Beyond hollowing out: Public sector managers and the use of external management consultants

RESEARCH ARTICLE Beyond hollowing out: Public sector managers and the use of external management consultants Ian Kirkpatrick 1 | Andrew J. Sturdy 2 | Nuria Reguera Alvarado 3 | Gianluca Veronesi 4,5 1 York Management School, University of York, York, UK 2 University of Bristol Business School, The University of Bristol UK, Bristol, UK 3 Faculty of Business and Economic Sciences, University of Seville, Seville, Spain 4 University of Bristol Business School, The University of Bristol, Bristol, UK 5 Crown Prince Frederick Center for Public Leadership, Aarhus University, Aarhus, Denmark Correspondence Gianluca Veronesi, Tyndall’s Park Road 11-13 1.08, University of Bristol Business School, Bristol BS8 1SN, UK. Email: [email protected] Abstract Expenditure on management consultants in public sector organizations is generally seen as contributing to a “hollowing out”of the state through the substitution of internal management capability. However, there is little systematic evidence for this view which also ignores how public sector managers may, themselves, drive consulting use. Looking at 125 English public hospitals over 6 years, we explore the relationship between the development and composition of management functions and spending on consultants. Our findings show the absence of a substitution effect and, therefore, challenge the “hollowing out”thesis. Instead, they point to a more active, occupationally varied and political use of consultancy. We find that larger management functions overall are associated with greater reliance on consultants— a complementary relationship. However, where a higher proportion of managers are engaged in internal consulting functions, this results in the lower use of external management consultants, with implications for theory, research and policy. Evidence for practice •The findings challenge the widespread view that external management consultancy use substitutes for in-house management functions. •The use of external management consultancy does not lead to a ‘hollowing out’ of in-house management functions. •Public sector organizations with larger management functions use more consulting advice, suggesting a complementary relationship. •Public sector organizations with more managers involved in internal consulting work tend to make less use of external consultants. “We contend that one of the major reasons for the heavy use of outside experts is the belief prevalent in our society that government …is inefficient, ineffective, and staffed with idle, incompetent, security-seeking individuals.”(Kline, 1979, 226) Debates about the hollowing out of the state are long standing in the public administration literature (Milward & Provan, 2000;Rhodes,1994). Hollowing out denotes a numberofprocessesthatleadtoasmallerandalteredrolefor the state, including the loss of authority to supra national bodies (such as the European Union) and moves to privatize and outsource public services (Skelcher, 2000,5).Most recently, the latter has become increasingly salient in the case of expert advice drawn from management consulting firms (Craft & Halligan, 2020). While initially concentrated in the United States and Westminster-based governments, consultancy use has spread further on the back of New Public Management (NPM) reforms (Saint-Martin, 2012). Globally, it is estimated that consulting revenue from governments exceeded $85bn in 2018 (IBIS World, 2019). The appeal of these services to clients includes the promise of fast and efficient delivery, the acquisition of new skills Received: 2 June 2022 Revised: 26 January 2023 Accepted: 26 January 2023 DOI: 10.1111/puar.13612 This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. © 2023 The Authors. Public Administration Review published by Wiley Periodicals LLC on behalf of American Society for Public Administration. Public Admin Rev. 2023;83:537–551. wileyonlinelibrary.com/journal/puar 537 and reputational benefits, especially when consulting advice is associated with ‘modernization’(Saint-Martin, 2012). However, as the opening quotation from a special issue of this journal, over 40 years ago, suggests, the use of consultants is also controversial. Regardless of the validity of the belief expressed, the quotation illustrates how internal and external expertise are often pitched against each other and management consultancy represents a challenge to the practices and values of public administration (Galwa & Vogel, 2023). As public sector organizations increase their use of consultants, this could be at the expense of developing internal management capabilities. Indeed, the notion that consultants pose a challenge to public managers is widespread in the media, policy and academic research (Ylönen & Kuusela, 2019). Often, this trend in the use of consulting advice is presented in terms of a hollowing out process, not in the dominant sense, linked say to outsourcing of shared services (Milward & Provan, 2000), but specifically in relation to the downsizing of internal management expertise (Althaus et al., 2021; Howlett & Migone, 2013;Lapsley,2009). Momani (2013, 396), for example, concludes that, in the United States, ‘the increased use of management consultants is further evidence of the hollowing of the state’.In Australia, Browne (2021, 17) also points to an ‘over-reliance on consultancy contracts which are a symptom of the hollowing out of the public service’. Specifically, it is claimed that ‘governments have moved to reduce or “hollow out” permanent staff, assuming that external help can be hired on demand’(Vogelpohl et al., 2022, 374). Common to all these accounts is the belief that the greater dependency on the use of external consultants has gone hand in hand with cuts (or caps) on internal management staff numbers, as responsibility for certain key internal functions associated with the planning and co-ordination are effectively outsourced. However, the hollowing out idea in this context is problematic. First, the evidence base is limited to a small number of crosssectional organizational cases or interview-based studies (e.g., Lapsley, 2009), making it hard to generalize. To date, there has been no systematic research showing a link between consulting use and the development of management functions across wider populations of public sector organizations. Second, there exist theoretical problems. In most accounts of hollowing out, managers are assigned a largely passive role, as victims of others’decisions to cut their numbers, but this is clearly an over-simplification. In practice, using consultants is rarely obligatory (Fincham, 1999) so managers are better conceived of as principal agents or clients, who often have considerable discretion over how much consulting advice public sector organizations use and for what purposes. In some contexts, they may oppose using this advice, especially where consultants are perceived as ‘intruders’(Menon & Blount, 2003), posing a threat to managers’ethos, expertise, jobs and prospects or simply as being unnecessary (Ruef, 2002; Sturdy & Wright, 2011). On the other hand, managers could support the use of consulting as a means of introducing new expertise from the commercial sector (Sturdy et al., 2021) and/or to enhance ‘agency reputation’, legitimacy and autonomy (Carpenter & Krause, 2012). Either way, it seems erroneous to assume that the growing ‘use of consultants’will ‘necessarily “hollow out”capacity’(Althaus et al., 2021, 1). Given these concerns, the aim of this article is to critically investigate the notion of a substitution effect. Specifically, it seeks to address the following research questions: (1) what is the relationship between the development of management functions in public sector organizations and the use of external consulting advice; and (2) does the use of consulting advice lead to decreasing investment in managers (i.e., hollowing out)? Answering these questions is theoretically important as a possible corrective to the hollowing out thesis which dominates the public administration literature on this topic. There are also potentially significant implications for policy around governing consulting use and associated costs. This is especially in light of evidence suggesting that rising expenditure on consultants can have negative consequences for performance (Kirkpatrick et al., 2019). The article first draws on the extant literature on consulting and management and highlights a possible relationship between the development of the management function and its influence over decisions about how much external consulting advice is used. Attention is then drawn to the largely neglected issue in consultancy research of how public management is itself occupationally diverse with competing political interests (Sturdy & Wright, 2011). This could shape whether consulting advice is perceived by managers either as a threat (lowering its use) or opportunity (increasing its use). In the main body of the article, these assumptions and related hypotheses are then tested, focusing on the illustrative case of the English National Health Service (NHS). This case is theoretically relevant for two main reasons: (a) a growing annual expenditure on management consulting (Oliver, 2016) and; (b) the relatively low numbers and internally weak occupational status and position of managers (Kirkpatrick et al., 2017). The presence of a weaker occupational group could exaggerate the tension between managers and consultants and the risks of substitution. Conversely, it could mean that managers are strongly motivated to control and promote the use of consultants as a way of extending their own power base and supplementing their expertise. The results of the analysis, focusing on 125 English public sector hospitals over 6 years, provide no evidence to support the hollowing out view. They also suggest that organizations with larger management functions tend to use more consulting advice, while the opposite is true in situations where a greater proportion of managers have overlapping, internal consulting expertise. 538 PUBLIC SECTOR MANAGERS AND THE USE OF EXTERNAL MANAGEMENT CONSULTANTS 15406210, 2023, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/puar.13612 by Universidad De Sevilla, Wiley Online Library on [04/01/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License THEORY While definitions of management consultancy are often debated, there is some consensus over the core roles of providing expertise in the form of advice, support for implementation and, more politically, legitimacy for client decisions (Saint-Martin, 2012;Sturdyetal.,2017). As noted earlier, recent trends have seen a growing uptake of this advice by public sector organizations, which many believe has resulted in a hollowing out of internal management capabilities. Specifically, this implies a substitution effect, whereby certain roles, usually performed by managers, are passed over to external consultants. For example, in UK public health care, a review by the Royal College of Nursing (2009) found that 78% of consulting expenditure was focused on non-clinical projects (e.g., assisting with governance, IT implementation, and public finance initiative contracts) rather than “direct patient care.”Thus, external consultants were mostly covering activities that normally fall under the purview of in-house management functions. Over time, this might contribute to the wider process of hollowing out of the state, as private sector organizations come to play a bigger role in running and shaping public services. The assumed drivers of this localized hollowing out process are complex. For some, it is understood as an outcome of a rational “make or buy”decision, linked to the perceived greater efficiency of external consultancy which offers economies of scale and scope in the provision of standardized advice (Armbrüster, 2006; Canback, 1999). For others, hollowing out of internal management capabilities is more an ideological matter, signaling the growth of a “consultocracy”(Ylönen & Kuusela, 2019), and the preference of policy makers for commercial expertise (Galwa & Vogel, 2023). But whether the push toward hollowing out is driven by rational or ideological considerations (or both at the same time), the end result is assumed to be a substitution effect. As Laage-Thomsen (2022, 472) puts it, “the literature tends to emphasize the replacement of long-term civil servant work with shortterm outsourced expert knowledge production.” However, as mentioned earlier, these assumptions are theoretically problematic for two reasons. First, they fail to acknowledge the active role of managers (the assumed subjects of hollowing out) as clients who often decide the amount of consulting advice that is used. Second and directly related to this, they do not pay adequate attention to the occupational dynamics within the management functions of public sector organizations and how different configurations might either support or resist the use of consulting services. The active role of managers in the use of external consulting When seeking to explain how much consulting advice is sought (the level of demand), an important starting point is role and interests of the “client.”In the public sector, the demand for consulting advice can sometimes be mandated by regulators, politicians, or even NGOs. However, in most situations, consulting is a non-obligatory service, with managers acting (directly or indirectly) as the primary purchasers (Lonsdale et al., 2017). This is even more so in organizations which have greater formal autonomy (with private sector-like governance arrangements) and semi-independent status such as in the case of English hospital trusts (Kirkpatrick et al., 2017). In such contexts, managers typically act as the primary gatekeepers and mediators of consulting use, with consultants themselves in structurally subordinate position as contractors (Waisberg & Nelson, 2018). Essentially, this reflects a principal-agent relationship, with managers as principals and consultants as agents (Sharma, 1997). More specifically, consultants can be seen as the “agent’sagent”or an “extrusion of management power” (Fincham, 2002, 67): while “the agent is no slave to the principal’s (i.e., client’s) interests…still, there is no doubting the basic asymmetry”(ibid., 73). Rather than being passive victims, managers are normally the main arbiters of consulting services, determining how much they are used and for what purpose. Highlighting this active role of managers (as principals) has a number of implications. Firstly, it suggests that the size and power of managers, as a function within organizations, will shape how much consulting advice is used. While only a small minority of managers will act as the direct clients of external consultants, one might expect their decisions to be influenced by the collective priorities of managers as an interest group. It also begs further questions about when and how managers (as a collective interest) are likely to favor or oppose using consulting advice. On the one hand, popular management and media discourse is replete with criticism of consultants, including jokes which imply rivalry and a direct challenge to management jobs and careers (Sturdy, 2009). This is also evident in research findings, identifying management resistance or, at best, wariness (Faust & Schneider, 2014). Managers can also feel threatened by consultants and seek to maintain a distance or “arms-length embrace” (Jackall, 1988, 140). Nevertheless, against this notion of consultants as a threat, other critical accounts point to their role as “servants of power”or “foot soldiers”(Hodge & Bowman, 2006). They offer managers, collectively and individually, the opportunity to enact their policies and, in an informal political sense, enhance their prospects, status, and control (Wilson, 1989). Additionally, managers sometimes use consultants as scapegoats to protect them from difficult or controversial decisions. For example, public managers may try to leverage the legitimacy of external consultants’ assumed independence to signal their own strategic importance to other managers, civil servants, or politicians and advance their individual, departmental, or functional prospects (Abrahamson, 1996). PUBLIC ADMINISTRATION REVIEW 539 15406210, 2023, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/puar.13612 by Universidad De Sevilla, Wiley Online Library on [04/01/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License This does not, of course, mean that consultants are entirely passive. It is well-established how consultants can wield power over the priorities and decisions of clients (Sturdy et al., 2021). Indeed, over time, clients may become dependent on, or even addicted to, consulting regardless of any “objective”need (Ernst & Kieser, 2002). Important here are ‘demand creating’skills of consultants, including the ability to exploit client insecurities and ambition in ways that generate and prolong demand. An example is the practice of consultants translating organizational conditions into problems which can then be solved through pre-existing consulting products. In the terms of the garbage can theory of decision-making, this is a case of “solutions chasing problems”(Sturdy, 2018, 83). Thus, the opportunities that managers perceive from using these services are to some extent co-produced through the agency of consultants as well. Hypothesis development The above observations have important implications for our understanding of the relationship between managers and consultants and debates about hollowing out. First, if managers are viewed as principals, when are they likely to support or oppose consulting use, viewing it as either an opportunity or threat? Second, how likely is it that, over time, the use of consulting advice will act as a substitute for internal management expertise (hollowing out)? With regard to the first question, the answer is likely to be highly context specific, depending on the nature of problems faced and services used as well as individual participants’technical expertise and experience as clients (Althaus et al., 2021; Fincham, 1999). For example, an inexperienced client (or one with limited expertise) might be more likely to see external consultancy as an opportunity (Sturdy & Wright, 2011). Also important here are the occupational characteristics of the management function and how this might shape interests and orientations toward external consultants. While public sector managers can be described as a single function or occupational group within organizations, they are by no means homogeneous (Kirkpatrick et al., 2017; Wilson, 1989). Accordingly, it is possible that sub-groups within the management function will perceive the opportunities and threats posed by consultants differently and that this, in turn, will impact on their support for using their services. Developing this argument further, it can be argued, on the one hand, that managers in general will tend to favor the use of consultants. This follows from the points made earlier about the opportunities that consultants provide for managers, as an occupation or function, to enhance their power base and expertise and/or allay their insecurities. If anything, this is even more likely in contexts such as health care where managers are likely to face continuing challenges to their dominancy from powerful clinical professions, especially medicine (Battilana, 2011). As such, it is possible that managers will perceive the consulting advice and extra help as a means of supplementing their expertise, advancing their collective interests or enhancing “agency reputation”(Carpenter & Krause, 2012). Implied here is that managers, as a whole, will push for the greater use of external consultants and that their ability to do so will be enhanced in situations where the management function itself is larger and (arguably) more influential. Our initial hypothesis, therefore, states that: Hypothesis H1. Public sector organizations with a larger management function will tend to make greater use of external consulting advice. While managers as a generic function may tend to perceive opportunities from consulting use, this is likely to vary between sub-groups. Important for understanding this assumption are the core roles of managers, the nature of their specialist expertise and how far they duplicate or compete with the work of external consultants. Indeed, the thrust of recent work on consultants has been to conceptualize them essentially as an occupation within management, drawing on some similar bodies of knowledge and expertise (Fincham, 2012). This creates the potential for overlap, especially in situations where managers are involved in work focusing on planning and the creation and implementation of new projects. An obvious example are managers involved in “internal consulting,” change management roles or what Mintzberg (1993, 15) loosely terms the “technostructure.”As a sub-group, technostructure managers comprise “analysts…who serve the organization by affecting the work of others.”Although “removed from the operating workflow”or line, these managers “design it, plan it, change it, or train the people who do it”(ibid). In this respect, the activities of technostructure managers may replicate, to a considerable extent, the work of all but the most specialist external consultants—essentially providing similar kinds of advice, expertise, and support (Daudigeos, 2013). Thus, it is possible that a sub-group within the management function associated with internal consulting activities (technostructure) will perceive greater threats and/or fewer opportunities from the use of external consultants. These managers—who themselves often specialize in knowledge development and transfer—are also less likely to be convinced about the superior quality and utility of externally sourced knowledge (Menon & Pfeffer, 2003). This, of course, does not rule out the possibility that internal “consultant-like”managers and external consultants will sometimes work in partnership. However, the overall orientation of internal consulting managers will be likely less favorable toward external consultant use. In practice, this could involve acting as gatekeepers, controlling or resisting the outsourcing of work to external consultants, including actively blocking it or limiting subsequent use, once contracts have been awarded (Sturdy & Wright, 2011). 540 PUBLIC SECTOR MANAGERS AND THE USE OF EXTERNAL MANAGEMENT CONSULTANTS 15406210, 2023, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/puar.13612 by Universidad De Sevilla, Wiley Online Library on [04/01/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License Hence, while managers as a generic function are likely to favor the use of consultants (Hypothesis H1), where a sub-group within management with overlapping expertise and interests is more strongly represented in organizations, one might expect the opposite scenario to apply. Specifically, this leads to a further hypothesis: Hypothesis H2. Public sector organizations where technostructure (internal consulting) managers are strongly represented within the management function will tend to make lower use of external consulting advice. Lastly, the arguments presented so far have implications for the broader notion of a substitution effect or hollowing out process over time. In principle, the growing use of management consulting services can arguably have a directly negative impact on the development of internal management capabilities in situations where consulting use is mandated by regulators. Yet, this seems less likely in those contexts where managers themselves are the primary clients, especially in more independent and semi-autonomous public sector organizations like in the English NHS (Kirkpatrick et al., 2017). Here, managers are likely to have greater control over outsourcing, including scope to use consulting services in ways that further their own interests. In these situations, one would also not expect consulting services to be used in ways that directly undermine the position of managers—substituting their expertise, practices and numbers with the use of outsiders. Accordingly, the final hypothesis predicts: Hypothesis H3. A greater use of external consulting advice by public sector organizations will not lead to a subsequent decline (“hollowing out”) of management functions. METHODS To test our hypotheses, thestudyfocusedontheuse of management consulting services in one area of the UK public sector—the English NHS—and more specifically the acute (secondary) care hospital sector. The NHS is publicly owned, funded and (for the most part) independently managed. Over the past four decades, therehavealsobeensustained attempts to develop management functions and capabilities (The King’s Fund, 2011). However, contrary to media hype, managers in the NHS are relatively few in number: approximately 2% of the total workforce (Veronesi et al., 2019). This, combined with its scale, complexity, and rate of change, makes the NHS a large and expanding market for management consulting services (see Kirkpatrick et al., 2019). We sought to identify and explain patterns of consulting use beyond the individual project or organizational case which are the focus in almost all consulting research. Data sources To undertake this study it was possible, for the first time, to combine three main data sources: a. data on the expenditure of NHS organizations, including on consulting, publicly available in the annual report and accounts of each hospital; b. official statistics relating to the organizational and employment characteristics of NHS organizations, sourced from NHS Digital. This includes the National Workforce Data Set (for employment records), the Hospital Episode Statistics database (for organizational activity), and the NHS Reference Costs Data Set (for tariffs and costs related to treatments); c. a database owned by Wilmington Health care Ltd. which includes a complete breakdown of NHS managers (individuals with decision-making power over budgeting, financial management, and allocation of resources) by job responsibility and organization type. The latest edition (n. 67) used in this study dates from May 2014 with information on 30,000 managers in over 100 job functions. Drawing on these sources, the study employed an unbalanced panel over 6 years (2008/09–2013/14) for a total of 641 hospital trust (hereafter HT) year cases. The final sample represented around three quarters of the total population of hospital trusts. Variables employed Dependent variable For the main dependent variable—the use of management consultants –the analysis focused on the annual expenditure of HTs on “consulting services.”To reduce the influence of outliers and skewness in the distribution of the variable, the use of consulting advice (Lnconsult_exp) was calculated as the natural logarithm of the total amount spent for the hiring of management consultants as recorded in each HTs’financial statement. In accordance with the NHS Manual for Accounts, consulting expenditure comprises any “objective advice and assistance relating to strategy, structure, management or operations of an organization in pursuit of itspurposesandobjectives”(DoH., 2013; Annex 5). This includes activities such as providing advice on the management of services, process management, and rationalization of services (like procurement) and strategic consulting for financial matters and is therefore PUBLIC ADMINISTRATION REVIEW 541 15406210, 2023, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/puar.13612 by Universidad De Sevilla, Wiley Online Library on [04/01/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License consistent with definitions used in other consulting research (Kirkpatrick et al., 2019). Explanatory variables Turning to the main predictors, the analysis focused on the employment of managers overall in each HT and specialist management functions in particular. The first variable accounted for the overall size of managers in each HT (All_managers) and was created by combining the data from the Database of NHS Management and the National Workforce Data Sets (from NHS Digital). It was calculated as the proportion between total number of managers and the total number of full-time equivalent (FTE) employees in each HT. A manual check was conducted to confirm alignment between databases for management occupational codes. Following Mintzberg’s( 1993) definition of technostructure introduced earlier, the analysis also identified roles from the Database of NHS Management which corresponded to the work of advising on, designing, planning, and (sometimes) running the formal systems to achieve coordination. The first variable (Techn1) was obtained as the proportion between the number of technostructure managers and the total number of managersineachHT.Thesecondvariable(Techn2)was calculated as the proportion between the number of technostructure managers and the total number of FTE employees in each HT. As all predictor variables were ratios, we followed best practice by calculating the coefficient of variation for each of the predictors. According to Certo et al. (2020), low levels of dispersion in the ratio of independent variables reduce the likelihood of Type II errors (i.e., false negatives) in the analysis. All coefficients turned out to be well below the threshold value of 1, indicating low dispersion. Control variables Drawing from other research in the sector (Kirkpatrick et al., 2019), eight main control variables were introduced in the analyses to account for factors that might influence the use of external consultants and its relationships with managers. These included controls such as organizational size, measured as the natural log of overnight beds (Size_log); the “busyness”of HTs, in terms of activity levels in relation to patient admissions (Admissions) and occupancy rates (Availability); and structural complexity (Multisites), based on the number of sites or units within each HT. Further controls accounted for differences in the governance of HTs, differentiating between those that had acquired the status of Foundation Trust with greater formal autonomy (Foundtr) and those which had more basic decision-making autonomy (see Kirkpatrick et al., 2017). Related to this, as a sixth control a dichotomous variable (Teaching) was created, to differentiate between (university) teaching and non-teaching HTs. Seventh, the level of competition between HTs in a given area (HHI) was accounted for. This was calculated as the Herfindahl–Hirschman Index of providers’concentration based on the contiguity of HTs in 10 different areas. Lastly, a measure of contracting out (Outsourcing) of non-core activities in each HT such estates and facilities management was included. In addition to these controls, it was necessary to account for the potential influence of past organizational performance (at t-1) on consulting expenditure and the presence of managers in each HT. Two main indicators were used for this purpose. The first (Efficiency) captured the operational efficiency of an organization, proxied by the Reference Cost Index (RCI) (found in the NHS Reference Costs Data Sets). This represents an indicator of (both production and allocative) efficiency of each trust in comparison to the mean level of efficiency for the NHSasawhole(seeVeronesietal.,2019). The second indicator is a traditional measure of profitability—Return on Assets (ROA)—which was calculated as the ratio between the net profit generated and the net value of the assets. In some of the regression analyses, a further measure of efficiency (Adjust_eff) was added (obtained dividing total expenditure by the total number of FTE employees). Lastly, to support the main analysis, two other performance indicators relating to the broad quality of HT services were used. These included the overall patient experience (Experience) (available from the NHS Adult Inpatient Experience Survey), and the safety of the health care provided (Safety). The latter was derived by looking at infection levels in HTs, specifically the rate of C. difficile infection (calculated by the NHS as a rate of infection per 100 bed days). Methodology The empirical approach adopted sought to test the hypotheses while controlling for the possibility of endogenous relationships between the proxies for the presence of managers in each HT and the use of management consulting services. Endogeneity occurs in the presence of omitted variables bias, selection bias, misspecified or erroneous variables, measurement errors, and joint simultaneity (Verbeek, 2008). Here, it was important to exclude the potential existence of a feedback loop between the variables of interest and causation running in both directions, such that the dependent and independent variables are jointly determined leading to spurious inferences. As formulated in Hypothesis H3, one potential outcome of a higher usage of management consultants could be a subsequent decline or hollowing out of management (substituting internal with external management expertise). 542 PUBLIC SECTOR MANAGERS AND THE USE OF EXTERNAL MANAGEMENT CONSULTANTS 15406210, 2023, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/puar.13612 by Universidad De Sevilla, Wiley Online Library on [04/01/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License The empirical model used in the analysis is represented through the following estimation equation: Ln_consultexpi,t¼β0þβ1All_managersi,tþβ2Size_logi,t þβ3Admissionsi,tþβ4Availabilityi,t þβ5Multisitesi,tþβ6Foundtri,t þβ7Teachingi,tþβ8Outsourcingi,t þβ9Efficiencyi,tþβ10ROAi,tþεit, where β 0 is the intercept and β i is the coefficient of each independent variable. Subscript iidentifies the individual, subscript tidentifies the time, and ε it is the error term. To control for endogeneity, a Two-Stage Least Squares (2SLS) regressions (Berry, 1984) was initially employed. Here, the use of instrumental variables corrects the issue of correlation of the explanatory variable with the error terms in a regression model. The instruments need to be highly related to the endogenous independent variable and unrelated to the dependent variable (Larcker & Rusticus, 2010), as well as being determined outside the model and uncorrelated with the error. The procedure involves regressing the independent endogenous variable on its instrumental variable (IV) and all other predetermined parameters in the model (Finkel, 1995), obtaining a reduced form equation through Ordinary Least Squares (OLS) regressions. The coefficient obtained should then be a reliable measure of the impact of managers on consulting use. Three instrumental variables were used: (i) the lag of the dependent variable consultancy expenditure (a customary approach); (ii) the Herfindahl–Hirschman index of concentration (HHI) (under the assumption that higher competition in the catchment area of a HT leads to a heightened pressure to employ managers); and (iii) the total costs (which could logically impact on overall resource availability). The Sargan- Hansen test of over-identifying restrictions confirmed the validity of the instruments, which as a group were exogenous. In the second stage, the surrogate instrument is added to the original equation so that all coefficients and associated standard errors are then estimated in another regression consistently with the assumptions of OLS. As the statistics are calculated on the basis of the original values (and not the predicted ones) of the variables of interest, these steps provide correct standard errors, standardized coefficients and R 2 estimates for the second-stage equation (Berry, 1984). In order to test Hypothesis H3, further tests were performed which also helped to confirm the robustness of the association between managers and expenditure on management consultants. To address this issue, a system of two equations through a Three-Stage Least Squares (3SLS) method was estimated (Zellner & Theil, 1962). 3SLS regressions entail a set of equations with imposed crossequation constraints that produces estimates which are consistent and asymptomatically normal. This method is fundamentally a combination of multivariate regressions (SUR estimations) and 2SLS, where the first equation becomes the general model of the study. By contrast, the TABLE 1 Descriptive statistics Variables Definition Mean SD Min Max Q1 Median Q3 Ln_consultexp Natural log of annual consulting expenditure 6.714 1.144 0.000 11.016 6.054 6.813 7.440 All managers Ratio of total number of managers to total number of FTE employees 0.022 0.009 0.007 0.069 0.016 0.020 0.026 Techn1 Ratio of technostructure managers to total number of managers 0.158 0.048 0.048 0.315 0.125 0.154 0.189 Techn2 Ratio of technostructure managers to total number of FTE employees 0.003 0.002 0.001 0.017 0.002 0.003 0.004 Size_log Natural log of total number of overnight beds 6.410 0.599 3.988 7.694 6.120 6.459 6.768 Admissions Ratio of total number of patient admissions to total number of FTE employees 21.559 5.111 3.097 34.281 18.874 22.213 24.943 Availability Percentage of overnight beds available 0.866 0.062 0.423 0.997 0.835 0.867 0.908 Multisites Number of sites providing care and treatment 5.969 8.867 1.000 72.000 2.000 3.000 6.000 Foundtr Dummy for Foundation Trust status 0.593 0.492 0.000 1.000 0.000 1.000 1.000 Teaching Dummy for teaching trust status 0.185 0.389 0.000 1.000 0.000 0.000 0.000 HHI Herfindahl–Hirschman Index 0.118 0.003 0.115 0.122 0.116 0.117 0.118 Outsourcing Percentage of non-clinical services outsourced 34.552 27.194 0.000 100.000 12.955 27.000 51.900 Efficiency Reference Cost Index 99.170 7.893 78.010 156.630 94.248 98.427 102.393 ROA Return on Assets 0.022 0.382 3.826 4.506 0.032 0.005 0.024 Quality Overall inpatient experience score 76.111 3.374 67.100 88.200 74.100 75.800 77.525 Safety Rate of C. difficile infection for 100 bed days 24.817 15.776 0.000 117.911 14.469 21.043 31.166 Adjust_eff Ratio of total revenue expenditure to total number of FTE employees 67.884 15.479 5.931 146.638 58.648 65.389 74.987 PUBLIC ADMINISTRATION REVIEW 543 15406210, 2023, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/puar.13612 by Universidad De Sevilla, Wiley Online Library on [04/01/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License second equation attempts to explain the determinants of management functions in the presence of other variables. In this equation, the analysis incorporated as explanatory factors of management functions the HHI and two additional measures of performance, capturing the perceived quality and safety of the service provided: patient experience (Experience), and the rate of C. difficile infection (Safety). The assumption here was that the size of the management workforce in HTs could be affected by: (a) competition from other trusts in their area; and, (b) from the performance of their clinical services and the public scrutiny this attracts (Kirkpatrick et al., 2017). In line with the underlying reasoning for adopting simultaneous equations, contemporaneous rather than lagged values of the proxies for management functions were entered in the regressions. TABLE 2 2SLS: The effect of internal management consultants on the use of management consulting services. (1) (2) (3) (4) (5) (6) (7) Variables ln_consultexp ln_consultexp ln_consultexp ln_consultexp ln_consultexp ln_consultexp ln_consultexp All managers 34.432* (19.926) All managers*HSCA2012 12.483 (8.928) Techn1 4.229*** (1.625) Techn1*HSCA2012 0.950 (1.665) Techn2 338.192** (136.972) Techn2*HSCA2012 40.185 (43.785) CEO change 0.044 (0.079) Size_log 0.249 0.293 0.239 0.514*** 0.467*** 0.515*** 0.436*** (0.416) (0.419) (0.449) (0.147) (0.162) (0.134) (0.160) Admissions 0.053** 0.016 0.046 0.024*0.019 0.021** 0.017 (0.024) (0.012) (0.029) (0.014) (0.012) (0.010) (0.011) Availability 0.916 1.297*1.781** 0.521 0.404 0.384 0.379 (0.783) (0.762) (0.844) (0.838) (0.605) (0.608) (0.606) Multisites 0.011* 0.012** 0.012** 0.009 0.009*0.009*0.009* (0.006) (0.006) (0.006) (0.006) (0.005) (0.005) (0.005) Foundtr 0.478*0.487*0.458*0.185 0.227*0.226*0.231* (0.255) (0.250) (0.265) (0.190) (0.127) (0.127) (0.126) Teaching 0.024 0.006 0.028 0.405* 0.411** 0.396** 0.403** (0.469) (0.470) (0.505) (0.234) (0.187) (0.189) (0.186) Outsourcing 0.006* 0.007** 0.007** 0.005 0.005** 0.005** 0.005** (0.003) (0.003) (0.004) (0.004) (0.002) (0.002) (0.002) Efficiency 0.008 0.010 0.011 0.015** 0.016*** 0.016*** 0.017*** (0.007) (0.007) (0.008) (0.007) (0.006) (0.006) (0.006) ROA 0.083 0.052 0.001 0.032 0.040 0.042 0.040 (0.094) (0.094) (0.103) (0.089) (0.082) (0.082) (0.082) Observations 540 540 540 655 656 656 656 R-squared 0.050 0.039 0.106 0.219 0.235 0.230 0.237 F-test 2.09** 2.44*** 2.14** 51.38*** 87.41*** 82.42*** 87.81*** Chi 2 16.82*** 10.82*** 12.59*** Note: Clustered robust standard errors in parentheses. Column 4 reports the regression coefficients of panel data random effects estimations. Columns 5–7 report the regression coefficients of difference-in-difference analysis. ***p< .01. **p< .05. *p< .1. 544 PUBLIC SECTOR MANAGERS AND THE USE OF EXTERNAL MANAGEMENT CONSULTANTS 15406210, 2023, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/puar.13612 by Universidad De Sevilla, Wiley Online Library on [04/01/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License RESULTS Table 1reports the descriptive statistics for the sample employed in the estimations for the period 2008/09 to 2013/2014. In the last year investigated, there were 129 HTs with an average of 125 organizations per year. On average, this sample of HTs each managed around 713 beds, provided care from approximately 6 units and had 4144 FTE employees. The average ratio of managers to staff stood at around 2.2% per HT which is in line with previous research (Kirkpatrick et al., 2017). Technostructure managers comprised on average around 16% of all managers and less than one percent of the whole workforce. HTs spent on average £1,517,670 on consulting advice, with a yearly growth in expenditure across the period of circa 27% which led mean consulting use to almost treble TABLE 3 3SLS: The effect of internal management consultants on the use of management consulting services and causality loop. (1) (2) (3) (4) (5) (6) Variables Ln_consultexp All_managers Ln_consultexp Techn1 Ln_consultexp Techn2 Ln_consultexp 0.009 0.066 0.001 (0.015) (0.210) (0.002) All managers 169.950*** (29.523) Techn1 18.529*** (3.145) Techn2 432.785** (179.374) Size_log 2.615*** 0.004 0.275** 0.066 0.487 0.002 (0.368) (0.010) (0.126) (0.149) (0.468) (0.002) Admissions 0.097*** 0.000 0.022*0.003 0.007 0.000 (0.018) (0.001) (0.012) (0.010) (0.017) (0.000) Availability 1.470 0.001 0.539 0.052 0.260 0.001 (0.916) (0.010) (0.846) (0.140) (0.802) (0.002) Multisites 0.006 0.000 0.003 0.000 0.000 0.000 (0.007) (0.000) (0.006) (0.001) (0.005) (0.000) Foundtr 0.087 0.001 0.066 0.007 0.162*0.000 (0.116) (0.003) (0.107) (0.036) (0.092) (0.000) Teaching 0.188 0.002 0.081 .029 0.287** 0.001 (0.169) (0.002) (0.167) (0.032) (0.134) (0.000) HHI 0.627*10.404** 0.038 (0.344) (5.008) (0.059) Quality 0.000 0.001 0.000 (0.001) (0.008) (0.000) Safety 0.000 0.001 0.000 (0.000) (0.001) (0.000) Efficiency 0.008 0.000 0.027*** 0.001 0.039*** 0.000 (0.009) (0.000) (0.007) (0.003) (0.010) (0.000) ROA 0.141 0.001 0.102 0.004 0.056 0.000 (0.144) (0.003) (0.131) (0.042) (0.114) (0.001) Adjust_eff 0.000 0.001 0.000 (0.000) (0.006) (0.000) Outsourcing 0.007*** 0.010*** 0.007*** (0.002) (0.002) (0.002) Observations 641 641 641 641 641 641 R-squared 0.442 0.250 0.200 1.618 0.089 0.520 Note: Clustered robust standard errors in parentheses. ***p< .01. **p< .05. *p< .1. PUBLIC ADMINISTRATION REVIEW 545 15406210, 2023, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1111/puar.13612 by Universidad De Sevilla, Wiley Online Library on [04/01/2024]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License