• Open Access

The Politics of Patronage and Coalition: How Parties Allocate Managerial Positions in State-Owned Enterprises



While commonly regarded as a democratic pathology, party patronage can also be understood as an inherent feature of party government and thus as a linkage mechanism between political parties and the government executive. Therefore, theories of government formation, portfolio allocation and coalition governance can potentially add analytical leverage to the study of party patronage. Starting from this presumption, this article derives a number of hypotheses from the field of coalition theory and tests them on an original data set of over 2,000 appointments made to managerial boards in 92 Austrian state-owned enterprises between 1995 and 2010. The empirical analysis strongly supports the hypotheses, showing that patronage appointments vary with the partisan composition of government, the allocation of portfolios and junior ministers, as well as the importance of corporations.

In a most general way, party patronage can be understood as ‘the use of public resources in particularistic and direct exchanges between clients and party politicians’ (Müller, 2006, p. 189). It is hence typically viewed as an immoral, if not outright illegal, activity. Most early accounts of patronage have therefore been generated under the assumption that patronage is a democratic pathology, characteristic of a malfunctioning political system (Bearfield, 2009, p. 67; Boissevain, 1966; Eschenburg, 1961; Sorauf, 1956), which is why studies of patronage are often focused on developing countries (Kopecký, 2011). Due to the perception of patronage as an evil, theorists of party politics have been reluctant to recognise that it constitutes an inherent feature of party government. However, as Jean Blondel (2002, pp. 240–5) argues, party patronage is one of several forms of linkage between the government and the parties that support it (for a similar argument with regard to clientelism, see Kitschelt, 2000; Kitschelt and Wilkinson, 2007). As such, it represents one link in the chain of delegation between voters, parties and the state apparatus (Müller, 2000b, pp. 311–2), even though this link is typically not regarded as strengthening the process of democratic representation or accountability (see, however, Flinders and Matthews, 2010). Yet the notion of patronage as a linkage mechanism between parties and government opens up new theoretical perspectives on party patronage. Specifically, it leads to the conclusion that a better understanding of party patronage can be developed once it is conceived of as a manifestation of party government.

All conceptualisations of party government crucially revolve around the relationship between political parties and the executive (Katz, 1986; Mair, 2008; Rose, 1974). One of the sub-fields of political science most comprehensively concerned with this relationship is coalition research. It examines the processes by which parties form governments, distribute ministerial offices, bargain over policy output and manage the termination of cabinets (Laver and Shepsle, 1990; Strøm et al., 2008). Even more importantly for the present purpose, coalition research has over the past decades developed a rich arsenal of theories and concepts for the study of how parties enter and act in government. If one regards patronage as a linkage mechanism between political parties and the executive, then it becomes clear that coalition theory can add analytical leverage to the study of party patronage. The purpose of this article is therefore to add to our understanding of party patronage by applying concepts derived from theories of government formation, office distribution and coalition governance to the analysis of top-level appointments in Austrian state-owned enterprises between 1995 and 2010. Coalition theory thus becomes an analytical tool to enhance our knowledge of party patronage.

The article proceeds as follows. First, the theoretical section discusses the concept of party patronage and relates it to coalition theory, thus generating systematic expectations about patterns of patronage in the empirical data. The next section outlines the case selection, explains the process of data collection and presents an overview of the variables for the analysis. The statistical analysis submits the hypotheses outlined in the theoretical section to a multivariate test. After a discussion of the results, the final section concludes.

Party Patronage and Coalition Theory

As cited above, patronage can be conceptualised as an exchange relationship in which a variety of goods may be traded between the patron and the client: a party politician may offer expertise, legislation, access to the bureaucratic apparatus, public subsidies, housing or jobs in exchange for (alleged) electoral support, labour, campaign contributions, party membership or information available to the client.

Yet the empirical focus of this article is on a very specific patron–client relationship, namely the appointment of party loyalists to managerial boards in state-owned corporations.1 This perspective is very much compatible with the definition offered by Petr Kopecký and Gerardo Scherlis (2008, p. 356), who understand ‘party patronage as the power of a party to appoint people to positions in public and semi-public life’ (see also Kopecký et al., 2008, p. 4; 2012). They thus delineate the concept of patronage from related concepts such as clientelism (the exchange of material goods in return for votes), pork-barrel politics (the distribution of funds and legislation to territorial units in return for electoral support) and corruption (making public decisions in exchange for private gain). While in practice these concepts overlap to some degree, the narrower definition of patronage fits the focus of this article and thus serves as a theoretical starting point.

However broad, all conceptualisations of patronage revolve around four elements: the characteristics of the two actors involved (patron and client) and the nature of the two goods to be exchanged (from patron to client and vice versa); the patrons in the present study are political parties, and the clients are party loyalists. The goods that parties as patrons award their clients are positions on managerial boards in state-owned enterprises. Yet what parties receive from their clients in return is, in essence, contingent on the purpose of the respective patronage appointment. Previous studies broadly distinguish between two motivations for patronage: reward and control (Kopecký et al., 2012). The former implies that parties hand out jobs and appointments to fellow partisans in return for their loyalty,2 whereas the latter suggests that parties intend to exert influence over some area of public policy.

At a most general level, this distinction provides a first link between patronage research and coalition theory. The reward–control dichotomy is clearly related to the notions of office seeking and policy seeking that drive theories of rational party behaviour (Müller and Strøm, 1999; Strøm, 1990) which, in turn, inform the bulk of theoretical concepts in coalition research. Yet as party patronage, according to the above definition, involves the distribution of appointments, it is even more apt to equate reward and control with the concepts of intrinsic and instrumental office seeking (Strøm and Müller, 1999, pp. 8–11). Parties that are instrumental office seekers will use patronage as a means to an end, most typically as a mechanism to exercise control over public policy, whereas intrinsically office-oriented actors will perceive patronage appointments as an end in themselves. Party leaders may also use appointments to buy support from party activists, thus mitigating the risk of intra-party rebellion in the face of electoral defeat or unpopular decisions made by the government.

Empirically, the most obvious intersection between research on coalition governments and patronage is thus the study of office pay-offs, even if this line of research has largely been limited to analysing the distribution of ministerial portfolios. There are only a few exceptions that have extended scholarly scrutiny to junior ministers (Lipsmeyer and Pierce, 2011; Manow and Zorn, 2004; Thies, 2001; Verzichelli, 2008) or members of parliamentary boards (Carroll and Cox, 2012). This is in stark contrast to the vast range of other office-related pay-offs that coalition parties have at their disposal. Michael Laver and Norman Schofield (1990, pp. 42–3) provide an idea of the scope of office perks that are potentially at stake:

Our search might extend to the judiciary and to the civil service, senior appointments to both of which typically require at least formal executive approval. Probably the most important sphere of patronage, however, can be found in the parastatal agencies, such as the many nationalized industry boards; water, electricity, and other service authorities; health boards; development authorities, and so on. … This, to the best of our knowledge, is a largely unresearched area.

In supplementing the study of party patronage with a theoretical perspective drawn from coalition theory, this article aims to address this research gap. Despite the lack of such studies on non-ministerial office pay-offs, the underlying assumption here is that coalition research provides the analysis of party patronage with a firm theoretical footing. Specifically, it is theories and concepts of quantitative and qualitative portfolio allocation, portfolio salience, ministerial government and coalition governance that inform the development of hypotheses for this study.

The single most important finding in the literature on office pay-offs is the empirical regularity termed ‘Gamson's Law’. As William Gamson (1961, p. 376) famously conjectured, ‘[a]ny participant will expect others to demand from a coalition a share of the payoff proportional to the amount of resources which they contribute to a coalition’. The most powerful resource that parties in parliamentary systems have is, of course, their share of legislative seats (Browne and Franklin, 1973, p. 457). Along these lines, the proportionality proposition has been tested time and again, and it has been found that parties in coalition governments distribute portfolios in almost perfect proportion to their parliamentary seat shares (Browne and Franklin, 1973; Browne and Frendreis, 1980; Schofield and Laver, 1985; Warwick and Druckman, 2006). Intuitive as this finding may seem, it has sparked a lively debate about the gap between the predictions derived from bargaining power theory (Banzhaf, 1965; Shapley and Shubik, 1954; Von Neumann and Morgenstern, 1953) and the regularities found in the real-world data (Carroll and Cox, 2007; Schofield and Laver, 1985; Verzichelli, 2008, pp. 240–2). A recent effort by Stephen Ansolabehere et al. (2005; see also Snyder et al., 2005) to reconcile bargaining power models with empirical data has not only been criticised on theoretical grounds but also shown to rely on operationalisations that are empirically problematic (Laver et al., 2011).3 The evidence thus clearly supports the notion of a proportionality norm guiding the quantitative allocation of ministerial portfolios in coalition governments. The first hypothesis is therefore a simple translation of Gamson's proportionality proposition to the realm of top-level positions in public sector corporations:

  • H1. The partisan composition of managerial boards reflects the partisan composition of government in a proportional manner.

The second hypothesis concerns the effect that individual ministers have on the composition of managerial boards within their jurisdiction. The theoretical foundation for this argument is drawn from the literature on ministerial government (Laver and Shepsle, 1994, p. 8), most prominently elaborated in the portfolio allocation approach to government formation. According to Michael Laver and Kenneth Shepsle's (1990; 1996) concept of ministerial discretion, Cabinet ministers are policy dictators within their jurisdictions and thus have considerable leeway in shaping the policy output of their departments (for an empirical application of this approach to local government coalitions, see Laver et al., 1998). This argument can easily be translated from the policy domain to the realm of appointments, which then creates the expectation that the partisan composition of managerial boards is biased in favour of the party whose minister is directly responsible for the respective corporation.

  • H2. The partisan composition of managerial boards reflects the partisan affiliation of the minister under whose jurisdiction a corporation falls.

One of the more recent developments in coalition theory is the focus on coalition governance (Kim and Loewenberg, 2005; Müller and Meyer, 2010; Strøm et al., 2010; Timmermans, 2006; Timmermans and Moury, 2006). This line of research examines the means that parties employ to safeguard the policy bargain struck during the process of government formation. Applying a principal–agent perspective, it theorises that, as the divergence of policy preferences among coalition parties increases, the danger of incurring agency loss in the delegation from parties to individual ministers rises. However, politicians have a range of ex ante and ex post control mechanisms at their disposal that can be employed in order to minimise agency loss.

One widely used monitoring device is the appointment of ‘watchdog’ junior ministers (Lipsmeyer and Pierce, 2011; Manow and Zorn, 2004; Thies, 2001) for the purpose of containing the discretion of senior ministers. With respect to party patronage, this would mean that the freedom of individual ministers to hand out positions on managerial boards is constrained if there is a junior minister belonging to a different party in the same department. The third hypothesis thus postulates that the presence of a ‘hostile’ junior minister will diminish the share of board members held by the senior minister's party:

  • H3. The presence of a ‘hostile’ junior minister in a department leads to a decrease in board member shares for the senior minister's party.

The first three hypotheses concern primarily the quantitative inter-party distribution of appointments, thus mirroring the classical portfolio allocation literature. Yet coalition researchers have long argued that not all pay-offs are equal in value. Some ministerial posts are clearly more prestigious and/or influential than others. The first systematic attempt to gather data on ‘portfolio salience’ was Michael Laver and Ben Hunt's (1992) expert survey that included a question on the ordinal ranking of a substantial number of ministries for each country. Paul Warwick and James Druckman (2001) used these measures to reassess the proportionality proposition against the predictions from bargaining power models. Yet dissatisfied with the limitations inherent in the data provided by Laver and Hunt, the same authors conducted a more comprehensive expert survey on the salience of portfolios in fourteen West European countries (Druckman and Warwick, 2005), which allowed for a more convincing demonstration of the empirical validity of ‘Gamson's Law’ (Warwick and Druckman, 2006).

For the present purpose, the main implication of the research on portfolio salience is that, just as the value of ministerial portfolios differs, there is clearly variation in the importance of corporations in the public sector. However, absent a genuine measure of the importance of state-owned enterprises, one needs to resort to proxy indicators. Two most simple and intuitive indicators of the importance of a corporation are its capitalisation and its number of employees. Whereas the former may be indicative of its relevance for the delivery of public services (think of state-owned railway corporations, banks or electricity producers), the latter may be a reasonable proxy for a company's value as a patronage resource. From the perspective of an intrinsically office-seeking politician, there is yet another measure by which corporations may vary in ‘salience’: the value of a board membership in a public sector company may, first and foremost, be determined by the remuneration that comes with it. Hypotheses 4 to 6 are therefore contingent on the extent to which a party values policy-making capacities, patronage resources or the most immediate perks of office:

  • H4. Party patronage in state-owned enterprises is more pronounced the higher a corporation's capitalisation.
  • H5. Party patronage in state-owned enterprises is more pronounced the larger a corporation's staff.
  • H6. Party patronage in state-owned enterprises is more pronounced the higher a corporation's remuneration for board members.

These three hypotheses account for the notion that, as ministerial portfolios vary in importance, so do public sector companies. To be sure, the idea of varying portfolio salience can be developed further. The importance of ministries may not only vary in general terms, but also across parties. Indeed, coalition research has produced good evidence that parties are not only interested in how large a share of ministerial posts they are awarded, but also care about the qualitative distribution of ministries. Eric Browne and Karen Ann Feste (1975) were the first to show that parties do seem to have specific preferences as to which portfolios they are assigned. This finding was later corroborated by Ian Budge and Hans Keman (1990, pp. 89–131) in their study of twenty Western democracies. More recently, Hanna Bäck et al. (2011) provided the hitherto most comprehensive evidence that variation in issue emphasis is a significant predictor of the qualitative allocation of ministerial portfolios in coalition governments. In other words, parties genuinely care for some policy areas more than for others – a notion that also lies at the heart of the saliency theory of party competition (Budge, 1982; Budge and Farlie, 1983).

To the extent that parties understand patronage as an instrumental good for the purpose of influencing policy, this argument can be translated to the realm of public sector patronage. One would, for instance, expect a party with a specific interest in farming and forestry to have a higher presence in the agricultural sector, whereas a party that emphasises transport and mobility may be more inclined to appoint loyal adherents to corporations in the railway and roadwork sectors. This is the logic behind Hypothesis 7:

  • H7. The partisan composition of managerial boards reflects variation in policy emphases across parties.

Case Selection, Data and Method

By all standards, Austria is a most likely case with respect to the occurrence of party patronage (Ennser-Jedenastik, 2013; Müller, 1988; 1989; 2000a; 2006, pp. 148–9). In part, this is due to the fact that, historically, state-owned industries accounted for a sizeable share of economic activity. During the 1970s, a whopping 12 per cent of the Austrian labour force was employed in state-owned enterprises (OECD, 1985, p. 76). In the aftermath of the Second World War most industries engaged in iron and steel manufacturing, coal mining, oil extraction, electricity generation and banking4 were nationalised, thus providing patronage resources that parties were quick to exploit. While it may thus be a trivial statement to assert that parties in Austria engage in patronage activities, the degree of partisan penetration of state-owned corporations may still be mind-boggling to anyone not acquainted with the functioning of the Austrian party state (Dobler, 1983; Fehr and Van der Bellen, 1982). As a case in point, consider that all coalition agreements during the first era of grand coalition governments (1945 to 1966) explicitly stated that managerial positions in state-owned industries be allocated to government parties according to their seat shares in parliament, thus indicating that control over these positions was, in fact, subject to inter-party bargaining. While such direct evidence for party patronage is nowhere to be found in today's coalition agreements, there is little reason to suggest that control over these appointments is no longer a substantial part of the coalition bargain. To be sure, the public sector has undergone substantial transformation since the heyday of the post-war grand coalition governments, and so has the nature of patronage (Treib, 2012). One of the most important institutional changes is that the formal power to make appointments nowadays rests with the minister under whose jurisdiction a public sector corporation falls.5 For much of the early post-war decades, appointments to the corporations in nationalised industries were under the control of the Ministry of Transport and Public Enterprise, the Ministry of Finance or the Federal Chancellery, and had to take the parliamentary strength of parties into account (Dobler, 1983, p. 326).

During the past decades privatisation and reorganisation efforts have substantially shrunk the share of corporate activity directly controlled by the federal government (Aiginger, 2003; Meth-Cohn and Müller, 1994). Still, a substantial number of corporations remain under government control, not least since the rise of regulatory capitalism in Western Europe (Gilardi, 2008; Thatcher and Stone Sweet, 2002) has found its expression in the establishment of new regulatory agencies in Austria.

The seven hypotheses outlined in the theoretical section are tested on data from state-owned corporations in Austria between 1995 and 2010, including only firms that were majority owned by the federal state. These corporations can easily be identified in the annexes to the federal budgets of the respective years. The names and tenures of all members of executive and supervisory boards6 are taken from the official commercial register (Firmenbuch). This yields more than 2,000 individuals nested in 92 corporations, ranging from large entities such as the Austrian Federal Railways (with a capitalisation of almost 2 billion euros and a staff of around 45,000) to small regulators such as the Railroad Control (0.75 million euros, twelve employees). The distribution of corporations across business sectors is presented in Figure 1 (see Appendix for a complete list of corporations).

Figure 1.

Distribution of Corporations across Business Sectors

To be sure, there are some formal differences in the appointment procedures of these individuals. Most importantly, executive appointments in stock corporations (Aktiengesellschaften) can only be indirectly controlled by politicians through their influence over the composition of supervisory boards. Even so, anecdotal evidence gathered during the data collection process suggests that Austrian politicians still have a firm grip on these appointments. Also, since these formal differences concern only a tiny minority of all posts (88 individuals), excluding these cases from the analysis does not alter the results significantly.

Through an extensive research of biographical encyclopedias, official government documents, election lists, parliamentary records, media databases and party and government websites, the partisan affiliation for each individual board member was determined. These data were then supplemented with information provided by a number of well-informed journalists and (former) civil servants working in the ministerial bureaucracy.7 Party affiliation was operationalised as a person (1) having held public or party office, (2) having served as an aide to an MP or in a ministerial cabinet, (3) being a member of a party or (4) being closely affiliated with a party. Overall, about half of all individuals in the data have identifiable party ties.

Two caveats need to be discussed here: first, while the first three categories of affiliation are objective, the last is potentially problematic in terms of validity as it is usually based on information found in journalistic accounts. However, given the importance of party in the Austrian public economy, it can be assumed that domestic and economic journalists are among the best-informed sources when it comes to identifying the partisan ties of public sector managers. Also, some anecdotal evidence in the data suggests that these ‘close affiliates’ are often, in fact, party members whose ‘true’ degree of affiliation is not perfectly observed by journalists.

Second, it is perfectly possible, even likely that the data gathered on individual party affiliation are incomplete. Not each party affiliation may be discernible, especially for low-profile individuals. Yet in coding as non-partisan those individuals who could not be identified as party adherents, the analysis, if anything, understates the level of party patronage in the data, thus loading the dice against the expectations outlined above.

As additional evidence for the validity of the data collection, consider the fact that the thousands of documents researched for this study returned contradictory information on party affiliation only in one case (which could be resolved after some additional research).

Each of the seven hypotheses is operationalised through one independent variable. Seat shares are calculated from the initial distribution of parliamentary seats (H1); two dummy variables indicate ministerial responsibility, one pertaining to unconstrained ministers (H2) and one to ministers monitored by watchdog junior minister (H3). Data on capitalisation (H4) and staff (H5) are taken from official government reports and annual reports obtained from the corporations; data on the remuneration of board members (H6) are obtained from reports compiled by the Austrian Court of Audit (http://www.rechnungshof.gv.at), ministers' responses to parliamentary questions and corporations' annual reports.8 However, due to the absence of uniform standards for the disclosure of financial information, remuneration data are missing for about 100 observations.

Finally, the policy emphasis variable (H7) is operationalised as the percentage of statements in a party's most recent election manifesto that refer to the sector in which a corporation operates (see Figure 1). For example, the SPÖ devoted 3.5 per cent of its 1999 manifesto to research-related matters. During the first measurement period, the policy variable thus takes on the value 3.5 for all observations relating to the SPÖ's share of board members in research-oriented corporations (e.g. the Austrian Space Agency or the Austrian Research Centres).9 Note that the policy, capitalisation, staff and remuneration variables have been log-transformed in order to conform to the normality assumption. The summary statistics are reported in Table 1.

Table 1. Summary Statistics of Independent Variables
VariableMeanStd deviationMinimumMaximum
  1. Note: n = 424, except for remuneration where n = 326.
Seat share0.5000.1550.1850.814
Minister, no watchdog JM0.3770.48501
Minister plus watchdog JM0.1340.34201
Capitalisation (logged)0.5482.753−3.3527.550
Staff (logged)4.4302.008010.867
Remuneration (logged)7.4301.2604.6059.923
Policy emphasis (logged)−0.1481.765−4.6051.923


In analogy to the portfolio allocation literature, the following analysis uses the share of partisan board members for each corporation and government party as the dependent variable. The period of observation stretches from January 1995 to December 2010. During that period, Austria had six governments.

In principle, one could take measurements of the dependent variable for each of the cabinets. However, this would likely lead to distorted results, since the two periods of grand coalition government in the data (1995–2000 and 2007–2010) display very little variation, for example with regard to the relative strength of parties and the distribution of portfolios. Therefore, the dependent variable will be measured only for those four periods that are delimited by ‘major’ changes in the government composition (see rightmost column in Table 2).

Table 2. Austrian Governments, 1995–2010, Cabinet Shares and Measurement Periods
GovernmentStartEndSenior partyJunior partyMeasurement periods
  1. Note: Figures in parentheses are percentage shares of the coalition's parliamentary seat total.
  2. *In April of 2005, the FPÖ's ministers and most MPs split from the party to form the BZÖ. The bulk of the party's rank and file, however, remained loyal to the Freedom Party.
Vranitzky IVNov 1994Mar 1996SPÖ (55.6)ÖVP (44.4)Period I (grand coalition)
Vranitzky VMar 1996Jan 1997SPÖ (57.7)ÖVP (42.3)
KlimaJan 1997Feb 2000SPÖ (57.7)ÖVP (42.3)
Schüssel IFeb 2000Feb 2003ÖVP (50.0)FPÖ (50.0)Period II (centre-right)
Schüssel IIFeb 2003Jan 2007ÖVP (81.4)FPÖ/BZÖ (18.6)*Period III (centre-right)
GusenbauerJan 2007Dec 2008SPÖ (50.7)ÖVP (49.3)Period IV (grand coalition)
FaymannDec 2008SPÖ (52.8)ÖVP (47.2)

It also needs to be taken into account that, unlike ministerial portfolios, patronage appointments are not necessarily determined – much less effected – on day one of a cabinet. Depending on their specific status, the tenure of board members is regulated by law10 or contract. Early dismissals may therefore impose costs, which is why many appointments are made with considerable time lag. In addition, parties may be constrained by having earlier campaigned on an anti-patronage platform or by low organisational capacities that impede the recruitment of adequately qualified personnel.11 Figure 2 illustrates how, after the FPÖ enters government in early 2000, its share of board members rises only slowly. By contrast, the proportion of SPÖ affiliates picks up rather quickly after the party returns to power in early 2007. Given these time lags, the values of the dependent variable are specified as averages across the last six months of each measurement period. This constitutes a reasonable approximation to the supposed ‘equilibrium distribution’ of patronage appointments as resulting from the bargaining process between coalition parties.

Figure 2.

Aggregate Share of Board Members by Party, 1995–2010

Note: Horizontal grey bars indicate government composition; remainder is share of non-partisan board members.

To be sure, not all corporations were in existence or under federal control during all four measurement periods. Some were established only very recently, while others were privatised, merged or simply abolished at some point in time. The data set hence contains 54, 59, 48, and 51 corporations for the four respective measurement periods; with two parties present in each cabinet, this results in a total of 424 observations ((54 + 59 + 48 + 51) * 2 = 424).

Since the dependent variable is measured as each party's proportion of board members, it takes on values between 0 and 1 (see Figure 3). Furthermore, the fact that not each and every managerial position is held by a partisan leads to a heavily skewed distribution (skewness = 1.19). Given that the dependent variable therefore violates the assumptions for OLS regression, generalised linear models (GLM) are used in the multivariate analysis.

Figure 3.

Histogram of Dependent Variable

Note: mean = 0.16, median = 0.13, skew = 1.19.

The GLM framework requires the choice of a distributional family and a link function. A modified Park test (Manning and Mullahy, 2001) suggests that the Poisson distribution is the most appropriate modelling choice.12 A link test based on Daryl Pregibon (1980) identified the log function as the most preferable link function. In line with the portfolio allocation literature (Warwick and Druckman, 2006), standard errors are clustered on corporations and measurement periods. The analysis thus accounts for the fact that, at any point in time, one party's share of board members is not independent of another's.

The overall picture suggests that the hypotheses referring to the institutional make-up of cabinet (H1 to H3) have substantial explanatory power. Likewise, the assumptions pertaining to the relevance of corporations (H4 to H6) are supported by the data, albeit that these variables are correlated to a certain degree (between 0.44 and 0.63) so that the effects of the capital and staff predictors are cancelled out by the remuneration variable in model 7. The policy-related hypothesis (H7) must be rejected on the basis of the figures in Table 3.

Table 3. Explaining Party Patronage in State-Owned Enterprises
 Model 1Model 2Model 3Model 4Model 5Model 6Model 7
  1. Note: Figures are coefficients from generalised linear models using a Poisson distribution and a log link function; t-statistics in parentheses; standard errors clustered on corporation-years.
  2. *p < 0.05; **p < 0.01; ***p < 0.001; †p < 0.1.
Seat share1.832*** 1.091***1.063***0.852**1.099***0.868**
(7.95) (3.72)(3.82)(2.83)(3.84)(2.93)
Minister, no watchdog JM 1.055***0.945***0.928***0.899***0.939***0.894***
Minister plus watchdog JM 0.607***0.637***0.659***0.551**0.647***0.560**
Capitalisation (logged)  0.0806***  0.0672***0.00763
  (5.84)  (4.35)(0.39)
Staff (logged)   0.0829*** 0.03560.0207
   (4.12) (1.65)(0.85)
Remuneration (logged)    0.245*** 0.215**
    (5.21) (3.28)
Policy emphasis (logged)  −0.0564*−0.0438−0.0532−0.0546*−0.0527
Log likelihood−168.9−163.9−161.2−162.0−127.2−161.1−127.2
cor (Y, Ŷ)0.230.450.510.490.530.520.54

To begin with, the seat share predictor (H1) is positive and highly significant, thus indicating that, while the influence of individual ministers to bias the partisan composition of managerial boards in their favour is considerable, the proportionality norm can clearly be detected in the data. To be sure, the relationship between parliamentary seat shares and the dependent variable is much weaker than has been found in the traditional portfolio allocation literature. Figure 4 shows that, compared to the ‘Gamson line’, the predicted share of board members rises rather slowly as the value of the seat share variable increases from 0 to 1. While Gamson's (1961) proportionality proposition thus clearly has its relevance with regard to the allocation of board memberships, the relative strength of parties impacts only moderately on the share of board members.

Figure 4.

Predicted Share of Board Members by Seat Share

Note: Calculations based on model 6; all other variables held constant at their respective means or modes. Solid lines indicate 95 per cent confidence intervals.

As a simple and intuitive measure of the model fit, Table 3 reports the correlations between the predicted values and the dependent variable (Zheng and Agresti, 2000). The value for model 2 suggests that ministerial responsibility is the single most important determinant of patronage patterns in state-owned corporations. While the overall distribution of power in a coalition is far from irrelevant (as evidenced by the highly significant seat share variable and a correlation of 0.23 between Y and Ŷ from model 1), ministers appear to have great leeway over appointments to public corporations. To some extent, this finding, which is extremely robust across all model specifications, is indicative of a departure from the strict proportionality norm that structured patronage appointments during the golden age of the Austrian nationalised industries (Dobler, 1983, pp. 327–31). It suggests that patronage today is primarily structured along ministerial jurisdictions.

However, not all ministers are created equal. The coefficients for ministers shadowed by watchdog junior ministers are consistently lower than those for unconstrained ministers. This confirms H3. There is thus good reason to believe that the presence of a ‘hostile’ junior minister has a negative effect on the share of board members in a corporation that is affiliated with the senior minister's party. This finding is all the more important since earlier studies (Lipsmeyer and Pierce, 2011; Thies, 2001) have generally tested the observable implications of the assumed ‘watchdog’ role of junior ministers without actually demonstrating the empirical effects of this assumption. The models in Table 3 thus provide some of the first evidence that watchdog junior ministers do have a practical impact on the governance of coalition cabinets.

In order to illustrate the relationship between ministers and junior ministers in more detail, Figure 5 shows the marginal effect of the two minister variables based on model 6, with all other variables held constant at their mean. All else being equal, the predicted share of board members thus increases from 9 to 23 per cent when the respective party holds the portfolio with no interference by a watchdog junior minister. In the presence of a ‘hostile’ junior minister, the senior minister's party is predicted to control only 17 per cent of the board members. In terms of party patronage, the net effect of the average watchdog junior minister is therefore a reduction in the share of board members affiliated with the minister's party by no less than 6 per cent.13 Taken together, this is strong support for the assumption that, to a considerable extent, party patronage in public corporations follows the logic of ministerial government. Hence, ministerial partisanship is arguably the single most important predictor of the partisan composition of management boards in state-owned enterprises, albeit that the freedom of individual ministers to hand out appointments can be severely constrained by watchdog junior ministers.14

Figure 5.

Predicted Share of Board Members Conditional on Ministers/Junior Ministers

Note: Calculations based on model 6; all other variables held constant at their mean. Solid lines indicate 95 per cent confidence intervals.

Also, the predictors for the saliency of corporations do yield significant results. When regressing the dependent variable on each of these predictors individually, the model fit measures (correlations between Y and Ŷ) yield values of 0.20 (capitalisation), 0.17 (staff) and 0.24 (remuneration). This suggests that their explanatory power is similar to that of the seat share variable, but considerably lower than for ministerial partisanship. It should also be noted that the three variables (capitalisation, staff and remuneration) correlate to a considerable degree. Indeed, larger corporations tend to have higher amounts of capitalisation, employ more people and pay higher reimbursements to their board members. Thus, the remuneration variable cancels out the effects of the other two variables when all three predictors are included in one single model. It is hence difficult to disentangle empirically the effects of these three variables. Still, with regard to Hypotheses 4 to 6, it can be concluded that the ‘value’ of a corporation corresponds with the overall level of patronage.

This argument can be supported by a closer inspection of some corporations that do not score above average on any of the measures of ‘corporation saliency’ employed in the analysis, but can be said to be of special importance due to their cultural significance or for other reasons. Corporations such as the Spanish Riding School, Schönbrunn Palace or the Schönbrunn Zoo, all three epitomes of Austrian cultural identity (and pompous tourist attractions), display considerably higher levels of patronage than the average corporation. An even more extreme case is the Wiener Zeitung, the government's official gazette, which displays one of the highest patronage levels of all corporations. The motivation here appears to be a desire to exert control over the distribution of information.

The negative and (weakly) significant coefficient for the policy emphasis variable runs counter to H7. Also, at cor(Y, Ŷ) = 0.07 the model fit for a simple regression of the dependent variable on this predictor is very low. Yet one should be cautious to dismiss the underlying theoretical argument and interpret these results as indicating that parties do not politicise boards in those policy realms that they deem important. Rather, the findings may be due to the fact that many corporations operate in sectors that are not politically salient, such as transport, culture or administration. Measuring policy emphasis through references in election manifestos may therefore fall short of capturing adequately the party-specific importance of a policy sector that features low in electoral contests.

Taken together, the results of the above analysis suggest that party patronage in public sector corporations is structured by a number of factors: the relative parliamentary strength of the government parties plays a certain role, although the influence of individual ministers is dominant. Junior ministers, however, can constrain their departmental superiors in handing out appointments. As a general rule, patronage is more pronounced in corporations with higher capitalisation, larger staff and more extensive reimbursements for board members. Yet the net effect of each of these variables is difficult to determine due to high inter-correlations. The assumption that patronage patterns reflect parties' policy emphases is not borne out by the data.


The analysis above presents one of the most extensive single-country studies of patronage appointments to date. In light of the evidence gathered, there is ample support for the presumption that theories of government formation, portfolio allocation and coalition governance have substantial explanatory power when it comes to analysing party patronage. This, in turn, reinforces the conceptualisation of patronage as a linkage mechanism by which parties exert control over integral parts of the state apparatus beyond the Cabinet. Patronage patterns can thus be expected to co-vary with changes in the partisan make-up of the executive. To be sure, the present analysis focuses exclusively on the patron's activity (handing out appointments) and disregards the reciprocal part of the exchange relationship between patron and client. Since it thus remains unclear what type of good government parties receive in exchange for the appointments they make, one cannot draw definite conclusions as to whether reward or control is the central motivational driver of party patronage. Yet some of the findings (e.g. those relating to the varying importance of public sector corporations) suggest that control aspects are an important determinant of patronage (see also Treib, 2012).

In addition, some qualifications with respect to the generalisability of the above findings are in order. Given that the analysis is limited to one country that is not necessarily representative of other West European democracies in terms of party patronage, it remains to be seen to what extent the findings generalise to other political systems. To be sure, one could be more confident about the general validity of the hypotheses if they had been found confirmed in a low-patronage political system. This implies that the extent to which the above findings can be generalised is likely to be a function of the extent of party patronage that is present in a political environment. For the present purpose, however, the ‘likely case’ of Austria provides a fertile testing ground for the application of theoretical concepts from coalition theory to the study of patronage.

Furthermore, it is not the main ambition of this study to extrapolate from the substantive results generated for the Austrian case to other countries, but to demonstrate the analytical potential of coalition theory for the study of patronage. The generalisability of this theoretical approach may therefore be higher than that of the actual results. While the findings presented above may be specific to the country under study, it is conceivable that the benefits of studying party patronage through the lens of coalition theory apply to a larger set of cases, thus providing a useful framework to explain variation in patronage patterns not only between but also within countries. Since most concepts in coalition theory have been developed and tested on a wider sample of (West) European post-war democracies, it can plausibly be argued that they can be employed for the study of patronage in a variety of contexts, even if the specific results may differ substantially from the ones presented here. This perspective opens up new possibilities for research on party patronage and encourages further empirical investigation of a phenomenon whose dynamics are still insufficiently understood.

Appendix: Appendix: List of Corporations

AIT Austrian Institute of Technology GmbH

ASFINAG Autobahn- und Schnellstraßenfinanzierung-AG

Austria Wirtschaftsservice GmbH

Austria-Film und Video GmbH

Austrian Business Agency GmbH

Austrian Development Agency GmbH

Austrian Space Agency GmbH

AustriaTech – Gesellschaft des Bundes für technologiepolitische Maßnahmen GmbH

Austro Control GmbH

BIG Bundesimmobilien-GmbH

Brenner Eisenbahn GmbH

Buchhaltungsagentur des Bundes

Bundesbeschaffung GmbH

Bundespensionskasse AG

Bundesrechenzentrum GmbH

Bundessporteinrichtungen GmbH

Bundestheater Holding GmbH

BÜRGES Förderungsbank für wirtschaftliche Angelegenheiten GmbH

BUWOG Bauen und Wohnen GmbH

EBS Wohnungs-GmbH Linz

Eisenbahn Hochleistungsstrecke AG

Energie-Control Austria

Entwicklungsgesellschaft Aichfeld-Murboden GmbH

ESG Wohnungs-GmbH Villach

Familie & Beruf Management GmbH

Felbertauernstraße AG

Finanzierungsgarantie-GmbH Ost-West-Fonds

Gemeinnützige Wohnbau-GmbH Villach

Gesundheit Österreich GmbH

Graz-Köflacher Bahn und Busbetrieb GmbH

Großglockner Hochalpenstraßen AG

Hypo Alpe-Adria-Bank International AG

IEF-Service GmbH

Innovationsagentur GmbH

Internationales Amtssitz- u Konferenzzentrum Wien AG

KA Finanz AG

Kärntner Flughafen Betriebs-GmbH

Lagereibetriebe GmbH

Landwirtschaftliche Bundesversuchswirtschaften GmbH

Lokalbahn Lambach-Vorchdorf-Eggenberg AG

Marchfeldschlösser Revitalisierungs- und Betriebs-GmbH

Monopolverwaltung GmbH

MuseumsQuartier Errichtungs- und Betriebsgesellschaft mbH

ÖBB Holding AG

Österreich Institut GmbH

Österreichische Agentur für Gesundheit und Ernährungssicherheit GmbH

Österreichische Autobahnen- und Schnellstraßen AG

Österreichische Bibliothekenverbund und Service GmbH

Österreichische Bundesbahnen

Österreichische Bundesfinanzierungsagentur GmbH

Österreichische Bundesforste AG

Österreichische Donau-Betriebs-AG

Österreichische Forschungsförderungs-GmbH

Österreichische Industrieholding AG

Österreichische Mensen Betriebs-GmbH

Österreichischer Austauschdienst GmbH

Österreichischer Bundesverlag GmbH

Österreichischer Exportfonds GmbH

Rundfunk und Telekom Regulierungs-GmbH

Schienen-Control GmbH



Schloß Schönbrunn Kultur- und Betriebs-GmbH

Schönbrunner Tiergarten GmbH

Spanische Hofreitschule – Bundesgestüt Piber

Technologieimpulse GmbH

Telekom-Control GmbH

Timmelsjoch Hochalpenstraße AG

Umweltbundesamt GmbH

Verbund AG

via donau – Österreichische Wasserstraßen GmbH

Villacher Alpenstraßen Fremdenverkehrs-GmbH

WBG Wohnen und Bauen GmbH Wien

Wiener Zeitung GmbH

Wohnungsanlagen-GmbH Linz


The author would like to thank the participants of Wolfgang C. Müller's research seminar at the University of Vienna's Department of Government and three anonymous reviewers for their valuable comments. This research has considerably benefited from my work under the auspices of the Austrian National Election Study (AUTNES), a National Research Network (NFN) sponsored by the Austrian Research Fund (FWF) (S10903-G11).

  1. 1

    This is what Müller (1989, p. 334) refers to as power patronage (see also Eschenburg, 1961).

  2. 2

    Note that this may imply a non-simultaneous exchange between patron and client, where the partisan's loyalty precedes the reward of a patronage appointment.

  3. 3

    More specifically, it was shown that the empirical evidence for the formateur advantage, one of the prime implications of bargaining power models, is due to endogeneity in the coding of the formateur status variable.

  4. 4

    See BGBl. Nr. 168/1946 and BGBl. Nr. 81/1947 in the Federal Law Gazette.

  5. 5

    See, for instance, BGBl. 439/1986 of the Federal Law Gazette.

  6. 6

    Like Germany, Austria has a two-tier system which requires all stock companies (Aktiengesellschaften) to have two separate boards, an executive (Vorstand) and a supervisory board (Aufsichtsrat). Similar regulations apply to limited liability corporations (GmbH).

  7. 7

    Three journalists and four (former) bureaucrats assisted in the data collection process on the condition of anonymity. These experts were asked specifically to provide information on individuals where little or no information was available from other sources.

  8. 8

    The remuneration variable uses the average annual remuneration paid to members of supervisory boards, since those make up the vast majority of all appointments.

  9. 9

    This operationalisation is based on an analysis of election manifestos conducted by a team of researchers within the framework of the Austrian National Election Study (AUTNES). The AUTNES manifesto coding scheme comprises more than 650 issue categories and thus allows for a measurement of policy saliency specifically tailored to the present purpose.

  10. 10

    Members of supervisory boards, for instance, can be appointed for a maximum term of four years. Appointments can, however, be renewed indefinitely.

  11. 11

    The author would like to thank one of the anonymous reviewers for pointing out this argument.

  12. 12

    For the modified Park test, the squared residuals from a GLM model with a gamma distribution and a log link function were regressed on the log of the predicted values. This yields a coefficient of 1.18 (SE = 0.14), suggesting that the variance of Y conditional on X is roughly proportional to the mean. Hence a Poisson distribution is appropriate (see Manning and Mullahy, 2001, p. 471).

  13. 13

    This reduction is statistically significant at p = 0.053.

  14. 14

    Interestingly, however, the appointment of watchdog junior ministers does not result in higher board member shares for the junior minister's party in corporations under the jurisdiction of the respective portfolio (results not shown). In terms of patronage, their role is thus limited to curtailing the actions of the senior minister.


  • Laurenz Ennser-Jedenastik is a researcher with the Austrian National Election Study (AUTNES) at the University of Vienna's Department of Government. His research focuses on political parties, party competition, coalition politics, political appointments and patronage. He is currently working on a project examining patronage appointments to regulatory agencies in seventeen European democracies. His publications include articles in Governance, West European Politics, Party Politics, Political Studies, Journal of Legislative Studies and Local Government Studies. Laurenz Ennser-Jedenastik, Department of Government, University of Vienna, Pramergasse 9/E10, A-1090 Vienna, Austria; email: laurenz.ennser@univie.ac.at