Professional managers, public values? The delicate balance between corporatization and stewardship to society

ABSTRACT Corporatization is a trend in local public service delivery. However, research into how public values can be preserved after corporatization has lacked. This study analyses what factors underlie the presence of ‘public values’ in municipally owned corporations (MOCs). I find that the best predictor of stewardship to society in MOCs is their strategic managerial autonomy. This finding challenges the literature on public value failure, which argues that government by proxy should lead to lesser public values. Instead, it is plausible that increasing autonomy after corporatization can shield service delivery from political rent-seeking, which can protect public values.

that the best predictor of perceived stewardship to society in MOCs is strategic managerial autonomy in performance target setting (contributing positively to stewardship to society). This finding challenges the literature on public value failure, which argues that government by proxy is likely to lead to lesser public values. However, the model only explains a small part of the perceived stewardship to society of MOCs; I tentatively suggest that public values in MOCs might depend less on their governance structures and background variables and more on the individuals that govern and manage them.
I organize the remainder as follows. In section 2, I describe the background of this study. In section 3, I discuss possible predictors of stewardship to society in MOCs. In section 4, I explain our methodology. In section 5, I present the results of our analysis and test the robustness of our data. In section 6, I discuss our findings in light of the theory and the limitations. In Section 7, I conclude.

Background and relevance
Municipally owned corporations (MOCs) are organizations that have independent corporate status, are majority-owned by municipalities, and typically have extensive legal and managerial autonomy (Voorn, Van Genugten, and Van Thiel 2017). The autonomy granted to MOCs is a driving force behind their increasing popularity (Grossi and Reichard 2008). Municipalities hope that MOCs, through their autonomy, can improve efficiency (Krause and Van Thiel 2019). However, MOCs are also utilized as a step towards privatization (Cruz et al. 2014;Marra 2007), to alleviate fiscal stress (Citroni, Lippi, and Profeti 2013;Tavares 2017;Grossi and Reichard 2008;Ferry et al. 2018), or to charge user fees (Tavares and Camões 2007).
A substantial part of the literature on (local) corporatization is concerned about the absence of public values in MOCs, and particularly about MOCs' stewardship to society (Bourdeaux 2007;Da Cruz and Marques 2011;Ferry et al. 2018; Van Genugten, Van Thiel, and Voorn 2020). In MOCs, decision-making power is shifted to appointed boards, which shifts responsibility, accountability and democratic authority away from elected actors (Skelcher 2000). Municipalities cannot fully observe the interests and behaviour of the professional managers on these appointed boards, and that causes some problems for MOCs' societal accountability (Van Genugten, Van Thiel, and Voorn 2020).
However, academics have conducted little large-scale research about what contributes to public values like stewardship to society in MOCs. While various researchers have argued that the extent to which MOCs fulfill public objectives should not be taken for granted (Van Genugten, Van Thiel, and Voorn 2020;Da Cruz and Marques 2011;Bourdeaux 2008;Ferry et al. 2018), actual academic knowledge on how to retain public values after corporatization is still missing. This is problematic particularly for practitioners, as it remains unclear what governance mechanisms can be implemented to safeguard public values in their arm's length bodies.
Knowledge of this is particularly important, because while we know little about the predictors of stewardship to society in MOCs, we know that MOCs are taking up increasingly prominent roles in local public service delivery (Citroni, Lippi, and Profeti 2013;Voorn, Van Genugten, and Van Thiel 2017;Ferry et al. 2018;Aars and Ringkjøb 2011). Aars and Ringkjøb (2011) note that the number of MOCs in Norway increased from 850 in 1997 to more than 2500 in 2008. Bergh et al. (2018), using longer time series data, find an increase of the number of municipal enterprises in Sweden from 250 in 1965 to 1750 in 2015. Ferry et al. (2018) find a 50% increase in corporatization in the United Kingdom between 2010 and 2016; Voorn,  point out that going back further in time based on the data of Crediop (2004) shows that the bulk of corporatization in the United Kingdom already occurred before this period. Gradus and Budding (2018), using time series data for every year from 1999 to 2014, find that the frequency of the use of MOCs for service delivery in the Netherlands increased from use for about 6% of services in 1999 to use for just over 30% in 2010, after which it remained somewhat steady. Increased use of MOCs is also observed in Germany (Grossi and Reichard 2008), Italy (Citroni, Lippi, and Profeti 2013) and Portugal (Tavares and Camões 2007). Given this growth in the use of corporatization in many European countries, it is important for local governance to understand how their public values can be safeguarded.

Public values and stewardship to society
There is a substantial body of research on the values of public sector organizations (cf. Bozeman 2002;Jørgensen and Bozeman 2007;De Bruijn and Dicke 2006;Van Thiel and van der Wal 2010;Lyons, Duxbury, and Higgins 2006). Fuelling the academic debate on public values in public sector organizations is the increasing use of businesslike approaches to public service delivery following New Public Management in many countries (Hood 1995). In the face of this trend towards 'government by proxy' (Bozeman 2002), many researchers have voiced concerns about the government's ability to safeguard classical public values in a modern administration (Bozeman 2002;De Bruijn and Dicke 2006).
It is especially surprising that given this background of the public value literature, there has been little research into the effects of corporatization on public values (Van Thiel and van der Wal 2010). Research has focused extensively on the distinction between public and private organizations, while differences within the public sector have received less attention (Van der Wal, and Van Hout 2009). Despite the fact that much of the literature on public value failure has challenged 'government by proxy' (Bozeman 2002), arm's length bodies remain understudied, and particularly little is known about public values at the local level in municipally owned corporations (Van Genugten, Van Thiel, and Voorn 2020).
In this article, I focus on procedural public values (De Bruijn and Dicke 2006, 719;Bozeman 2002, 64), i.e. the public values surrounding the 'responsibility, accountability and authority for public action' of MOCs (following Skelcher 2000, 10), that I call 'stewardship to society'. Specifically, the procedural public values of responsiveness, accountability, and democratic service delivery, following Skelcher (2000), are the values that are most at risk when the state is hollowed out and when services are outsourced to arm's length bodies such as MOCs. Therefore, a strong adherence to these values suggests a strong stewardship to society in MOCs.
Procedural public values have been measured in public sector organizations before (Andrews and Warner and Hefetz 2002). However, to the best of my knowledge, there has not been research into predictors of public values in arm's length bodies, and certainly not in MOCs.I consider various predictors of the presence of procedural public values in MOCs.

Managerial autonomy
One key reason to engage in 'governance at arm's length' through MOCs is to allow autonomous (non-political) delivery of public services (Grossi and Reichard 2008).
Two key theories address the consequences of such increased autonomy on public values. First, following principal-agent theory, autonomy creates distance, which is expected to negatively affect MOCs: as the distance between public service organizations and their political principal increases, public objectives may be less likely to be satisfied (Brown and Potoski 2003a) and accountability can be lost (Moynihan 2006). This distance also creates more opportunistic behaviour: politicians may use MOCs for blame-shifting, and may be less inclined to monitor Vishny 1994, 1997). Second, principal-steward theory challenges how often such opportunistic behaviour really takes place, particularly in a public sector setting. Following principal-steward theory, autonomy is expected to lead to increased feelings of responsibility and motivation towards public values in MOCs (cf. Schoorman, Mayer, and James 2007; Van Thiel 2011).
Generally, the literature frames the negative effect as stronger than the positive effect, i.e. managerial autonomy is expected to have a negative effect on the stewardship to society of arm's length organizations. This is argued in the literature on agencification (Overman 2017;Schillemans 2008), and is also often argued for MOCs (Bourdeaux 2004;Ferry et al. 2018;Garrone, Grilli, and Rousseau 2013;Bergh et al. 2018). Thus, the first hypothesis is: H1: Managerial autonomy negatively affects stewardship to society in MOCs.

Legal autonomy
MOCs can operate under public (government) law, but they can also operate under private (commercial) law. MOCs operating under public law tend to be more political and less autonomous than their counterparts operating under private law, because public-law provisions mandate forms of political control of these MOCs that do not exist under private law (usually, presence of political executives on the board of directors). This is true in MOCs in many countries (Van Genugten, Van Thiel, and Voorn 2020). To paraphrase Zijlstra (2008): in principle, private-law organizations are free to do anything that is not prohibited by the law, and public-law organizations are unfree except to execute what they are allowed by law. In other words, private-law organizations have more legal autonomy (Verhoest et al. 2004a).
Although the literature on expectations about legal autonomy is much scarcer, the same reasoning about expectations should apply as for managerial autonomy: legal autonomy reduces political control, which reduces the chance that the objectives of the public principles are met. Simultaneously, legal autonomy (potentially) makes political rent-seeking more difficult. I expect the former effect to be more dominant than the latter. Thus, the second hypothesis is: H2: MOCs operating under public law demonstrate more stewardship to society than MOCs operating under private law.

External governance
Another factor with a predicted effect on stewardship to society in MOCs is external governance, and in particular performance management. This effect is more ambiguous.
On the one hand, according to principal-agent theory, performance management helps to enforce the wishes of political principles and thus acts as an accountability mechanism for MOCs that encourages stewardship to society. Clear objective setting can make the behaviour of MOCs more measurable, and evaluations and monitoring procedures help the government assess whether objectives are actually achieved, with positive effects for accountability (Pollitt 2005;Wholey and Hatry 1992;Mayne 2017). Next, performance measurement makes governance of MOCs easier, and makes it easier for political principles to encourage their democratic nature and responsiveness to society.
On the other hand, according to principal-steward theory, a focus on extrinsic motivation through performance management can 'crowd out' intrinsic motivation (Frey 1994;Benabou and Tirole 2003) and in that way lead to reduced motivation and negative effects on stewardship (Davis, David Schoorman, and Donaldson 1997b;Schillemans 2013;Van Slyke 2006). Moreover, it is argued in the literature that when objectives are precisely measured, this may inspire managerial performance only to the letter of the contract (Bernheim and Whinston 1998;Libby and Mayer 2011), which would reduce the responsiveness and democratic nature of MOCs. Thus, relational contracting is important, and a focus on governance through increasing alignment can motivate, reduce information asymmetries, and may make MOCs more 'public' in nature.
An increasing number of authors argue that principal-agent theory and principalsteward theory, and thus performance management and relational governance, are not necessarily mutually exclusive (Davis, David Schoorman, and Donaldson 1997a). For instance, Van Thiel (2011) argues that agency-based governance works when the professional manager is an agent, that relational governance works when the professional manager is a steward, and that stewardship is only discouraged when there is a mismatch. Thus, salient governance involves using governance strategies on a case-bycase basis, depending on individuals and circumstances. Therefore, the third hypothesis is: H3a: Performance management positively affects the stewardship to society of MOCs.
H3b: Relational governance positively affects the stewardship to society of MOCs.
H3c: Alignment positively affects the stewardship to society of MOCs.

Shared ownership
One common feature of MOCs in some countries, but not in others, is shared ownership, particularly in the form of intermunicipal ownership . This can potentially bring conflicts in governance (Bel and Sebő 2018;Blåka 2017;Cäker and Nyland 2017;Sørensen 2007), which may be problematic for public values for several reasons. In particular, when municipalities have different interests, MOCs often get more leeway to set their own goals and agendas (Garrone, Grilli, and Rousseau 2013). This makes them less accountable to elected actors. Therefore, I expect intermunicipal ownership to negatively contribute to stewardship to society in MOCs.
One way to increase accountability of MOCs under intermunicipal ownership is to make sure that MOCs are governed by only one of the municipalities, usually the largest municipality (Bel and Sebő 2018;Voorn, Van Genugten, and Van Thiel 2019). However, this policy can erode the democratic nature of multi-owned MOCs, and may lead to inequitable situations, where the wishes of the largest municipality, rather than the public interest, become dominant (Cäker and Nyland 2017;Spicer 2017). Thus, I expect that a key role for the largest municipality in governance negatively affects MOCs' stewardship to society.

H4a: Intermunicipal ownership negatively affects stewardship to society in MOCs.
H4b: A key role for the largest municipality in governance negatively affects stewardship to society in MOCs.

Internal governance: board structure
Board structure may also affect stewardship to society. First, board diversity is expected to contribute to MOCs' stewardship to society, as diversity of backgrounds and perspectives adds to board discussion and mitigates groupthink (Forbes and Milliken 1999;T Hart 1990), and may help MOCs to better reflect on the interests of the public. Among commonly perceived types of diversity to contribute to public values in boards are professional diversity, international diversity and gender diversity (Jehn, Northcraft, and Neale 1999;Miller and Del Carmen Triana 2009;Walt and Ingley 2003;Williams and O'Reilly 1998). Next, board size may affect stewardship: larger boards are likely to have more diverse thinking.
Thus, I formulate the following hypotheses: H5a: Board diversity positively affects stewardship to society in MOCs.
H5b: Board size positively affects stewardship to society in MOCs.

Background variables
Beyond these elements, I include some background variables that are potentially linked to stewardship to society. Organization size in terms of budget and number of employees may affect stewardship to society, increasing diversity in the organization, which may increase their responsiveness and accountability to society. The sector in which organizations operate may matter, because some sectors can be more technical than others, and regulators have more difficulty regulating more technical services because of lack of expertise or capacity, leading to more autonomy in those sectors Potoski 2003b, 2005; Tavares and Camões 2010) and thus lesser accountability and responsiveness. Next, organization age may affect stewardship to society, but we do not know in which direction: more settled organizations are more likely to have established salient practices, and so may be more accountable and responsive; conversely, information asymmetries are likely to grow over time, and so older organizations may be less 'public'.

Procedure
In February 2018, I submitted a survey to directors of all Dutch MOCs. In particular, directors were sent a letter through postal mail that contained links to an online survey administered through Qualtrics. This initial letter contained a deadline of two weeks from the date of submission, intended to bolster response. I sent a second letter in March 2018, again with a two-week deadline; following that deadline, I closed the survey.

Population and sample
I study in particular MOCs in the Dutch context. In the Netherlands, MOCs have the following characteristics (Voorn and Van Genugten 2021) describe these in more detail). Municipalities are the majority owners of the MOCs. MOCs have large legal and managerial autonomy to deliver services, and have independent corporate status. The executive branch of the municipality is responsible for the governance of MOCs. Similar to other countries (Van Genugten, Van Thiel, and Voorn 2020), corporatization can occur under both public law and private law. Under private law, MOCs must have a supervisory board, typically comprised of independent experts. Meanwhile, public law mandates that municipal executives are on this supervisory board, again not dissimilar to other countries (Van Genugten, Van Thiel, and Voorn 2020). Finally, specific about the Dutch context is the fact that intermunicipal MOCs are very common, making up three-quarters of all MOCs. I collected a database of all MOCs in the Netherlands. This could be done systematically: by law, Dutch municipalities are obligated to include a section 'Verbonden partijen' (translated: 'attached entities') that outlines (partial or complete) municipal ownership of any organization. From November 2017 to January 2018, I went through the most recent budgets of all 390 Dutch municipalities and included all organizations they owned in a database. I omitted 'empty' holding companies and organizations with no listed address. In total, this yielded a database of 809 MOCs. I sent out letters requesting survey participation to this entire population. Ten letters were returned indicating inability to deliver, leaving a total population of 799 MOCs.
In total, 243 surveys were returned (response rate 30.4%). 61 of those 243 surveys were incomplete (all under 30% completion); these were excluded for our research. Five more respondents suggested they had no budget or no employees, and were excluded. This left 177 complete surveys (completion rate 73.7%) for analysis. Table  1 shows descriptive statistics for the population (799) and our sample (177), specifically showing types and sectors, which are quite relevant in the context of public corporations (Park 2021). It shows that the sample is slightly skewed towards MOCs based in public law and towards larger MOCs, but is on the whole fairly representative of the population. Table 2 shows more descriptive statistics about our sample.

Survey measures
The survey was modelled after earlier surveys by the COBRA network that were sent to national agencies and quasi-autonomous government organizations across a variety of countries (COBRA 2010; Verhoest et al. 2004b). Like this study, studies done by the COBRA network (see, e.g. Christensen andLaegreid 2006, Verhoest et al. 2004;Verhoest, Verschuere, and Bouckaert 2007;Yesilkagit and Van Thiel 2011) used extensive survey questions to model governance issues in national agencies and quangos as completely as possible, and I follow their variable selection closely. In particular, I used items measuring autonomy and performance based on the fact that the factor loadings found in those earlier studies support the use of these items. The Dutch translations were based partially on earlier COBRA-based surveys in the Netherlands (Yesilkagit and Van Thiel 2011); the factor loadings of those translations validate the factors as well. The survey was furthermore tested in a pilot among former MOC directors and local government experts. I discuss the key measures below. All specific survey items can be found in the supplementary file.
The construct managerial autonomy was divided in financial autonomy, personnel autonomy, budget source autonomy, and strategic autonomy, four concepts found as defining sources varying in intensity among municipally owned corporations (Voorn, Van Genugten, and Van Thiel 2017) and all validated in the original COBRA survey (COBRA). For financial and personnel autonomy, respondents were given a list of types of autonomy they could have (10 items for financial autonomy, 9 items for personnel autonomy), which they could answer as 'yes', 'only with prior consent of shareholders', or 'no'. For measuring budget source autonomy respondents were asked to estimate the percentage of their budgets derived from various sources (municipal budgets, other public budgets, other private budgets, municipality tariffs, general tariffs, private assignments, and other). Higher scores indicate a higher percentage of budgets derived from each source. Strategic autonomy was measured with a single item question: 'who sets performance targets?', with five response possibilities: 'we set performance targets ourselves, without consulting the municipalities', 'we set performance targets, after consulting the municipalities', 'we set performance targets together with the municipalities', 'the municipalities set performance targets, after consulting us', and 'the municipalities set performance targets, without consulting us'. Legal autonomy was measured with a single direct question about the MOC's corporate status. I asked respondents to select one of the four possible corporate statuses of Dutch MOCs: enterprise (limited company), public-law organization (organization founded in government law, akin to a very autonomous agency), foundation (owner-controlled non-profit), or association (member-controlled non-profit), and retained both this four-category variable and a recoding of this variable to a binary variable (public law or private law).
External governance and performance management was measured using multiple single items. I measured (i) the presence, type, and formalization of target setting; (ii) the presence of, focus of, and party responsible for evaluations and monitoring; and (iii) the presence and types of bonuses and sanctions.
The construct alignment was primarily an own construct based on 5 items that measured the alignment between respondent organizations and owner municipalities (the 5 items were stewardship, information-sharing, flexibility towards the municipality, trust visible in input, and trust visible in output). The factor loadings were just enough to accept this as a factor (KMO measure .617; sig .000); for future studies, alternative measures would be preferred.
The presence of multiple principles was measured by asking for the number of owner municipalities, as well as for the presence of private or other public shareholders. I also asked respondents if they perceived one of the municipalities to be the dominant steering actor on a five-point Likert scale.
Board structure was measured using multiple single items. For this study, I measured (i) the number of board members; (ii) the diversity of board members in expertise, nationality, and gender; (iii) the existence of a supervisory board; and (iv) the role of the supervisory board. I recoded the diversity variable to measure diversity in relative numbers (the percentage of the board members). I also measured background factors with single items. I measured (i) size of organizations, both in number of employees and in size of budget; (ii) age of organizations; and (iii) sector, which we asked as an open question and then recoded ourselves by sector based on the ministries they would fall under if they were national organizations (to allow national-local comparisons). For the indicators organization size, I included both the log and the square root for our modelling to account for non-linear effects. I also created a measure 'labor intensity' by dividing the square root of the number of employees by the square root of budget size, to serve as a measure of service technicality. Lastly, I added a measure 'service measurability', a binary variable labelled as '1' when respondents described at least 1 performance indicator as measurable.
Lastly, the presence of stewardship to society was measured through self-assessment on a 10-point scale of 3 items, accountability to society, responsiveness to society, and democratic nature of service delivery, as validated in the original COBRA survey (COBRA, 2010). These items have the benefit of having already been validated in exploratory factor analysis, and also closely follow Skelcher's (2000) definition of stewardship to society as 'responsibility, accountability and authority for public action' of MOCs. This factor had acceptable loadings in our data (public values KMO = .662, sig .000).

Strategy of analysis
For the analysis, I opt for stepwise regression, a more automated approach that prevents type II errors and researcher bias that works by feeding all independent variables to the computer, which uses backwards elimination to discard the variable whose loss gives the most statistically insignificant deterioration of model fit, continuing to do so until the model reaches a specified criterion (Thompson 1995). I use the 'standard' criteria (probability of F for entry = .05, probability of F for removal = .10). I opt for stepwise regression particularly because the number of variables in our model is quite high compared to the number of observations, leading to modelling problems when including all variables in the analysis.
That said, stepwise regression has two key limitations. First, it can be prone to overfitting errors (type I errors) (Thompson 1995). This limitation is mitigated by the fact that the independent variables were collected after construction of our conceptual model. Next, stepwise regression assumes linearity in the relationships between dependent and independent variables. I make the data linear using various strategies, by normalizing the dependent variables budget size, number of employees, and organization age, that are expected to have non-linear relationships with public values, and by removing eight strong outliers for the public values factor. I include three robustness checks to try to offset the limitations of our stepwise approach. First, following the analysis, I shrink the 'best-fit' model, to observe if this has a substantial impact on model fit. Second, I re-run our regressions independently on the items in the stewardship measure, which will simultaneously help to uncover potential mechanisms. If results are inconsistent, hypotheses are not accepted. Third, for the results that are robust to this, I perform mediation analysis in order to test whether the findings are (partially) mediated by four potential mediators: managerial autonomy, legal autonomy, corporate status and alignment, which will likewise help to uncover potential mechanisms behind the findings and test the robustness of the findings at once. Table 3 reports ordinary least-squares (OLS) regressions of all our variables on stewardship to society (I do not report correlation statistics because of the too large number of variables). Column (1) shows the outcome of the stepwise regression on stewardship to society. The most critical factor predicting MOCs' stewardship to society seems to be the strategic autonomy of MOCs (significant at the p < 0.01 Table 3. Predictors of stewardship to society.

Findings
(1) level). It appears that autonomy in setting performance targets is related to a higher degree of self-reported stewardship to society. Another factor that significantly negatively affects stewardship to society is the extent to which the largest municipality is responsible for steering, calling attention to the problem of multiple principles. Lastly, alignment seems to be positively correlated with stewardship to society.
Columns (2) and (3) reduce the number of predictors to test the model's robustness. Reducing the number of predictors from (1) to (2) strongly reduces the model's R 2 , indicating the relevance of all predictors, but the variable 'alignment' loses predictive power and becomes non-significant. No significant loss of power occurs between models (2) and (3).
Column (4) shows the key model when controlled for the number of municipal owners, testing whether the responsibility for the biggest municipality steering affects organizations more as the number of owners grow. I find no evidence for this.
Finally, columns (5) and (6) include personnel and financial autonomy in the model to test the model's robustness. The model is relatively unaffected by their inclusion. Column (7) includes both forms of autonomy at the same time, again with no effect.
Overall, the results in Table 3 refute particularly hypothesis 1 -in contrast to expectations, managerial autonomy seems beneficial for stewardship to society. The other findings are not robust.
The model fit is low across all models (R 2 ranges from 0.077 to 0.193). This is unusual for stepwise regression and even more unusual for theory-informed stepwise regressions. Thus, either key factors have not been included in this model, or stewardship to society in MOCs depends more on the individuals responsible for governance and management than on organizational structure.

Sensitivity analysis
In the previous subsection, I offered one robustness check in the form of removal of or addition of key independent variables to our model. I also perform a robustness check on the dependent variable, by splitting up our stewardship factor into its three components: accountability, responsiveness, and democratic service delivery, and running OLS-regressions of our model (1) on the individual factor components.
This check demonstrates that strategic managerial autonomy is a robust predictor of stewardship to society, and is a significant predictor of all components of the factor public values (see Table 4). However, the findings are substantially less robust for the other three predictors. Alignment appears to correlate quite strongly with accountability to society and responsiveness to society, but not with the democratic level of service delivery. Interestingly, the governance role for the largest municipality is not a significant predictor for individual indicators, but is a significant indicator for them in conjunction, likely because of information gain after factorization. I run an additional robustness check to consider whether some key factors mediate the relationship between strategic autonomy and stewardship to society: personnel autonomy, financial autonomy, legal autonomy and alignment. I find no evidence for the existence of a mediating effect, and thus autonomy in setting performance targets appears to be the only direct, significant and robust predictor of stewardship out of the variables tested.
One possible reason for this could come from goal clarity. It could be that if MOCs have greater control over performance targets themselves, they create more measurable performance targets, which makes them perceive themselves as more accountable to the public. However, also goal measurability is no significant mediator of the strategic autonomy -stewardship to society relationship (see Table 5).

Findings
This study found three significant predictors of stewardship to society in MOCsstrategic autonomy, a steering role for the largest municipality, and alignment. However, because of the possibility of false positives in stepwise regression, I recommend conservative interpretation, and after sensitivity analysis, only one was a robust finding: strategic autonomy was a significant and robust predictor of the perceived stewardship of MOCs. Other factors lacked either robustness or significance.
The finding that strategic managerial autonomy is a positive predictor of stewardship to society runs counter to the hypothesis and the literature on this topic. While a large literature predicts that more autonomous MOCs should feel less public because of their distance to their political principles (Van Genugten, Van Thiel, and Voorn 2020;Da Cruz and Marques 2011;Bourdeaux 2008;Ferry et al. 2018;Garrone, Grilli, and Rousseau 2013), I find the opposite: professional managers of MOCs consider themselves to be greater stewards when they have more control over objectives. I have also shown that increased measurability of objectives is not the reason for this.
One explanation comes from public choice theory: by reducing the role of politicians in setting objectives for public service delivery, the goals of MOCs could be less 'political' and more 'public' -or at least directors of MOCs may perceive it as such. This is in line with findings from some countries of problems in MOCs that are not sufficiently autonomous from politicians (Da Cruz and Marques 2011;Bergh et al. 2018), and autonomy from political interference in general has been shown to have positive effects in municipalities (e.g. Bourdeaux 2008; Lee, Park, and Butler 2021). Thus, corporatization may take away political obstacles to public value delivery. An alternative explanation comes from stewardship theory: the strategic managerial autonomy granted to MOCs may make directors feel more responsible and motivated, and is reciprocated through increased public values (cf. Schoorman, Mayer, and James 2007;Van Thiel 2011). In other words, managerial autonomy may clarify responsibilities and consequently protect public values. Future research should help us understand this finding. A secondary key finding of this study was the surprisingly low fit of the explanatory model, despite it being drawn from a rather extensive amount of predictors all based in the literature. The most extensive model of the stewardship to society of MOCs explained under 20% of the variation in the stewardship to society of MOCs, which is unusually low, particularly for a stepwise regression, which due to its data-fitting approach tends to bring higher-fitting models. The relatively high number of indicators and our relatively low sample size both would also predict higher model fits. Therefore, it is likely that this study either missed a crucial predictor of performance in our study, or that the extent to which MOCs have public values is highly dependent on the characteristics of the individuals in the governance relationships ( Van Thiel 2011). Both of these possibilities point to the necessity of further research into predictors of the stewardship to society of MOCs, as to the best of my knowledge this is the first study into the public values of such organizations.

Limitations
This study has various limitations, relating to context, heterogeneity of our sample, and the subjectivity of particularly our stewardship measure. I will address these in turn.
First, these findings may be limited by context, as I addressed MOCs only in a single country: The Netherlands. One contextual factor that may particularly affect the findings is Dutch public law: the Wet gemeenschappelijke regelingen mandates the presence of politicians on supervisory boards of intermunicipally owned MOCs based in public law, which means that public law in the Netherlands inevitably politicizes MOCs. However, the finding that strategic autonomy is a predictor of public values without a mediating effect of legal autonomy implies that this may not be a serious limitation. Moreover, this type of public-law legislation exists in many other countries (Van Genugten, Van Thiel, and Voorn 2020). Nonetheless, further research should show whether these findings can be generalized to other countries.
Second, the findings may be limited by the heterogeneity of this study's sample. Since this study addressed the entire population of MOCs in the Netherlands, it dealt with a large variety of MOCs in organization types, sectors, sizes, and ages. In the context of such heterogeneous data and the relatively high number of variables, Type I errors (errors related to overfitting the data) become more likely. This appears not to be a problem in these data (judging from the low model fit and high significance of our finding), but I cannot verify this. I call for more data from other countries on the public values of arm's length bodies.
Next, as previously mentioned, the method of analysis, stepwise regression, is subject to a higher likelihood of false positives, and I therefore encourage conservative interpretation of our findings. In the context of this higher likelihood of false positives, I included three robustness checks, which helped narrow down the results to one core relevant variable for explaining stewardship to society -strategic autonomy, which is positively associated with stewardship to society.
A last limitation of the study is that all findings are based on survey data and selfassessment. While I used several procedural remedies in the survey design to limit common source bias (MacKenzie and Podsakoff 2012), such as reducing the potential for social desirability bias in item wordings, reducing ambiguity in scale items by using pre-validated scale items, and eliminating common scale properties (by using different scales both among independent variables and between independent and dependent variables), I cannot fully rule out the possibility of common source bias. Moreover, the stewardship measure is subjective and thus deals with perceptions rather than fact. For this reason, I encourage further research using more objective measures of stewardship to society.

Conclusion
Since MOCs are becoming increasingly numerous, and since it is hard to retain control over them due to their autonomy, it is important to understand how to ensure their commitment to public values. However, until now, no research has been conducted among them to understand what factors underlie such corporations' stewardship to society. This study uses a survey among all 799 MOCs in The Netherlands to try to uncover an answer to this question, and provisionally finds only one direct, robust and significant predictor of these corporations' perceived stewardship to society: their strategic managerial autonomy. In particular, MOCs that have more control over their own performance targets perceive themselves to be greater stewards to the public. This finding contradicts expectations from the literature, as the prevailing view is that MOCs will become less 'public' as they become more autonomous. However, I emphasize that more research is still necessary, particularly using more objective measures of stewardship to society, before governance implications can be drawn.

Disclosure statement
No potential conflict of interest was reported by the author(s).

Notes on contributor
Bart Voorn is assistant professor at the Institute for Management Research at Radboud University in the Netherlands. He studies the causes and effects of local corporatization, remunicipalization, and local government efficiency.