Hybrid Context, Management Practices and Organizational Performance: A Configurational Approach

Employing a configurational approach we explore how ‘hybrid context’ shapes organizations’ adoption, and performance implications, of management practice. We do because hybrid contexts have been a policy aim of many governments seeking to blurr the distinction between the public, private and not-for-profit sectors. To conceptualize hybrid contexts we employ the dimensions of market authority and (the multiple) political authority. Employing data from UK care homes, our findings suggest that: (i) the adoption, and performance effects, of management practices are conditioned by dimensions of hybrid context; (ii) there is significant variation across the configurations in terms of the mix of management practices that lead to high and low performance; and (iii) there is a high degree of symmetry between high and low performance, with good management practices being a necessary condition for high performing as compared to low performing organizations.


INTRODUCTION
In this paper, we explore how an organization's context influences its adoption of 'good' management practices and their performance implications. Traditionally the private, public and not-for-profit sectors have been defined as distinct sectors, with characteristics that influence the adoption of management practices (Allison, 1979;Rainey et al., 1976;Walmsley and Zald, 1973). Traditional sectoral differences, however, have been eroded over time as a result of government reforms aimed at harnessing the best features of private and public organizations (Bozeman, 2013;Newman, 2001). While the intersection of governments and markets has become more pronounced, scholarship of management and public policy have developed on parallel tracks (Bozeman, 2013) reflecting more traditional views of sectoral differences.
Since the 1980s governments have sought to implement reforms associated with new public management (NPM) in order to encourage the public sector to become more business-like, encompassing quasi-markets, performance management and contracting (Christensen and Laegreid, 2011;Hood, 1991;Lynn, 1998). Subsequently, governments have introduced post-NPM reforms to encourage forms of governance that differ from both traditional bureaucratic-hierarchical forms and more commercially oriented forms encouraged by NPM (Christensen and Laegreid, 2011;Currie and Martin, 2016;McNulty and Ferlie, 2002). Finally, there has been an expansion in the provision of traditional public services by not-for-profit sector organizations, contracting with both government and private clients, leading to the rise of the social enterprise (Skelcher and Smith, 2015).
To help understand the blurring of sectors, and its impact on organizations, scholars have increasingly looked to the concept of hybridity (Christiansen and Laegreid, 2011;Fossetol et al., 2015;Skelcher and Smith, 2015). Traditionally, hybridity has been viewed as an organizational level concept (Denis et al., 2015;Pollitt and Bouckaert, 2011), specifically structural hybridity around forms of governance (Denis et al., 2015;Miller et al., 2008). Hybridity, however, also relates to the context in which organizations operate (Skelcher and Smith, 2015). Recognizing the hybrid nature of contexts, Herbert Simon (1998) warned of the dangers of studying markets to the exclusion of government, and vice versa. In this paper we focus on hybridity of context, which has been a policy tool through which governments have sought to address pressing social issues, remedying the underlying resource constraints of public authorities by encouraging the introduction of private sector practices to improve organizational performance in the delivery of public services (Brandsen and Karre, 2011;Brandsen et al., 2005;Skelcher, 2005;Smith, 2010). In doing so we address the call for studies of contextual hybridity that move beyond narrow structural approaches that focus upon governance, to consider more broadly the interaction of policy, through hybrid context, and organizational practices that influence performance (Bryson et al., 2014;Denis et al., 2015;Moore, 2013).
To examine the complex relationships that exist between the different dimensions of contextual hybridity, management practices and organizational performance, we employ a configurational approach (e.g., Bensaou and Venkatraman, 1995;Doty et al., 1993;Ketchen et al., 1993;Ketchen et al., 1997;Meyer et al., 1993). The foundations for the configurational approach were presented by Ragin (1987), who emphasized the importance of causal complexity, with causality often being conjunctural, equifinal and asymmetric. Until recently, however, empirical studies employing a configurational approach have been hampered by the multivariate regression methods that involved additive, unifinal, and symmetrical effects (Fiss, 2007;Grandori and Furnari, 2008). In response, Fiss (2007) was amongst the first to employ the Qualitative Comparative Analysis (QCA) set-theoretic approach to study organizational configurations and complex causality, demonstrating how it might overcome the mismatch between theory and methods that had plagued earlier configurational theorizing. We adopt a set-theoretic approach to develop our understanding of the relationship between contextual hybridity, management practices and performance.
Through our study we make three contributions to the literature. First, we conceptualize organizations' hybrid contexts in terms of the interplay of market authority and political authority, as organizations are, to a greater or lesser degree, subject to both (Bozeman, 1987(Bozeman, , 2013. We operationalize market authority through the degree of competition in a market, and political authority through the concept of publicness (Bozeman, 1984(Bozeman, , 1987(Bozeman, , 2013Bozeman and Bretschneider, 1994;Johansen and Zhu, 2013;Nutt and Backhoff, 1993;Walker and Bozeman, 2011). To capture the multi-dimensional nature of publicness we focus on the three dimensions identified by Andrews et al. (2011) in their review of academic studies of publicness and organizational performance: ownership, funding and control. In doing so, we present the first study to bring together the dimensions of market authority and political authority in a single study to examine the influence on organizations' adoption, and performance effects, of management practices. Further, we align with Bozeman's (2013) call to employ publicness as a concept that can promote cross-fertilization between organization and management with public policy, because at its core the concept of publicness deals with elemental phenomena, related to the question 'what is public?'.
Second, we allow for the complex interdependencies that may exist between the dimensions of market authority and political authority, in shaping organizations' adoption, and performance implications, of 'good' management practices. Economic research has provided robust evidence for the productivity benefits of so-called good management practices, such as target setting, incentives and monitoring (e.g., Van Reenan, 2007, Bloom andVan Reenan, 2010). Similarly, three decades of work in the field of management indicates that better people management practices (e.g., 'high performance management') tend to lead to better financial and market performance in firms of all sizes and sectors (e.g, Combs et al., 2006;Hayton, 2003;Huselid, 1995;Ichniowski et al., 1997). However, contextual and environmental factors are expected to influence both the adoption, and the effects, of good management practices (e.g., Bloom and Van Reenen, 2007;Hayton, 2003;Jackson and Schuler, 1995;Ichniowski et al., 1997;Jackson et al., 2014;Patel and Cardon, 2010). By employing a configurational approach, in this paper we are able to examine whether or not the performance effects of management practices are conditional on the degree of market authority and political authority. In doing so, we extend current literature on high performance work systems that has only examined the effect of a limited set of contextual factors on practices, and which has been limited in its potential by employing a contingency rather than a configurational approach (Jackson et al., 2014;Meuer, 2017).
Third, hybrid contexts require organizations to pursue a blend of both social and economic outcomes. To date, studies of the effects of management practices upon the delivery of public services have produced equivocal results (Boyne, 2003), which may in part stem from the difficulty in measuring performance in this context (Boyne, 2003;Boyne and Walker, 2004;Boyne et al., 2005). We follow Andrews et al. (2010) and use a multi-dimensional measure of performance, encompassing social and economic outcomes, and both objective and subjective data. The data are drawn from inspection reports for each organization, which provide politicians, managers and the public information about © 2020 The Authors. Journal of Management Studies published by Society for the Advancement of Management Studies and John Wiley & Sons Ltd organizational success or failure in the delivery of public services (Audit Commission, 2001. Our study extends consideration of contextual hybridity beyond narrow structural governance, to dimensional publicness, and at the same time sees such hybridity as a multi-level interaction with organizational hybridity.
The remainder of our paper unfolds as follows. We begin by exploring the concept of publicness, delineating the dimensions of political authority and market authority, and explain the logic of configuration analysis and how it enables us to explore the relationship between hybrid context, management practices and performance. Second, we outline our empirical context, which is the market for old people's and children's care homes in the UK. We then outline the nature of our data collection, method and process of analysis. Third, we present and discuss the configuration analysis, which explores the relationship between hybrid context, management practices and performance. In our penultimate section we highlight our contribution to extant literature and suggest implications for policymakers, before drawing our conclusions.

HYBRID CONTEXT, MANAGEMENT PRACTICES AND PERFORMANCE
As outlined above, governments around the world have sought to blur the distinction between private, public and not-for-profit sector organizations, creating hybrid contexts as a means of addressing pressing social issues and encouraging the introduction of private sector practices to improve organizational performance in the delivery of traditionally public services (Brandsen and Karre, 2011;Brandsen et al., 2005;Skelcher, 2005;Smith, 2010). Below, we outline how contextual hybridity can be conceptualized in terms of the mix dimensions of market authority (competition) and political authority (ownership, funding and control), which can have direction and scale.
All organizations are subject to market authority to a greater or lesser extent. Private and not-for-profit organizations compete in markets (Desai and Snavely, 2012;Guo, 2006;Salamon and Elliott, 2002), and public organizations are increasingly being required to compete with private and not-for-profit organizations (Rashman et al., 2009).
In some cases what appear to be public sector organizations may be subject to even greater market authority than commercial organizations (Bozeman, 2013). Market authority, defined in terms of the level of competition in a market (Nutt, 2005), is assumed to bring higher efficiency, consumer responsiveness, and effectiveness of organizations operating in the public sphere. Market authority derives from the extent of competition and associated buying behaviour of clients to whom services are offered (Nutt, 2005).
Similar to market authority, all organizations are subject to political authority, including those in the private (Moulton, 2012;Moulton and Feeney, 2010), public (Bozeman and Bretschneider, 1994) and not-for-profit (Moulton and Eckerd, 2012) sectors. Political authority is not limited to government intervention, it also encompasses intervention from other stakeholders (e.g., public pressure groups) as a consequence of an organization's position in society and its pursuit of public value (Walmsley and Zald, 1973). Political authority is manifested in more intensive regulatory environments, increased accountability, increased 'red tape', more extensive political oversight and greater political visibility, not just from government but from other external stakeholders (Bozeman, 2013). Directive policy requirements and incentives/disincentives stem from the need to provide public value (Johansen and Zhu, 2013).
Drawing on the Andrews et al. (2011) review and synthesis of the literature on publicness, we focus on three dimensions of political authority: ownership, funding and (political) control. Ownership has been the traditional way of operationalizing the differences between public, private and not-for-profit sector organizations. Theoretical debate about the importance of ownership has centred on property rights theory (Clarkson, 1972;Demsetz, 1967). The traditional argument is that in the private sector property rights align incentives, with shareholders and managers accruing the private benefits of organizational performance. In contrast, property rights are more diffuse in publicly owned organizations, as individual voters have little incentive to monitor publicly owned organizations and managers do not have direct financial performance incentives.
Funding focuses on the organization's source of revenue. Employing a public choice theory of bureaucracy logic, Niskanen (1971) argues that organizations that obtain their revenues from a political sponsor are inherently unresponsive to the people who receive their services. Andrews et al. (2011) note, however, that public sector reforms have strengthened the ability of government to monitor organizational performance where the state is funding the service. Furthermore, governments have sought to increase consumer choice in the delivery of public services to increase organizations' responsiveness to users.
Finally, political control over organizations means that organizations are subject to the priorities set on them by government, which may alter with any change in government (or changing of personnel within government) (Nutt and Backoff, 1993). Rather than being judged on more market-based assessments of performance (e.g., efficiency, effectiveness etc.), performance is measured in terms of fulfilment of political expectations.
The creation of hybrid contexts also means that many government services may now be provided by private, public or not-for-profit sector organizations in competition with one another (Rashman et al., 2009). Further, governmental attempts to blur private, public and not-for-profit sectors has not led to market authority simply replacing political authority. Rather, the dimensions of market and political authority co-exist, sometimes complementing and other times conflicting. For example, ownership, funding and control may be interdependent, and may also exert combined influenced upon the adoption of management practices and performance (Andrews et al., 2011). At times, market and political authority may act as incompatible forces that public service providers struggle to accommodate simultaneously, hampering organizational performance (Boyne, 2003;Currie and Suhomlinova, 2006;Entwistle and Martin, 2005;Entwistle et al., 2007;Vancoppenolle, 2011). Thus, the influence of the contextual dimensions may be less certain than universalist models of management practices suggest. The challenge of the operating in a hybrid context is rendered even greater by the ebbs and flows of emphasis upon political and market authority, which arise from changes in governments and governmental agendas (Addicott et al., 2007;Currie and Suhomlinova, 2006;Currie et al., 2011;Pollitt, 2008;Skelcher, 2000).
Extant studies suggest that managerial practices may vary according to the mix of political authority and market authority, and that dimensions of publicness show a modest effect, if any, on organizational outcomes (Andrews et al., 2011;Heinrich and   . The challenge facing scholars is that extant studies have not incorporated all dimensions of market authority and political authority, which may in part be due to the studies employing correlational and/or regression-based techniques (including contingency based modelling) that are not able to account for the complex interrelationships between multiple dimensions of publicness, management practices and performance simultaneously (Anderson, 2012;Bozeman, 2013). To address the problem of modelling complex relationships we employ a configuration-based approach (see : Fiss, 2007;Greckhamer, 2011;Greckhamer et al., 2008;Schneider et al., 2009), which we outline in the next section.

Hybrid Context: A Configurational Approach
Based on set-theory, QCA is a useful method for developing theories entailing complex causal patterns (Ragin, 1987(Ragin, , 2000(Ragin, , 2008. Many theoretical arguments in management research are best stated in terms of sets and their relations, rather than correlations or net effects (Fiss, 2007). In contrast to conventional regression-based modelling, which treats variables as competing in explaining variation in outcomes, QCA seeks to show how variables combine to create outcomes. Central to the configurational approach are the concepts of conjunction, asymmetry and equifinality. Conjunction emphasizes that outcomes rarely have a single cause but rather results from the interdependence of multiple conditions (Ragin, 2008). Asymmetry allows for the causes leading to an outcome to be different from those leading to the absence of the outcome (Black and Boal, 1994;Miller, 1990;Ragin, 2008). Equifinality refers to a situation where a system can reach the same final state from different initial conditions and by a variety of different paths (Katz and Kahn, 1978). Furthermore, both causal asymmetry and equifinality may be interrelated in that equifinality may change depending on outcome levels; as one moves across outcome levels, different sets of equally effective configurations may arise (Doty et al., 1993;Gresov and Drazin, 1997;Payne, 2006).
The QCA approach enables us to identify commonalities in the form of configurations between theoretically relevant attributes and outcomes of interest (Ragin, 2008). Our interest centres on the set of relations between management practices and the dimensions of publicness, and using QCA we seek to identify the necessary and sufficient conditions that are associated with our outcome of interest, performance (Ragin, 2000). Causal conditions are necessary when the outcome cannot occur without them; whereas causal conditions are sufficient when the outcome occurs with the condition present (Ragin, 1987).
The advantage of employing the configurational approach is that it facilitates the exploration of complex models, however, complex configurational models are difficult to specify a priori. Hence, the configurational approach is well suited to inductive inquiry, in which theory is used first to identify a set of factors to study, and then QCA is used to reveal the empirical configurations of those factors, an approach we follow in this study. Accordingly, we employ the output from the configurational modelling to develop new insight into the conditions associated with the outcome, and how they may be necessary or sufficient for an outcome, based on an organization's hybrid context and management practices.

DATA AND METHOD
Our empirical focus is old people's and children's care homes in the UK. Care homes predominantly provide residential care for older people and children who cannot live with their own family. The two areas of activity constitute clearly demarcated and regulated organizational fields, are populated by multiple for-profit, public and not-for-profit organizations, and constitute the two largest sectors in residential social care in the UK (Gamble, 2008). We collected detailed information on management practices from 204 care homes in England; details of which are presented in Table I.
As our method of analysis does not involve statistical inference or rely on probability theory (Greckhamer et al., 2013;Ragin, 2008;Verkuilen, 2005), rather than obtaining a nationally representative sample, we focused on generating a sample that had sufficient diversity in relation to the ownership form. We randomly selected children's care homes across regions, and proportionally by local authority, to ensure that our sample was representative of the national population. The resulting sample was relatively evenly distributed across the public, for-profit and not-for-profit sectors. For old people's care homes our approach differed as more than 90 per cent of care homes are in the for-profit sector. To ensure that we generated a sample that had sufficient diversity in relation to the ownership form, we over-sampled not-for-profit and public sector old people's care. Our final sample consists of 103 children's care homes and 101 old people's care homes.
Our data collection strategy employed both interviews and the collection of archival data to address the problem of common methods bias. The interviews were undertaken with the managers of the care homes to develop our measurement of management practices and the three components of political authority (ownership, funding and control). Details of the interviews are provided below with description of our measures. In addition, archival analysis was employed to generate our market authority (competition) and performance measures. We describe all variables, measures, and data sources, in the next section.

Measures
Performance. Our measure of performance was gathered from inspection reports provided by regulators, which is a common practice in studies of health and social care (Andrews Notes: While the BAAF is a trade body, it claims to represent more than 90 per cent of all agencies in the industry. In addition, we have confirmed that there is no systematic sampling bias in terms of performance outcomes and other population characteristics, such as number of care home beds. et al., 2010). Both types of care homes are regularly inspected and given a rating by an independent regulator: the Care Quality Commission (CQC) for old people's case of care homes; and the Office for Standards in Education (OfSTED) for children's care homes. The inspections generate a composite indicator, which is an aggregation of underlying performance indicators into a single measure. While conducted by different inspection authorities, both OfSTED (OfSTED, 2009) and CQC (CQC, 2010) employ similar methodologies: developing standards; developing indicators and the data to measure standards against; assessment of performance of providers across a range of metrics; and publication and presentation of ratings publicly. Policy dictates that ratings for all providers come from one 'official' trusted source, and OfSTED and the CQC have a memorandum of understanding which sets out areas in which they will work together, coordinating their roles and activity, in recognition of their similarity of approach to inspection and reporting.
The index maps across a number of the different dimensions of performance identified by Boyne (2002): whether a provider of care is safe, effective, caring, responsive to peoples' needs and well led (CQC, 2010). Each organization is rated on every dimension by the independent regulator, based upon a mix of (external) objective performance indicators, external subjective results from the inspection of services, and service plans and standards. The ratings are then converted to an overall performance score (ranging from 1 [inadequate] to 4 [outstanding]) that is comparable across all organizations.
As a validity check we compared this performance measure with information derived from our interviews with care homes managers, in which we asked respondents to rate the performance of their organization on a scale of 1 to 10 (10 indicating high performance). Following Andrews et al. (2010), we analysed bivariate correlations between our measures of internal and external performance. The magnitude of the correlation between our measure and the internal subjective rating (Spearman's rank coefficient = 0.6) provides evidence for the validity of this measure of performance.

Management practices.
To measure management practices we employed the method developed by Van Reenan (2007, 2011;Bloom et al., 2017) in their global project to assess the effectiveness of management practices across countries and sectors. The tool provides a comprehensive and robust measure of the four practices, and has been applied and validated across a wide range of sectors (including hospitals, retailing and manufacturing) and countries (including Europe, North American and Asia). The method involves two interviewers administering a survey instrument, which includes 18 indicators of management practices, grouped into four subcategories: process management, monitoring, targets and human resource management (see appendix for details). Scoring is based upon a set of open-ended questions designed to allow a trained interviewer to make a reasonable assessment of the quality of management practices in the organization (e.g., 'can you tell me how you promote your employees?'). For each indicator the interviewer scores between 1 and 5, with a higher score indicating a better performance. To ensure that the concept-measure consistency of the indices was high we paid particular attention to this, drawing upon our substantive knowledge of the Bloom and Van Reenan indices. The indices themselves are fine-grained and wellestablished measures of good management practices, which are qualitatively derived and quantitatively scored. While the indices are numeric, they are calibrated in terms of sets, and so lend themselves to being assessed as necessary and sufficient conditions, not as linear scales where all variation is treated as being equally meaningful.
All interviews were carried out in spring 2010, with training on the use of the survey provided by the manager from the original Bloom and colleagues' management survey team. Obtaining interviews with managers followed full ethics approval and was facilitated by a supporting letter from the CQC and in the name of the University. All interviews were independently double-scored by two interviewers: one conducting the interview, the other listening. Any differences in scores were discussed and reconciled at the end of the interview. Each interview lasted 45-60 minutes. All interviews were conducted by telephone and the scoring was not discussed with the respondents. The interviewers were not aware of the organization's performance (or sector) in advance of the interview, so their scoring was not influenced by a priori perceptions of the quality of the organization.
For robustness purposes all 18 indicators were subjected to a principle components analysis. The analysis indicated that there are four distinct factors (as outlined above: operations, monitoring, targets and incentives). We then calculated an overall score for each of the four management practices by taking the average for the scores for each of the separate practice indicators. We then calculated the Cronbach's alpha for each of the four management practices, which all indicated a high degree of internal reliability, as follows: operations (Cronbach's alpha = 0.71), monitoring (Cronbach's alpha = 0.74), targets (Cronbach's alpha = 0.79) and incentives (Cronbach's alpha = 0.70).
Ownership. Consistent with Amirkhanyan et al. (2008), we included an ownershipbased dimension in our measure of political authority, allowing for three categories of ownership: private, public and not-for-profit. Respondents were asked during the interview to indicate which category that best represented their organization's ownership type.
Funding. We defined funding in terms of the extent that the organization is subject to political interference through non-market funding (Nutt and Backoff, 1993). Respondents were asked during the interview to provide the percentage of their organization's total revenue that was derived from government or local government sources (see: Crow and Bozeman 1987;Emmert and Crow, 1988).
Control. There are no direct measures of the degree to which an organization is subject to political control by the government (Anderson, 2012). Indirect measures of political control include the influence over the organization's goals and agendas and/or the degree of influence over its delivery of services (Anderson, 2012). Accordingly, we measured control through the extent to which an organization's client base was deemed to be vulnerable, being drawn from at-risk or disadvantaged groups (Moulton and Feeney, 2010). The logic being that the provision of services for vulnerable groups receives a high degree of government attention (and hence control), as compared to non-vulnerable groups. All organizations were asked to estimate the proportion of their client base that were deemed 'vulnerable', based on the following definitions. For children and young people, we defined vulnerable in terms of whether or not they come from settings where alcohol and drug abuse, or mental health problems of carers are evident. For old people, we defined vulnerability if their background was one of poverty, frailty or cognitive impairment including conditions such as dementia and Alzheimer's disease.
Competition. The markets for care homes are geographically fragmented, and so we employ a spatial definition of competition (see: Propper et al., 2004;Propper et al., 2008). Specifically, we used the fixed radius linear distance approach, which defines markets as circular areas centred on the organization's geographic location. Employing publicly available data from OFSTED and the CQC we measured the level of competition in each market as number of competitors within a pre-determined distance, which we set at 30km radius .
Organizational attributes. We included the size and age of the organization as additional variables (Glisson and Martin, 1980). Unlike econometric regression-based modelling, where additional variables are added to a model as controls to remove their effects from the analysis, there is no such thing as a control variable in QCA. Employing a configurational logic, conditions are included if, and only if, they are considered to be among the paramount causes of the outcome. Hence, QCA requires researchers to balance the need to include conditions that are deemed to be important with additional conditions increasing the complexity of the causal explanations, which may lead to a smaller share of configuration that will be observed (Schulze-Bentrop, 2013). Both size and age are well-established as conditions that may shape organizational performance and in combination with other conditions (Meier and O'Toole, 2002). For example, extant research suggests ownership type and size, in combination, may influence performance (see: Gage et al., 2009;Lemke and Moos, 1989;Zinn et al., 1992). Data on firm size and age were obtained from the CQC, Ofsted and the organizations' webpages.

Calibrating the Measures into Fuzzy Sets
Rather than dealing with ideal types, which commonly describe 'only the extremes' (Smithson and Verkuilen, 2006;Verkuilen, 2005), we employ fuzzy sets. Fuzzy sets require the transformation of the measures into sets that are then calibrated regarding full membership, full non-membership and the crossover point (or point of maximum ambiguity) in the set of interest. Following Ragin (2008), we calibrate our measures using a combination of the indirect for categorical measures and the direct method continuous measures. For the indirect method, while the number of categories for the calibration is not limited, we choose four in that they appeared to be the sufficient distinction in variable (e.g., performance). The choice to calibrate the measure along four categories was based on examining the data and bringing to bear our knowledge of the measures and the context of the inspection authorities. For the direct method we define full membership as 1, non-membership as 0, and the crossover point of 0.5 (i.e., where we cannot tell whether the case is more in or out of the set). The crossover point is the point is qualitatively assessed and anchored as the mid-point between full and non-membership of the set. Our calculation of the crossover points is detailed below, and is designed to reflect meaningful thresholds, which is a significant advantage over the conventional use of indices in regression analysis that tend to be centred on sample means (Fiss, 2011;Ragin, 2008). Full details of the means and standard deviations of our measures and details of the transformed measures are given in Table II. Our performance measure employs a scale (1 = inadequate; 2 = adequate; 3 = good; 4 = outstanding), which we calibrated using the indirect method. Accordingly, we use a four-value fuzzy set, but note that the indirect method is not necessarily limited to four categories (Ragin, 2008). In calibrating the measure, we were aware of the lack of spread  (4) is fully in and given the value of 1; poor performance (1) is fully out given the value of 0; good performance (3) is the point that is more in than out and given the value of 0.67; and adequate performance (2) is the point that is more out than and given the value of 0.33 Operations 3.4 (0.5) We calibrated fully in (good management) above the 90 th percentile; fully out (poor management) below the 10 th percentile; and the crossover point 3 on this measure Monitoring 3.2 (0.5) We calibrated fully in (good management) above the 90 th percentile; fully out (poor management) below the 10 th percentile; and the crossover point being 3 on this measure Target 3.0 (0.5) We calibrated fully in (good management) above the 90 th percentile; fully out (poor management) below the 10 th percentile; and the crossover point being 3 on this measure Incentives 3.1 (0.4) We calibrated fully in (good management) above the 90 th percentile; fully out (poor management) below the 10 th percentile; and the crossover point being 3 on this measure Ownership 1.7 (1.0) We calibrated public sector firms as fully in; private sector firms as fully out; and NFP firms as the crossover point Funding 48.8 (51.2) We calibrated 90% or more funding from government as the fully in; below 10% as the fully out; and the crossover point being 50% Control 34.2 (47.2) We calibrated 65% or more vulnerable clients as the fully in; below 34% as the fully out; and 50% as the crossover point because this marks the difference between those who provided full care for vulnerable clients and those who did not Competition 60.1 (70.2) We calibrated agencies with 250 firms or more in 20 Km radius as fully in; agencies with less than 10 firms in a 20 Km radius as fully out; and agencies with 50 firms in a 20 Km radius is set as the crossover point Size 22.1 (25.3) We calibrated the above the 90 th percentile as fully in; below the 10 th percentile as fully out; and the crossover point as the 50 th percentile Age 25.5 (24.9) We calibrated the above the 90 th percentile as fully in; below the 10 th percentile as fully out; and the crossover point as the 50 th percentile in the regulatory ratings with nearly 60 per cent of the population are rated as good. To address the issue, we drew on our substantive knowledge of the measure, and the linguistic forms of the variable lend themselves to a direct translation into fuzzy sets. The indirect method then proceeded by allocating cases to groups according to their degree of membership of the set; i.e., fully in, mostly in, mostly out, fully out (Ragin, 2000). The different degrees of membership scores are given fuzzy scores; defining outstanding performance (4) as fully in (set to the value of 1), poor performance (1) as fully out (set to the value of 0), good performance (3) as a point that is more in than out (set to the value of 0.67) and adequate performance (2) as a point that is more out than in (set to the value of 0.33).
For the continuous measure of management practices, we employed the direct method of calibration. We set these thresholds based on extant theory and substantive knowledge and utilized the direct method of calibration in the fsQCA software to transform the measures into set memberships (e.g., Fiss, 2011;Ragin, 2008). The work of Bloom and Van Reenan indicates that management practices do vary across firms, tending towards a normal distribution, with a significant tail of very badly managed organizations (Bloom et al., 2014). Our distribution of management scores was consistent with work of Bloom and Van Reenan and colleagues. Hence, we followed the lead of Ragin (2008) in suggesting the use of 90 per cent and 10 per cent cut-off points for those with the best and worst management practices. The 'fully in' (good management) were identified as the 90 th percentile high scoring firms and the 'fully out' (poor management) as the 10 th percentile of low scoring firms. To identify the crossover point, the maximum point of indecision, we followed Ragin (2008) in looking closely at respondents use of language, to identify evidence of indifference in the responses corresponding to the rating at the halfway mark (the score 3). As the crossover point represents neither good nor poor management of firms. We felt confident in the choice of crossover point, given the distribution (note: that the crossover is close to the means in the distribution of scores for each practice, but as outlined above this was not our motivation for choosing it).
We calibrated the dimensions of political authority as follows. First, as ownership is a categorical variable, we employed the indirect method. As over half our participating organizations were private, we calibrated for public sector organizations as fully in, and private sector firm as fully out. As not-for-profit firms are deemed to be more private than public (Amirkhanyan et al., 2008;Heinrich and Fournier, 2004), we set the crossover value as 0.33 (as a robustness check we also ran our analysis employing not-for-profits as a crossover point of 0.5, as a fuzzy measure, which produced the same results). Second, funding was calibrated via the direct method using the percentages with 90 per cent or more funding from government as the fully in, below 10 per cent as the fully out, and 50 per cent as the crossover point because this marks the point of indifference. Finally, control was calibrated via the direct method using the percentages with 65 per cent or more vulnerable clients as the fully in anchor (1), below 34 per cent as the fully out anchor (0) and 50 per cent as the crossover point (0.5) because this marks the difference between those who provided full care for vulnerable clients and those who did not.
For market authority, we calibrated competition via the direct method. Organizations with 250 or more competitors in 30 km radius were classified as fully in, with less than 10 competitors in a 30 km radius as fully out, and with 50 competitors in a 30 km radius is set as the crossover. To convert our variables, calibrated using the direct method, into fuzzy sets we translated the set scores into the metric of log odds using the fQCA software program.

Data Analysis
After calibrating our measures, our analysis progressed as follows, using fsQCA software. The first step was to examine the causal conditions that are sufficient for the outcome to occur constructing a data matrix (or truth table) with 2 k rows, where k is the number of causal conditions used in the analysis. Each row of the matrix is associated with a specific combination of attributes, as well as the number of cases high on these conditions (i.e., with fuzzy set membership scores greater 0.5). We then refined our analysis based on the criteria of frequency and consistency (Ragin, 2008).
We applied a frequency cut-off to ensure that the assessment of the fuzzy set relations occurs only for those configurations considered substantial enough; i.e., exceeding a specific minimum number of cases, which we set at three (see: Greckhamer et al., 2013). We felt comfortable with our choice considering the relatively large number of cases in our sample. Using our frequency cut-off, we were able to perform further analyses on 85 per cent of cases. With our cut-off set at three, we then analysed the configurations with three or more cases for consistency, which we estimated using the consistency coefficient for each configuration for a given outcome. The consistency coefficient is defined in terms of the proportion of cases consistent with the outcome, which is calculated by dividing the number of cases that exhibit the same configuration of attributes and the outcome from the number of cases that exhibit the same combination but do not show the outcome (see: Ragin, 2008). The resulting number is analogous to a fuzzy conditional probability, with numbers closer to 1 signifying a higher likelihood that a configuration leads to the outcome (Verkuilen, 2005). Following Ragin (2006Ragin ( , 2008 a value of 0.80 is considered sufficient to indicate that a given configuration of causal conditions is reliably associated with the outcome of interest. In addition, we followed Ragin (2006) in examining the proportional reduction in inconsistency (PRI) measure, alongside the raw consistency measures. As suggested by Ragin (2006Ragin ( , 2008 the cut-off value for PRI should not be less than 0.75 and higher value is recommended. Accordingly, we looked at the truth table and identified a natural break around 0.79. Using this value, we assigned a positive outcome (a score of 1) to all the configurations with a score of greater than or equal to 0.79 and set the other configurations in the table to zero.
The second step was to examine the complex causal relationships based on fuzzy sets, employing the Quine-McCluskey algorithm to logically reduce or minimize the matrix rows to simplified combinations (see: Ragin, 2008). Each of the final reduced solutions was then evaluated with respect to its coverage of the outcome. We performed the analysis using the Quine-McCluskey algorithm in fsQCA software, where we specified a priori the minimum threshold of the consistency score and our cut-off of frequency of cases per configuration as outlined above. The third step was to evaluate and interpret the results generated in step two. As our approach is inductive, we present the complex solution, which is the minimal solution derived without the aid of any logical remainders (i.e., we exclude possible configurations not actually observed) (Rihoux and Ragin, 2009).
We chose to present the configurations of the complex solution, following Ragin (2008), in assuming that the logical remainders are not sufficient for the outcome; i.e., we exclude logically possible combinations that we do not observe (Ragin, 2008;Rihoux and Ragin, 2009). Given that we have a large sample size, the error of removing logical remainders (less than 10 per cent of the sample) is very much reduced, and we have no clear theoretical reason to include the sample logical remainders (Cooper and Glaesser, 2011).
The complex solution provided us with a good number of configurations. We then assessed the empirical importance of each configuration using a measure of coverage, which is are indicators of the percentage of cases that follow a given pathway to the outcome (Fiss, 2011;Ragin, 2008), and if multiple combinations are sufficient for an outcome. Raw coverage indicates the relative importance of the configurations, and unique coverage indicates which cases are only explained by this solution (Ragin, 2008). We note that the coverage statistics are similar the R-squared in regression (Fiss, 2007).
Following the protocol for large-N QCA studies, in the absence of detailed case information, we examined the robustness of results to the choices made for the calibration of the management measures (Schneider and Wagemann, 2012), the frequency thresholds and consistency levels (Ragin, 2008). For our calibrations we checked several crossover points for the management practice score (2.5, 3 and 3.5), and found that that changes in the crossover points had only a marginal influence on the solutions. Most of the prime configurations were either identical or very similar to the baseline result. For the frequency threshold, as stated earlier, we set the cut-off at three cases due the size of our sample. As a robustness test we also set the frequency thresholds at two and four cases. The results were very stable, but relaxing the thresholds to two did lead to a small increase in the number of solutions generated, but the solutions were all logically substitutable. Finally, for consistency levels, we varied the consistency level whilst holding everything else constant. Unsurprisingly, changing the threshold to a more demanding level (0.85) led to a small decrease in the number of configurations, and a less demanding level (0.72) to a small increase in the configurations. Across all of the robustness checks, we conclude that our analysis is robust as found no meaningful deviation from our presented results.

EMPIRICAL FINDINGS
Our analytical strategy involved assessing causality as a complex set of interrelationships between management practices, political authority, market authority and performance. To do so we need to distinguish between necessary and sufficient conditions (Fiss, 2007). We began by checking for necessary conditions, which appear in all combinations leading to an outcome (Ragin, 2000). To analyse necessity, we use a consistency score (the range being from 0 to 1), which takes a value of 1 when a causal condition is always required for the outcome to occur. As a score of 1 is rare, we follow convention to only consider conditions with a consistency value above 0.9 (Greckhamer, 2011;Ragin, 2006).
We found that all conditions fell below the 0.9 cut-off, and so we have no necessary conditions, confirming that all conditions should be included in the sufficiency analysis. We also conducted a robustness check by testing all the conditions in their original, negated and re-calibrated form in relation to the outcome. In comparing the results of the test, changing the calibration anchors had no effect on the conclusions drawn for necessary conditions.
We now report the results of the sufficiency analysis and the fuzzy set analysis in relation to high and low performance. We look at both high and low performance outcomes because, unlike regression analysis, fsQCA enables us to explore asymmetry, i.e., whether the set of causal conditions associated with the presence of high performance is different from the set of conditions associated with the presence of poor performance. Tables III and IV present the configurations and their statistics that are sufficient for the outcomes of high performance and low performance respectively, employing the notation of Ragin and Fiss (2008). At the bottom of the Tables III and IV the overall consistencies and  coverages scores are presented for the configurations as a whole, as well as raw coverage and unique coverage for each configuration. Overall, our tables show high overall consistency scores, indicating that the sufficient conditions are met, and the causal relations hold for each solution. Also, the overall solution coverage scores in each of the tables are good, being greater than 0.5 (Ragin, 2008). We now seek to understand how the different configurations of management practices, political authority, and market authority, may lead to either high performance or low performance. Starting with Table III, to reduce or minimize the number of configurations of conditions leading to high performance (as outlined above) we set the frequency cut-off to configurations with 3 or more cases, and the consistency cut-off to 0.80, using the Quine-McCluskey algorithm. The overall solution reveals a total of nine configurations associated with high performance, with an overall consistency measure of 0.83, which indicates that all nine configurations should be treated as equally important as each provides a sufficient solution to achieve high performance. In Table IV, we set the frequency cut-off to configurations with 3 or more cases and the consistency cut-off to 0.80, using the Quine-McCluskey algorithm to minimize the number of configurations of conditions leading to poor performance. Here, we found seven configurations that led to low performance with an overall consistency of 0.75.
To explore the profiles of our configurations we grouped them on a taxonomic basis, based on four distinct groups of hybrid context, as defined by the mix of political authority  and market authority. 'Weak hybrid context' is defined by low political authority (across all three components) and low market authority; 'market hybrid context' is defined by high market authority and low political authority (across all three components); 'political hybrid context' is defined by high political authority (in one or more component) and low market authority; and 'strong hybrid context' is defined by high political authority (in one or more component) and high market authority. We employed the taxonomy as our initial baseline interpretation, from which we unpack and explore the findings below.

High Performance Configurations
From Table III (high performance), it is clear that all solutions for high performance require a high condition for either monitoring or target-setting. The weak hybrid context grouping had the highest number of solutions (1a-1d), which reflects the fact there are multiple ways in which a firm can perform well with fewer contextual imperatives. All weak hybrid context configurations for high performance involve a low condition for ownership (i.e., no public sector [1a, 1b and 1d] or are independent of ownership [1b]), with funding and control a low condition for all four solutions, with market authority absent for 1a but low for 1b-1d. Across all four configurations we found that at least one form of management practices is highly present; the findings suggesting that in a weak context, either monitoring or targets can condition for high performance, and may be substitutes for one another in conditioning high performance. All configurations had a low condition for ownership (apart from 1b, where ownership was non-binding), with most of the organizations being private sector. Taking each configuration in turn, configuration 1a requires the high presence of monitoring to condition high performance (raw coverage 0.18) and is associated with the complete absence of market authority and low condition for the three dimensions of political authority. The cases associated with this configuration are predominantly private sector organizations. Configuration 1b shows a causal condition involving the low presence of political and market authority and the high presence of operations and target-based practices that lead to high performance. We note that operations is only present as a high condition in this configuration and is a configuration in which the majority of organizations' ownership type is defined by public ownership. Further, 1c and 1d show that incentives can work in tandem with either monitoring or target-based practices. Here both monitoring and targets seem to be substitutes, when we have low political authority and low market authority. Overall, we note that for weak hybrid context we see a necessary condition for the presence of monitoring or targets, suggesting that they may act as a substitute for one another.
The second group we termed market hybrid context, as defined by high market authority and low scores across all three dimensions of political authority. This group had only one solution (2a). The sample of cases in this configuration is drawn mostly from firms that are non-public sector organizations, being largely private sector (raw coverage = 0.15). The configuration shows that high performance is conditioned by the high level of the three people management practices: monitoring, targets and incentives. This suggests that high performing organizations in contexts characterized by market competition without political authority, are those which have adopted a wide range of good management practices.
The third group we termed political hybrid context, as defined by high political authority (in one or more component) and low market authority. We found two configurations that led to high performance (3a -coverage 0.12; 3b -coverage 0.14). The two configurations differ in that 3a comprises mostly public sector organizations and 3b comprises mostly not-for-profit organizations. Both configurations were conditioned by a low condition for competition, and a high condition for funding, but differed in that 3a was conditioned by low control and 3b by high control. For both 3a and 3b, high performance was conditioned by the presence of both operations and monitoring, which suggests that in the political hybrid context the two practices should be viewed as complements. For configuration 3b, we also found high performance to be conditioned by incentives. We note that operations appears to be more important in the political context, as compared to the weak context and market context, suggesting that operations becomes more important when organizations face a higher degree of political authority.
The final group we termed strong hybrid context, as defined by high political authority (in one or more component) and high market authority. Only one configuration was present for high performance (4a), with a coverage of 0.10, and a limited unique coverage (0.01). The configuration is the least common, but still important, in the overall set of configurations for high performance. The high binding conditions of hybrid context are funding and market competition, with control a non-binding condition. High performance in configuration 4a is conditioned by the presence of high performance across all four management practices. The configuration is comprised of organizations that are largely non-public sector organizations, who face a high and binding condition for both political authority (funding) and market authority (competition). We consider this group of organizations to be particularly interesting, from both an empirical and theoretical point of view, as their context is the most difficult in which to operate (Boyne, 2003;Currie and Suhomlinova, 2006;Entwistle and Martin, 2005;Entwistle et al., 2007;Vancoppenolle, 2011). The configuration suggests that a strong hybrid context requires firms to more fully develop a range of management practices, in comparison to weak hybrid context, political hybrid context and market hybrid context.
In drawing together our findings for high performance we make the following observations. In terms of hybrid context, the most important dimensions in shaping the relationship between context, management practices and performance appear to be funding and competition. Where political authority is high, funding is always a high condition for our high performance configurations. In relation to our four taxonomic groups we draw then following conclusions. First, for a weak hybrid context, monitoring or target-based practices act as substitutes for one another in conditioning high performance. Second, for a market hybrid context, organizations are required to develop a greater range of management practices to condition high performance, specifically, monitoring, targets or incentives-based practices. Third, for a political hybrid contexts, there is a requirement for both operations and monitoring based practices to condition for high performance; and incentives emerge as important in the presence of high public control. Finally, for a strong hybrid context, organizations are required to have the most extensive repertoire of management best practices to condition for high performance.

Low Performance Configurations
We now turn to the six configurations associated with low performance (see Table IV). The group for the weak hybrid context included only one solution (1e: coverage 0.31), with poor performance being conditioned by the low scores on all management practices except operations. In comparing the finding for management practices to configurations 1a-1d, that led to high performance, we highlight the absence of monitoring or target-based practices that appear to be important in driving high performance. Further, a high condition for operations, in the absence of high conditions for other management practices, does not condition for high performance in a low hybrid context.
The market hybrid context configuration consists of one solution (2b: raw coverage 0.17), which is largely drawn from organizations that are non-public sector. The solution involves a high condition for competition. The low performance market context configuration suggests that poor performance is conditioned by the absence of well-developed management practices. When compared to the configuration that led to high performance in a market hybrid context (2a), we suggest that the findings appear to be a near mirror image.
In a political hybrid context, we found three configurations associated with poor performance (3c-3e), all of which are largely drawn from public sector organizations. The solutions all involve the high condition of public funding, and the low condition of competition. All solutions involve either none (3c) or just one management practice, which are operations (3d) and monitoring (3e). When compared to the configurations that condition for high performance in a political hybrid context, which indicates the necessity of a high and binding condition for operations and monitoring, it appears that the presence of fewer management best practices is associated with poor performance.
Finally, we only found one configuration (4b) associated with poor performance in strong hybrid contexts. Configuration 4b involves both high levels of competition and public ownership being a high condition. In this low performance configuration, all four management practices are also low. Again, this finding is the mirror image of the configuration for high performing firms in a strong hybrid context.
In drawing together our findings for low performance we make the following observations. First, we found configurations that led to poor performance across all hybrid contexts, and across all organizational types, as defined by ownership. Second, control is not associated with low performance configurations. Third, there is a low prevalence of management best practices across all six configurations leading to low performance. Fourth, in all but one configuration, monitoring and target-based practices are found to be low.

DISCUSSION
In this paper we have explored the idea of hybridity as context (see: Skelcher and Smith, 2015), motivated by governmental reforms that have blurred the traditional boundaries between sectors with the aim of harnessing the best features of each (Bozeman, 2013;Newman, 2001). To operationalize the idea of contextual hybridity we draw on the concept of dimensional publicness, as defined by the degree to which an organization © 2020 The Authors. Journal of Management Studies published by Society for the Advancement of Management Studies and John Wiley & Sons Ltd is subject to political authority and market authority (Bozeman, 1984(Bozeman, , 1987(Bozeman, , 2013Bozeman and Bretschneider, 1994;Johansen and Zhu, 2013;Nutt and Backhoff, 1993;Walker and Bozeman 2011). To examine the complex relationships that exist between the different dimensions of hybrid context, management practices and organizational performance, we employed a configurational approach (e.g., Bensaou and Venkatraman, 1995;Doty and Glick 1993;Ketchen et al., 1993;Ketchen et al., 1997;Meyer et al., 1993) and method (see : Fiss, 2011).
The key findings from our research are as follows. First, the conditioning effects for high and low performance are conjunctural in nature, in that outcomes result from the interdependence of multiple conditions (Ragin, 2008), although configurations are different across the four taxonomic groups. Second, these configurations demonstrate equifinality, in that there is significant variation in terms of the mix of management practices and hybrid contexts that lead to high performance. Third, there is significant symmetry between our findings for the configurations that lead to high and low performance, with the presence of good management practices being a necessary condition for high performing organizations, while the absence of these practices was a condition for low performing organizations. We expand on each point below.
The conjunctural nature of our results attest to the importance of examining the interaction between an organization's hybrid context and its practices. Our findings show that a hybrid contexts have a material impact on an organization's adoption of management practices, and the coexistence of both contextual characteristics and the adoption of good management practices impacts on performance. Further, our findings attest to the fact that hybrid contexts cannot be reduced to the simple distinction between sectors: there are different sets of hybridity resulting from different combinations of the dimensions of political authority and market authority. We observe that there can be both weak hybrid context, in which there is a balance of low public ownership and low market competition, as well as strong hybrid context, in which both political authority and market authority co-exist. These two forms of hybridity are associated with quite distinct requirements for good management practice in high performing organizations. Two forms of intermediate hybrid context exist in which the market and political authority dimensions are high and low respectively.
We further observe that the organizations operating in a political hybrid context are not uniform, and may face different combinations of the three dimensions of political authority. In turn, these different configurations appear to have implications for which combinations of management practices are associated with high performance. All of this offers evidence that a multi-dimensional approach to address the issue of political authority is helpful to better clarify what 'good management practice' might look like when such blurring of sectors is acknowledged (Bozeman, 1984(Bozeman, , 1987(Bozeman, , 2013Bozeman and Bretschneider, 1994;Johansen and Zhu, 2013;Nutt and Backhoff, 1993;Walker and Bozeman 2011).
These results extend debate about the performance implications of the generic transfer of management practices beyond the binary divide, between private and public sectors, that has been prevalent to date (Andrews et al., 2011;Boyne, 2002;Boyne and Walker, 2010). We now might ask two interrelated questions: 'what management practices?' and 'in what context?'. In doing so we accept management practices are important in the provision of public services, but take a more nuanced approach to their application in what is now a hybrid context for organizations delivering public services.
Our study suggests that we need to get away from a simplified debate that assumes a binary condition of either public sector or private sector in examining management practices and performance. Prior work has suggested that good management practices are important for productivity or other performance measures in manufacturing, education, and healthcare (e.g., Bloom et al., 2014) and indeed much of the 'strategic HRM' literature has indicated similarly that 'more is better' when it comes to understanding the role of best management practices across the entire economy, public and private (e.g., Huselid, 1995;Jackson et al., 2014;Kaufman, 2012). The current evidence extends these observations. We do see some evidence of the 'more is better' hypothesis with respect to management practices, in that regardless of the contextual factors, high performing organizations in this study exhibit systematically more application of best practice than the low performing organizations. The finding appears consistent with prior evidence with respect to the importance of good management for performance (Bloom et al., 2014;Combs et al., 2006;Hayton, 2003;Huselid, 1995;Jackson and Schuler, 1995;Jackson et al., 2014;Prennushi et al., 1997).
It is commonly assumed that the association between competition and the adoption of management best practices should be positive (Bloom et al., 2014). Kaufman (2012), however, argues that since industry concentration results in lower margins the effect of competition may also be to erode employer's willingness to invest in their human resources and associated management practices. Our study provides support for the positive effects of market competition upon the use of good practices, with high performing organizations in the market hybrid context and the strong hybrid context exhibiting greater emphasis on management practices than those in the weak hybrid context with little competition. Our configuration analysis, however, allows us to observe something that might be lost in a simple linear model of the effects of market competition: that is, the absence of market competition does not lead to simple reduction in management practices. We see that even when market competition is low, the presence of different combinations of the dimensions of political authority are associated with different degrees of adoption, which are still more extensive than the weak context cases.
In addition to rejecting a simple binary model, these results also lead us to question the logic in which the political hybrid context and the market hybrid context are two ends of a simple continuum. The results not only demonstrate the need to consider the null case (our 'weak hybrid context') where there is little external authority driving adoption of practices; they also suggest that the end of the continuum is the strong hybrid context in which multiple forms of authority influence adoption of practices. The most complex hybrid environments, rather than the market hybrid or political hybrid contexts, seem to rely upon the most extensive adoption of good practices. Thus, rather than a pure effect of the market conditioning good practice, it is market authority plus political authority, which exhibit the greatest adoption. Rather than the market being the end of the continuum, it is the strong hybrid context which leads the way in the adoption of management. The implication is that in order to fully understand the drivers of adoption of good management practices in any context, it is important to consider the degree of political authority as well as market authority that is exerted upon the organization (Bloom et al., 2014;Simon, 1998).
Taken as a whole, it is clear that there is a positive relationship between the extent of good management practices and enhanced organizational performance, but that the nature of the practices that are required to condition for high performance is linked to the hybrid context and not only the degree of exposure to market conditions. That is, an organization needs to be able to configure their organization such that their set of practices interrelate with context for political as well as for market authority in a positive manner. In a seminal monograph about shaping strategic change, Pettigrew et al. (1992) critiqued as simplistic the debate about the generic transfer of management practices from private to public sector, by arguing that management practices matter, and the absence of performance improvement following their introduction is likely down to their implementation. Our study reinvigorates this debate to consider, 'what management practices, and where?' rather than simply 'are more always better?'.
In returning to our four taxonomic groups we make the following observations. For a weak hybrid context, with low political authority and low market authority, organizations have the least pressure to adopt better management practices. Yet, in high performing organizations we still see the adoption of a subset of practices involving one or other of the combinations of monitoring, targets and incentives. Conversely, in poor performing firms we see a total absence of attention to target setting, monitoring, or incentives. We face a societal challenge in persistent reports of poor care in old people's care homes in the UK, which operate in a less regulated domain of public services, and which may experience less market competition. Our findings study suggest policymakers should attend more closely to management practices in such poorly performing homes (Cooper et al., 2018). Using policy levers to address contextual factors alone, either increase competition and/or political authority through one or more of the dimensions studied here, could just as likely result in poor performing organization operating in market hybrid, political hybrid or strong hybrid contexts. Policy needs to address the core issue of management practices. Further research is needed to understand why organizations fail to adopt the most basic of practices with significant performance consequences. Yet some evidence from across the economy suggests that the principle barrier to the adoption of good practices is that of the skills of managers and leaders (Hayton, 2015).
The market hybrid context is the most similar to traditional markets, as defined by competition and the minimal presence of government. In a market hybrid context we see the strong conditioning effect of market competition leading to the adoption, and performance implications, of good management practices. There are some markets for public services that operate very much as private sector markets. Reduced budgets for adult social care in local authorities in the UK, alongside other policy forces, has effectively privatised old people's care in residential settings. With an ageing population, this has become an expanding and lucrative market in which care home chains have developed, bringing with them a more corporate orientation and associated management practices, which in turn have had a performance enhancing effect. In essence, it seems the weak regulatory regime for older people's care homes, which we note above contributes towards poor quality care (Cooper et al., 2018), may be countered by marketization, but only where appropriate management capabilities exist within and across organizations.
The evidence from our study makes the point that within a particular context, improved management practices can support improved performance. It is important to be clear, however, that the evidence does not suggest that simply changing contexts will automatically improve outcomes. Rather, without attention to management practices, no amount of marketization or political oversight can enhance outcomes.
In a political hybrid context, the autonomy and flexibility of organizations is more limited, being constrained by government mandate and client obligation (Johansen and Zhu, 2013). To date, scholars have suggested that organizations ranked high on political authority are reluctant to adopt organizational changes (Antonsen and Jorgensen, 1997). The low levels of market authority, in combination with high political authority, enable managers to eschew the adoption of newer management practices. Our finding for political authority, however, goes beyond prior studies that are unable to accommodate the complexities of multiple components of political authority (Antonsen and Jorgenson, 1997;Henrich and Fournier, 2004).
Specifically, we found that in a political hybrid context, organizations need to develop specific management practices to condition for high performance, namely operations and monitoring practices, which appear to be complements. The absence of operations and monitoring practices appears to condition for low performance. Government policy 'steers', rather than 'rows' those organizations delivering public services; that is, it sets the ends but not the means of public services delivery (Hood, 1991). Following which, management have discretion in the practices they enact towards high performance. Thus, despite political authority playing a significant role in conditioning an emphasis upon management practices that ensure delivery against government set targets, the contrast between the performing and non performing organizations highlights that there is still room for managerial discretion. The fact that there remain poor performing organizations, which are also characterised by a failure to implement operations and monitoring practices, indicates the potential for policy makers to impact outcomes through policies designed to promote and support better management practice.
The strong hybrid context is one that presents managers with a tension, between the demands of compliance with political authority and the demands of market authority (Boyne, 2003;Currie and Suhomlinova, 2006;Entwistle and Martin, 2005;Entwistle et al., 2007;Vancoppenolle, 2011). The strong hybrid context is the most complex studied here, with high performance being conditioned by the presence of all four management practices. The dimensional nature of political authority and market authority presents tensions that the organization has to address simultaneously. Hence rather than being able to substitute management practices, organizations are required to develop the full range of practices, which may act as complements (Huselid, 1995;Jackson and Schuler, 1995;Jackson et al., 2014).
Our results suggest that some organizations are able to balance and/or resolve contradictions from political and market authority better than others. The ability to manage the contradictions between political and market authority has parallels with the literature on ambidexterity, which relates to managers' ability to manage current demands whilst being adaptable to the changes in the environment (Gibson and Birkinshaw, 2004;Strauss et al., 2015). The low level of coverage observed for high performing organizations in the strong hybrid context suggests that the ability of management to resolve contextual ambidexterity that relates to the inherent tensions of dual authority is relatively scarce (Birkinshaw and Gupta, 2013). In order for organizations to respond to the pressure of becoming more ambidextrous the capability of senior management will need to be enhanced (Jansen et al., 2009;O'Reilly and Tushman, 2008). Also, there will also a need to develop middle manager capability in this respect, with hybrid managers, who have experience and expertise across contexts that exhibit political and market authority, well placed to balance and/or resolve any contradictions (Strauss et al., 2015).

Managerial and Policy Implications
The managerial and policy implications of our work are clear. First, in the face of ongoing debate about generic transfer of management practices from private to public sector (Andrews et al., 2011;Boyne, 2002;Boyne and Walker, 2010), our study cautions organizations against adoption of a universal set of best practices. Instead organizations need to take account of their hybrid context, and the different dimensions of political authority and market authority, in their choices about their adoption of management practices in pursuit of performance improvement (Pettigrew et al., 1992).
Second, from a policy perspective, our study highlights the significance of understanding the effects of the components of political authority and market authority, and how they may jointly and individually influence the achievement policy goals (Lindblom, 2002). We found evidence that the use of good management practices is associated with high performance across all possible configurations of market and political authority, but the groups differ in terms of the configurations and extent of these practices. The policy implication is that absent support and encouragement for the application of improved management practices in any of these hybrid settings, the manipulation of context alone (e.g., through 'marketization' of contexts characterized by political authority) will not produce satisfactory results. We note further, that the high performing organizations in the strong hybrid context grouping were fewer in number (coverage was 1 per cent) than the other groups, suggesting that it is rare for organizations to perform well in strong hybrid contexts due to the difficulty in balancing the tension between high political authority and high market authority.

Limitations and Future Research
As with all studies, ours has a number of limitations that merit further discussion, but also point to interesting new avenues for future research. First, our analysis relies on cross-sectional data, limiting our ability to infer whether and how management practices are altered in response to a particular external constraint or change. Given that government policy, as well as sectoral competitive forces, may change the emphasis placed upon political authority and market authority over time within a given context (Addicott et al., 2007;Currie and Suhomlinova, 2006;Currie et al., 2011;Pollitt, 2008;Skelcher, 2000), we encourage future research to consider the role of changing contexts in influencing management practices and the effect upon performance in longitudinal research designs.
Second, our sample of organizations is drawn from two sectors with vulnerable client group (Burns et al., 2013;Kirkpatrick et al., 2001). Organizations facing potentially conflicting sources of external authority, however, may operate in sectors that vary from refuse collection to public transport. It is important, therefore, to replicate this study across a range of different sectors, with different client groups, to see if our patterns of configurations and our findings generalize to other sectors.
Third, we have drawn upon fsQCA to investigate two outcomes: high performance and low performance. One of the many advantages of fuzzy set-theoretic analyses is that it can accommodate degrees in the focal outcome. Subsequent work should follow Greckhamer's (2011) example and investigate multiple gradations of performance, which enable the further investigation of (a)symmetry across possible configurations.

CONCLUSION
Employing dimensional publicness (see : Bozeman, 1984: Bozeman, , 1987: Bozeman, , 2013Bozeman and Bretschneider, 1994;Johansen and Zhu, 2013;Nutt and Backhoff, 1993;Walker and Bozeman, 2011), and a configurational approach, we have explored the relationship between organizational hybrid context, the adoption of management practices and performance. Our findings indicate that management practices are important in shaping firm performance, but that the volume and nature of the practices required is shaped by the nature of the hybrid context the firm faces. Hence, the assumption that universal management practices can be transferred successfully from the private sector to other sectors only holds under specific configurational conditions. Our study moves the debate about management practices from a binary one, based on ownership boundaries of private and public sectors, towards one that recognizes a continuum of contexts in assessment of the management practices that matter.