyncratic reasons for the twelve appointments of non-expert outsiders in our data-set that go beyond our theoretical model. For instance, Austria's first post-war cabinet included a member of the resistance movement 'O5' who was neither a partisan at the time nor a policy expert. 10 As an alternative, we present multinomial logit models predicting three ministerial types (non-expert insiders, expert insiders, and expert outsiders) in Table A3 in the online appendix. The substantive conclusions are identical to the ones presented here. 11 Note that the issue salience and budget share variables correlate (r=0.51). Still, eliminating either variable from the estimation does not alter the result for the other one. appointment (in line with Hypothesis 3) in Model II, the effect is insignificant. Overall, there is thus only limited evidence in support of Hypothesis 3.
The strongest finding in Models I and II pertains to the effect of party support in the bureaucracy. Here we find a statistically significant positive effect that is in line with Hypothesis 5: As predicted by the ally principle, greater party support among bureaucrats is associated with more outsider appointments. What is more, Figure 2 shows that this effect is by no means small. Across the empirical range of the party support variable, the predicted probability of an outsider appointment increases from below five to around 33%. This result holds when limiting the sample to experts or first appointments (thus excluding re-appointments of ministers in subsequent cabinets) or when examining only SPÖ or only ÖVP ministers (who, together, constitute the vast majority of appointments). Our analysis thus yields strong support for the expectation that the partisan leanings of the bureaucracy are correlated with the probability of outsiders being chosen as ministers.
Next, we turn to expert appointments. Model III yields strong evidence that experts are more likely to be appointed to high-salience portfolios as measured by manifesto data, thus supporting the argument that such appointments are less risky in salient portfolios. This result holds when limiting the sample to senior ministers (Model IV). As the graph in Figure 2 shows, the probability of an expert appointment increases from around 40% to around 60% when moving from a low- salience issue (2% of the manifesto) to a high-salience one (a third of the manifesto). Parties are thus more likely to put experts into portfolios that they deem more electorally important, as predicted by Hypothesis 4. Finally, there is no effect of party support among ministerial bureaucrats on the probability of experts being appointed to a portfolio and thus no support for Hypothesis 6. Hence, while nominating parties use loyal bureaucracies to monitor outsider ministers (as well as insider ministers to deal with bureaucracies of a different political leaning) (Hypothesis 5), they do not employ expert ministers to minimize the potential for sabotage posed by a hostile body of bureaucrats. Among control variables we find substantial effects of the social partnership portfolio variable on outsider and expert appointments, as expected. Likewise, outsider appointments are less likely in cabinet reshuffles (Models I and II) and under party leaders who worked their way to government through the party machine (Model II). Unexpectedly, we also find pronounced negative effects of the junior minister post variables (co-partisan and watchdog) on expert appointments in Model III, while outsider appointments are not affected.

Discussion and conclusion
This paper contributes novel perspectives to the study of ministerial appointments. Conceptualizing outsiderness and expertise as independent dimensions, the paper identifies distinct drivers of outsider and expert appointments based on theories of political delegation. By demonstrating how portfolio characteristics condition appointments of different types of ministers this study also extends the scope of the existing literature, which largely focuses on country and cabinet-level explanations.
The empirical analysis supports the theoretical argument that nominating parties anticipate the potential for agency loss in individual portfolios and select ministerial candidates accordingly. Outsiders are more likely appointed to portfolios where a friendly ministerial bureaucracy might contain deviations from the party agenda. Likewise, parties minimize delegation problems by appointing experts to portfolios where they have access to larger pools of ideologically compatible experts, while refraining from expert appointments elsewhere. Only one of three expectations derived from that line of reasoning, namely that expert ministers are more likely appointed to portfolios with a high risk of sabotage by the ministerial bureaucracy, is not supported by the statistical models.
In contrast, a portfolio's value to the party in terms of policy, budgetary resources and appointment powers does not affect outsider appointmentsat least not in a statistically significant way. Parties are thus no more willing to make risky appointments to portfolios where the stakes are comparatively low than to ministries where agency loss comes at a higher cost. Overall, our findings therefore indicate that, while parties minimize the risk of agency loss when making ministerial appointments, they avoid taking chances even for portfolios where the potential damage from agency loss is relatively small.
Due to specific characteristics of the case, studying Austria provides the opportunity to disentangle drivers of outsider and expert appointments, which are independent in theory, yet correlated empirically. This correlation is asymmetric, though: Most outsiders in our sample are experts, yet most experts are not outsiders. Thus, the appointment of ministers who cannot claim legitimacy from a party mandate needs to be justified on the basis of their portfolio-specific qualification. While results on both variablesoutsiderness and expertiseshould generalize reasonably well to other parliamentary democracies with multiparty systems, the high level of politicization in the Austrian ministerial bureaucracy and the social partnership model may constrain their external validity. Comparative research will therefore be required to test the validity of the findings in other contexts.