Vous êtes sur la nouvelle plateforme d’Érudit. Bonne visite! Retour à l’ancien site


Decentralization in the Public SectorThe Case of the U.K. National Health ServiceLa décentralisation dans le secteur public : le cas du Service national de santé du Royaume-UniDecentralización en el sector público : el caso del servicio de salud nacional en el Reino-Unido

  • James Arrowsmith et
  • Keith Sisson

…plus d’informations

  • James Arrowsmith
    Industrial Relations Research Unit, University of Warwick, Coventry, U.K.

  • Keith Sisson
    Industrial Relations Research Unit, University of Warwick, Coventry, U.K.

The research on which this article is based forms part of the work of the Pay and Working Time project of the ESRC-funded Centre for International Employment Relations Research (CINTER).

Corps de l’article

The decentralization of management authority and, with it, industrial relations, has been one of the most international and seemingly profound organizational developments over the past decade and more (Ferner and Hyman 1998). In the public sector, driven by fiscal crisis, governments have demanded greater cost-effectiveness and efficiency and sought to reproduce the fabled benefits of private-sector competition (Warrian 1996). Towards the end of the 1980s and early 1990s, it seemed as if almost all developed countries were in the grip of a new orthodoxy that demanded, if not the privatization of public services, the imposition of market relationships and tighter managerial control through decentralization (Traxler 1995; Bach et al. 1999; Supiot 2000). This was especially true of countries such as the United States, Australia, Great Britain, New Zealand and some provinces of Canada, where governing parties were influenced by neo-liberal ideology (Goldenberg 1998). However, similar pressures were experienced in traditionally statist countries such as France (Arrowsmith and Mossé 2000) and in the heartlands of the Scandinavian social democratic model (Christensen and Laegreid 2001).

As one of the largest budgetary expenditures, the health sector was often in the van of reform (Pettigrew, Ferlie and McKee 1992). Indeed, the changes introduced into health care systems in the 1980s and 1990s bore many similarities across different countries, though, of course, important differences remain (Saltman 1997; Bach 2000). This internationalized policy process has been referred to as the “Enthoven phenomenon,” after the U.S. economist who popularized the idea of “managed markets” in health care services (Moran and Wood 1996). In this, the U.K. experience serves as a paradigmatic case. The Conservative governments of the 1980s and 1990s systematically tried to shift the National Health Service (NHS) from a traditional European approach, characterized by resource stewardship and professional appeasement, to embrace a decentralized model with U.S.-style cost-consciousness and corporate management. The major change was the transformation of hospitals and ambulance services in the NHS into self-managing trusts in the early 1990s. These “providers” of health care were separated from public authority control, which now became responsible for “purchasing” health care services within an “internal market.” This article examines what the trusts did with their new found “freedoms” in terms of the management of labour, focusing on the pay and working time arrangements of the largest staff groups. The results suggest that little happened in terms of local pay, but there is evidence of changes in approach to working time. These findings have wider theoretical and policy relevance in terms of the limits and possibilities of decentralization.

In the next section, we set the scene by briefly describing the context of the NHS reforms and the literature relating to their effects. Much of this is partial and atheoretical. We then set out our own conceptual and methodological approach. The following two sections present the empirical results in terms of pay and working time. As well as implications for academic research, the final section draws policy implications that are of wider relevance to large public sector organizations.

Context and Research Design

Conservative governments held office between 1979 and 1997 in the U.K. Their policy towards the public services was dictated by a perception of “producer capture” and the need to re-assert managerial authority (Clark 2000). A principal mechanism of this was decentralization. The civil service was fragmented into executive agencies and, in education, schools were granted local management of staff and budgets. It was in the NHS, however, Britain’s largest and most geographically dispersed employer, that the most radical changes were made. Successive initiatives culminated in the 1990 NHS and Community Care Act, which introduced a division and marketization of relations between health care “purchasers” and “providers” in the form of self-governing hospital, community care or ambulance trusts. Each trust employed their staff directly and was entitled to determine their own industrial relations procedures and employment terms and conditions. Although the Labour Government elected in 1997 abandoned the so-called “internal market” and reinforced the national framework within which trusts exercized their “local flexibility,” it retained this key feature of the Conservative reforms (Pollock et al. 2001).

Early expectations, based on plans being developed by trust managers, were that far-reaching changes in patterns of union recognition and pay systems were likely to result from the initial shift to trust status (Corby 1991). This proved to be premature and an alternative view soon emerged that managers would instead concentrate on short-term cost-cutting (Bach and Winchester 1994). Empirical work generally supported this prognosis (Bryson, Jackson and Leopold 1995; Buchan 1994; Lloyd and Seifert 1995; Corby and Higham 1996), especially since there was very limited progress on local pay bargaining or performance-related pay (Bryson, Jackson and Leopold 1995; Bach 1995; Department of Health 1997; NHS Executive 1997; Grimshaw 2000). Pay decentralization was complicated by the Transfer of Undertakings (TUPE) Regulations 1981,[1] which meant that existing staff were entitled both collectively and individually to maintain their national terms and conditions of employment. Many workers preferred the apparent security of the national grade structures and the pay recommendations of the national Pay Review Bodies, especially given union hostility to local pay, and the trusts had little scope to introduce enhanced terms and conditions to induce staff on to new local contracts. Financial constraints and political and contractual uncertainty therefore promoted continuity in human resource management (HRM) and labour relations (Bach 1999).

However, the evidence of this article suggests that there were also significant changes that have been overlooked in the literature, above all in working time arrangements. Following the re-organization, there were concerted attempts by the trusts firstly to develop a general senior level policy role for HR and, secondly, to devolve operational responsibility to new clinical directorates (IRS 1996). This generally enhanced the credibility of the personnel function with local line management (Barnett et al. 1996). As a result, though institutional obstacles remained in the way of local pay, significant changes were able to be made elsewhere.

Part of the reason such developments have received little attention is the nature of the change and the research methods used. Industrial relations research, especially in the public sector, is often focused on issues that are formally bargained and policy-driven. Emergent changes to working-time patterns, which were often highly localized, even individualized, and did not always involve union representatives, were much less visible and controversial than the changes to pay and grading structures proposed in the 1990s. Furthermore, neither snapshot surveys nor individual case studies are sufficient to analyse extensive incremental change at the micro-level. Our research tried to address this by multiple case studies and a detailed survey repeated on an annual basis (see below).

Conceptually, this means we also need an understanding of “human resource management” (HRM) as well as industrial relations, to explore the realities of management decentralization in context. There are many definitions of HRM and the idea continues to be surrounded by controversy (Mabey, Salaman and Storey 1998; Purcell 1999). According to Guest’s (1987) influential model, however, its main dimensions involve the goals of integration (internal coherence; fit with strategic business plans and line managerial activity); employee commitment; flexibility; and quality. Essentially, this contrasts with the bureaucratic or routine servicing (“handmaiden”) style of traditional personnel management to which many large organizations, especially in the public sector, have long been attached (Storey 1992). One of the purposes of management decentralization in the NHS, a huge organization of around a million workers, where staffing accounts for three quarters of all costs, was to encourage the substitution of administrative personnel management with interventionist HRM (Nutley 1999; Ludbrook and Gordon 1999).

To analyse the diffusion of HRM in pay and working time arrangements we use the “firm-in-sector” analytical approach. This draws on contingency and neo-institutional theory to demonstrate how, at the level of the organization, choice is constrained by both internal capabilities and procedures, and the external environment (for more details, see Smith, Child and Rowlinson 1990; Arrowsmith and Sisson 1999; Hislop 2000). In an important article, DiMaggio and Powell (1983) argued that convergence between organizations need not proceed through an evolutionary process of competitive elimination, but that “institutional isomorphism” can occur in three ways: “(1) coercive isomorphism that stems from political influence and the problem of legitimacy; (2) mimetic isomorphism resulting from standard responses to uncertainty; and (3) normative isomorphism, associated with professionalization” (DiMaggio and Powell 1983: 150, original emphases). Significantly, each of these three dimensions is present in the health sector context. NHS trusts remain dependent on the state for funding and are subject to various forms of central policy guidance and control; managerial networks and “benchmarking” are widely used; and managers and workforce alike possess a strong sense of occupational and sector identity.

However, consistency does not necessarily preclude independent change. In contrast to ideas of change as a planned process of leadership, the emergent model of change (see Wilson 1992; Pettigrew and Whipp 1993; Mintzburg, Ahlastrand and Lampel 1998) sees “change as driven from the bottom up rather than the top down, and stresses that change is an open-ended and continuous process of adaptation to changing conditions and circumstances” (Burns 1996: 13). The research hypothesis suggested by the literature and generated by our analytical approach was, therefore, that environmental uncertainty and sectoral tendencies towards isomorphism would likely subvert the centrally-driven policy of local pay, but need not preclude the emergence of bottom-up HR initiatives over working time. In Kanter, Stein and Jick’s (1992) terminology, the “bold stroke” approach to change would be less likely to succeed than the “long march.”

The particular research questions used to examine the impact of management decentralization in a large public-sector organization, focusing on continuity, and change in pay and working time arrangements, were grouped as follows:

  • What was the extent of pay localization? What form did it take? What were the principle obstacles involved?

  • What was the extent of other changes in employment, in particular working time arrangements? What form did the changes take? Why and how were they implemented?

The research involved survey data combined with multiple case studies, to capture the extensive and elaborate nature of pay and working-time change. There are strong technical arguments in favour of a multi-method approach to research, to combine the strengths of different means of data collection (Martin 1990). This is particularly so in industrial relations research, since industrial relations is both an inter-disciplinary and an applied field (Kervin 2000: 539). Our survey was designed to broadly “map” developments in pay and working time by focusing on the main workplace occupational group in a large number of organizations over time. It was not intended for use in building statistical models. This is not just because of problems in identifying and measuring appropriate variables, something that is often hidden in the formalistic presentations of quantitative research (Morgan and Smircich 1980). More fundamentally, our view of reality is that it is much more complex than can be allowed for by the decontextualized application of statistical techniques, particularly where prior knowledge is incomplete. Here, case study material is especially valuable to enrich the survey data and provide insights that would not otherwise be available.

The survey was sent annually, with the co-operation of the NHS Executive, to the HR Director or equivalent in all of the 450 English NHS trusts.[2] In the first year, 1995, 85 useable questionnaires were returned, and 68 in 1998. The response rate possibly reflects some trade-off with the depth of issues covered, and should be borne in mind in the discussion of the results. In 1995 the mean employment in the trusts covered by the survey was 2,239, with a total employment of 190,000; in 1998 mean employment was 2,236 and the total covered by the survey was 150,000. Overall, there are no significant differences between the two samples in terms of key indices such as trust size, region, services, competition or income trends. Some comparisons can therefore be made over time, especially to chart developments within the changing national context of an incoming Labour government (which might have contributed to the fall in response rate). However, for practical reasons, the main focus of the analysis is from 1998. Most questions refer to the largest occupational group in order to closely examine the main issues without making the questionnaire overly complex. For the large majority of respondents in 1998 (87 per cent), the largest group was nurses, including nursing support workers. There were also seven ambulance trusts represented in the sample, reporting on their paramedic and patient transport groups, and two trusts where “scientific and technical” or “support workers” formed the largest workforce group.

The second data source is a series of semi-structured interviews conducted in seven trusts in early 1998, some of which built on previous interviews dating back to 1996. The trusts were selected from the survey returns to broadly represent different types of activity and characteristics, including as well as “mainstream” trusts, trusts identified from the survey responses and in the literature as “innovative” on issues such as performance pay. This approach was designed to capture some broad sense of developments in a range of contexts, rather than a purposive contingency sample. Though conceptually very appealing, this was practically difficult given multiple variables of, for example, trust size, age, region, services, leadership and strategy; especially since the divisionalization of trust organizational structures means there is likely to be increased diversity within as well as between NHS trusts (Ong 1998). The cases were therefore selected to provide a form of “theoretical sampling” rather than any strict notion of representativeness (Eisenhardt 1989: 537; Fincham 1994). The interviews explored the role of the HR function in the trust, and developments in pay, working time, work organization and relations with unions. Five general acute trusts, one ambulance service trust and one community health care trust were visited. Thirteen interviews were conducted with HR directors in each of the trusts and, where appropriate and feasible, local union representatives and other personnel specialists responsible for developing specific trust strategies such as local pay or process redesign. Relevant trust documents such as HR strategies, Reports to the Board and other HR Briefings were also consulted.

Pay: The Limits of Decentralization

At one level, the impact of decentralization is clearly demonstrated by the marked growth in trust-specific employment contracts. At the time of the first survey in 1995, 89 per cent of respondents said that the contractual position of the majority of the largest workforce group was the national (“Whitley”[3]) terms and conditions. Only 11 per cent said that the majority position was one of contracts “shadowing Whitley,” and none reported having most employees on specifically trust terms and conditions. By 1998, Whitley contracts were in the minority. Only 38 per cent had most of their largest workforce group with Whitley. In a further 40 per cent, most staff were employed on trust contracts which “mirrored” Whitley terms and conditions, but which provided for changes to be negotiated on a local basis. In the rest—five of the seven ambulance trusts and around one in seven trusts where nurses formed the largest group—respondents reported that most workers were now on substantially different trust contracts.

Although the growth of trust contracts was reflected in the wider use of job evaluation, which could be used as a building block for local pay, much of this applied only to new grades such as Health Care Assistants or to reorganized services affecting support staff. There was little evidence of job evaluation being used to affect the make up of pay in the larger staff groups. Whatever the main form of employment contract used, virtually all of the trusts in 1998 applied the national pay grades. Likewise, the survey recorded few other departures from national pay arrangements. The timing and level of the 1998 pay award was generally the same, including in different regions, whether or not staff were on local or national contracts. Moreover, few trusts linked the implementation of the award to other changes in pay, working time or other aspects of work organization more generally (table 1). Despite the growth in trust contracts in the intervening period, there were fewer attempts in 1998 than in 1995 to link the pay award to other changes in pay, benefits or work organization.

Table 1

Changes Made Alongside the 1995 and 1998 Pay Increase, per cent

Changes Made Alongside the 1995 and 1998 Pay Increase, per cent

-> Voir la liste des tableaux

Local bargaining was much more evident in the ambulance trusts, reflecting the more homogeneous workforce and distinctive managerial style of the ambulance sector.[4] In the ambulance trusts, four respondents reported that they consulted and negotiated with staff concerning the most recent pay increase and none said that they neither consulted nor negotiated. All negotiations involved a full-time union official, which is a likely indication of their substantive nature. In contrast, only seven hospital or community trusts reported having gone through both consultation and negotiation procedures. A further three had negotiated without prior consultation, and nine had consulted staff but not negotiated locally over the pay award. In the few cases where pay changes had involved some negotiation, only one in three required the services of a full-time union official.

The broad similarity of outcomes in pay and the limited nature of local bargaining in the hospital trusts reflect common financial constraints as well as trade-union opposition to local bargaining and the continued dominance of the national scene. Respondents were asked whether, apart from the outcome of national negotiations, other internal or external pressures were relevant to the decision to make the increase. For the nursing groups, answers were provided by all but five of the “Whitley” groups, all but three of which were “shadowing Whitley” and all except one of those largely on local contracts. The results show that resource considerations were vital to most of each type of trust (table 2). Revenue and cost considerations were important factors regardless of the contractual basis of employees, with those having most staff on independent contracts apparently facing very similar pressures and concerns to those fully committed to the Whitley arrangements. However, various labour market factors were more likely to be an issue in the trusts with mostly local contracts, even if they were as yet unable to have fully developed their own responses in terms of setting pay.

Apart from the pay round itself, the limited impact of local pay was also illustrated by the fact that earnings differences tended to reflect traditional pay components rather than rewards for performance, skills or competencies. As will be shown later, overtime remained a significant factor, as were additional payments for shift or “unsocial hours” working, and enhancements for length of service (table 3). Merit pay and pay related to individual appraisal were reported in only a few cases, reflecting union opposition and practical problems of measuring staff performance for pay purposes (Arrowsmith et al. 2001). In the case of nurses, appraisal pay was found in two of the nine trusts where most were on specific trust contracts (22 per cent), and only one (4 per cent) of those following Whitley and two (8 per cent) of those shadowing Whitley (x2, 2df, p = n.s.). However, only one respondent indicated either merit pay or appraisal related pay to be a “most important” factor explaining earnings differences, suggesting that the sums involved were relatively small. Furthermore, all but one of the respondents with appraisal pay reported that it covered only a small minority of the workforce. In answer to a separate question, there was also only two cases of collective incentives applied to work groups.

Table 2

Relevant Local Considerations in the 1998 Pay Award for Nurses, by Contractural Status of the Majority of Nurses in the Trust, per cent

Relevant Local Considerations in the 1998 Pay Award for Nurses, by Contractural Status of the Majority of Nurses in the Trust, per cent

-> Voir la liste des tableaux

The survey evidence therefore suggested that there had been very little localization of pay, despite the growth of trust employment contracts. In the case study interviews, we explored why this was the case. Of the seven trusts visited in 1998, four had not implemented local pay, outside of senior management, in any significant way. One of these (Acute 1) found its initiatives in this area frustrated by various internal and external pressures, while others (Acutes 2, 3 and 4) had chosen after early consideration to refocus activity elsewhere. One trust (Community 1) had made some progress but had not achieved its ambition to introduce performance pay. The other two trusts (Ambulance 1 and Acute 5) had introduced more or less comprehensive local pay initiatives, the latter being widely regarded as a “flagship trust” in local pay.

Table 3

Factors other than Overtime Explaining Earnings Differences within the Workforce, 1998

Factors other than Overtime Explaining Earnings Differences within the Workforce, 1998

-> Voir la liste des tableaux

A major obstacle to the implementation of local pay was that most of the trusts did not have the financial reserves to bear transitional costs. In addition, contracts for services with purchasers were normally renewed on an annual basis, making the context for planning major change extremely difficult. Nevertheless, in Acute 1, senior managers were initially enthusiastic and joined a consortium of trusts to commission a job evaluation system as a first step in developing local pay arrangements. The application of the scheme to a range of different staff groups proved logistically complex and daunting, however. Other delays resulted from concerns raised by the trade unions, reflecting their national policy of opposition to local pay, and practical difficulties, which the finance department faced in costing out a provisional reward package developed by an HR project team. In the end, the initiative was overtaken by events. Locally, as in some of the other trusts, the question of a merger had been raised, something which the development of a trust-specific pay system could only complicate. Nationally, the uncertainty surrounding the Conservative government’s prevarication over local pay, reflected in its pay restraint policy and its failure to dismantle the national Whitley system, was exacerbated by the prospect of a Labour government pledged to return to national pay arrangements. Faced with competing demands and priorities, there was no real champion of local pay at board level and the issue was eventually put to one side.

The context of practical local difficulties and national uncertainty also led to early plans for local pay being shelved in most of the other trusts. The fact that staff were able to stay with Whitley meant that the local pay package had to be particularly attractive to induce sufficiently large numbers on to local terms and conditions. However, running a significantly different trust package alongside even a minority of staff on Whitley was also recognized as divisive in care terms. The potential costs and complexity of introducing wider moves to local pay was also demonstrated in Acute 3 by a pilot project relating to “ward stewards,” which consumed considerable management time even though it affected only one per cent of the trust workforce. Similar difficulties were experienced by Community 1, where efforts to introduce a “performance element” to the local pay scheme was frustrated by complexities of design and implementation, particularly skills profiling, as well as the changing political context at national level. Even in Acute 5, which had managed to introduce a comprehensive local pay system with integrated scales and performance based progression, there was a feeling of exasperation within the HR team that local pay had consumed a major amount of resources and time without demonstrating significant outcomes in terms of costs, productivity, or staff retention. Such frustrations encouraged HR managers to redirect their energies elsewhere. In the words of one of the management team (a “Remuneration Development Adviser”) from Acute 2:

We thought, what’s the point spending a lot of time working on this sort of thing... our resources are much better channelled down other avenues.… Whereas two or three years ago we were saying what we really want is local pay, a single pay spine, we wanted to evaluate jobs... we didn’t progress that very far. We’ve taken the initiative off pay really and off looking at local arrangements, to look at what people are doing and how they are doing it, and the hours they are doing it for and those kind of issues.

These sentiments were echoed by the HR director of the longest-established trust visited, Acute 4, where an early decision had been made to stick with Whitley despite its acknowledged complexity, rigidities and inequities.[5] Local pay was seen as unnecessarily destabilizing, particularly given the financial structures governing the trust’s operations. Senior management felt that at least some staff would feel threatened by pay localization and, by reassuring them on pay, real gains could be made elsewhere to deliver the trust’s targets on costs, efficiency and quality of care:

It’s important not to get too excited about the amount of local (pay) flexibility you’ve got on paper.… One of the things we are trying to encourage is a feeling of stability and security and not rocking too many boats unnecessarily... I think that local flexibility on pay is, to an extent, a red herring—local flexibility on work organization is what I think is really important.

Not surprisingly, flexibility in working time has long been an important management concern in a service that is both continuous and subject to varying levels of demand. Working time arrangements have also become increasingly important in staff recruitment and retention (Audit Commission 1997). With pressures of local pay receding with the election of a Labour government in 1997, trust HR professionals were arguably better placed instead to help develop initiatives over working time.

Working Time: Squaring the Circle

The survey results identified few changes so far as the basic working week was concerned. In nursing, virtually all trusts in 1998 had a recognized 37.5-hour standard workweek, including every one of those with most staff on trust contracts. In the ambulance trusts, four respondents had a 40-hour basic, two had a 42-hour week, and one had a 39-hour standard working week. There was also still a great deal of overtime working. In 1998, the difference in worked time between the highest and lowest earning full-time workers averaged 15 hours a week. Seventeen respondents also reported that at least some staff typically worked in excess of 48 hours a week, the target “ceiling” of the 1998 Working Time Regulations that implemented the EU Working Time Directive of 1993. In fact, this figure increased from 1995, when only seven trusts reported such regular long hours working, reflecting internal pressures on staffing levels and difficulties in meeting recruitment needs.

There were significant changes nonetheless made to improve flexibility and reduce overtime costs. A number of trusts mentioned shift working as an important recent change to their working time arrangements (table 4). By far the main shift pattern has become a rotating mixture of “earlies,” “lates” and “nights.” In 1998, all seven ambulance trusts used this pattern, with five indicating that it was the most important. Where nurses formed the largest workforce group, this mixed shift pattern was used by 49 of the 59 trusts—up from 1995 when just over half reported its use—and 36 indicated it to be the most important arrangement. Other changes made in recent years were the increased employment of temporary workers and revised annual holidays, usually the consolidation of the NHS’s two extra statutory days holiday into normal leave. However, the most common change, at least for nurses, was an increase in part-time working. This was also seen as by far the most important of all the changes made. Other forms of increased working time flexibility were variants of “annualized hours” contracts. Five trusts had annual hours arrangements for their largest workplace group in 1995. In 1996 and 1997, one in ten trusts reported having annual hours; by 1998, this had increased to one in five, notwithstanding the opposition of the Royal College of Nurses (RCN).[6] There was also an increase in the number with “min-max” arrangements, whereby employees varied their weekly standard hours, rising from around a third of trusts in 1995 to 43 per cent in 1998.

Table 4

Recent Working Time Changes Reported in 1998

Recent Working Time Changes Reported in 1998

-> Voir la liste des tableaux

Respondents were asked which objectives and considerations, apart from developments at national level, had influenced the most recent changes in working time. Achieving greater flexibility and reducing costs were among the main factors. Recruitment and retention were also very important, especially for nurses and for trusts with most on local contracts. Reflecting earlier observations about the most recent changes in pay, respondents in these trusts were more likely to refer to labour market pressures in the development of part-time and shiftwork arrangements. Three quarters of respondents from trusts where most nurses were on local contracts gave recruitment and retention as a very important reason for the most recent changes in working time, compared with two thirds of those which were shadowing Whitley, and 57 per cent of those where most of the nursing workforce was still on Whitley contracts (x2, 4df, p = n.s). In fact, the trusts that had moved away from Whitley tended to lead the changes to shift arrangements and the balance between full- and part-time employees. Around 56 per cent of the trusts with Whitley arrangements had made some recent change to working time, compared to 63 per cent of those shadowing Whitley and nearly three quarters of those in which most nurses were on local terms and conditions (x2, 2df, p = n.s.). The expansion of part-time working in particular was mentioned by 56 per cent with nurses on mainly local contracts, 44 per cent where most were “shadowing Whitley,” but by only 17 per cent of trusts with a majority of nurses with Whitley. This is not because the latter already had a particularly high proportion of part-time nursing staff: trusts still with Whitley had an average 31 per cent of nurses part-time, the same figure as those with the majority of nurses on their own terms and conditions.

The survey evidence therefore suggested that the trusts were trying to reduce the extent and costs of high levels of overtime working by introducing greater flexibility in working time, especially changes to shift patterns, including annualization of hours, and expanding part-time work. The expansion of part-time working in particular seemed to reflect a concern to widen the pool of available labour and improve recruitment and retention. In the case studies we were keen to explore why the changes had been made, and how HR had managed the process, both vertically (through relations with unions and directly with employees) and horizontally (with line management). The significance of changes in shift patterns emerged in the interviews with both our ambulance and nursing trust respondents. The former was a fourth wave trust servicing a large metropolitan area with a total staff of around 1100 including paramedics and transport staff. Historically, drivers worked one of two main shift patterns, 7 a.m. to 7 p.m. and 7 p.m. to 7 a.m. The problem was that these times did not always coincide with demand, typically Thursday, Friday and Saturday nights. Just over a year before the research, after many meetings with both staff and union representatives, management had introduced a range of new shift starting times (4 p.m., 5 p.m. and 6 p.m.), coupled with arrangements to vary the length of the basic working week (42 or 39 hours). The breakthrough in shift patterns also allowed management to make significant progress on the issue of standby. Again, historically, crews waited in the station for emergency calls. Instead of being located exclusively at their base, it was accepted that ambulances could now be positioned nearer to where it could be predicted that accidents and emergencies were more likely to happen. Two factors were important in overcoming initial opposition from the ambulance crews. One was the evidence from the trust’s investment in information technology. This was capable of plotting where ambulances were stationed relative to accident location, which was important in proving the need for stand-by arrangements, and it could predict much more accurately the shifting patterns of demand for services more generally. The other was a willingness to accommodate employee concerns through the process of negotiation with union representatives. Individual stations and employees were not all required to go over to new shift arrangements. In a workforce of around 600, it was emphasized, exceptions could be accommodated. In addition, to meet staff concerns about security, it was also agreed that standbys would be used on a general basis only until 10 p.m. Thereafter, they were restricted to hospitals and healthcare premises.

Changing shift patterns were also strongly in evidence in several of the trusts where nurses made up the largest workforce group. Most significantly, management in three of the acute hospital trusts had introduced shift rotation for the bulk of their nurses. Rather than working a regular pattern of fixed shifts such as “days” or nights,” in the words of one of our respondents, “they all do nights; they all do weekends; they all do days.” In effect, they had achieved a core of nurses working on a rotating basis supported by nurses working part-time at weekends or on days. Crucially, this was managed on a gradual basis and no one was compelled to change. The main benefits cited were the greater flexibility it gave to ward managers in labour scheduling; continuity of care; employee morale, as it was suggested that staff exclusively working nights were felt to enjoy something of a double unfair advantage in terms of less work and higher pay; and improved opportunities for staff training and development, especially for those previously on permanent nights.

The interviews also suggested that the survey might have under-reported the scale of the changes in working time. In the community health trust, there had been experiments with what was in effect an annual-hours arrangement. The HR Director stressed that it could not be called an annual-hours arrangement, however, because of the formal national opposition of the RCN. Yet, in effect, this was more or less to what it amounted to given that the patterns of District Nurses working at night were now more effectively organized around care needs, with one of the incentives for staff being extra leave. It was a new initiative, the HR Director emphasized, that simply could not have been introduced under Whitley arrangements. No one was compelled to go on to the new arrangement, which was agreed on an individual basis with staff. The local union representative had been consulted and in turn she maintained close consultation with the staff. Having no complaints over the proposals and with reassurances over the voluntary approach, it did not become an issue that required a full-time union official to be involved.

As already indicated, often accompanying the introduction of shift rotation for full-time nurses was the encouragement of part-time arrangements. Here, two kinds of flexibility were at work. One was the need of management to get improved cover for critical periods. The other was the need to offer nursing staff “family-friendly” working hours as a means of recruitment and retention. In the case of Acute 2, for example, part-time work was supplemented by “additional hours” through an internal bank as a cost-effective form of working time flexibility. However, it was also viewed as a “family friendly” means of attracting and retaining the “women returner” workforce who could be offered a guaranteed floor of hours with the option of further work at hours to suit. In Acute 5, around 40 workers were employed on contracts that guaranteed them a minimum of 1000 hours per year. These could be called in as required, within certain limits around individual circumstances and commitments agreed in advance. In another of the acute hospital trusts, part-time work was explicitly linked to a “return to nursing” programme to encourage nurses to return to work after having left to have children or for other reasons. This included offering potential returners enough work each year to maintain their nursing registration so that they could eventually return if they wanted.

The fact of increased part-time working shown in the survey results does not therefore do justice to HR efforts to reconcile competing demands and develop a forward-looking approach in collaboration with line management colleagues, to meet operational needs, and through consultation with the workforce, as a “family friendly” initiative. These initiatives were rolled down from trust level HR but were also developed by line managers within the directorates. In one case, both the HR director and the chief nurse made a point of emphasizing that they could not easily describe in detail the patterns of part-time that had emerged because organizing shift patterns was now very much an issue for the ward managers. For example, they argued, some ward managers may have continued with 12-hour shifts, but they would not know without checking payroll records if one or two nurses were involved. Moreover, it did not necessarily follow that that the time would be split equally if two nurses were involved—it could be that one did seven hours and the other five, for example. Interviews with staff representatives confirmed that the arrangement of working time patterns, above all, for part-time staff, had effectively become a one-to-one “negotiation” between individual nurses and the ward managers. Union representatives only got involved if there was a particular problem or if there was a major issue of principle at stake, circumstances that had been the exception rather than the rule for some time.

In their relationships with line management, the HR function enjoyed two advantages. The first was that in a service compartmentalized into separate and often conflicting professional interests, they were often able to credibly present an overall view. The second is that they have a premium of the skills required to make significant changes in working arrangements, helping to extend the HR sphere of influence. Directorate HR managers had a dual role in firstly servicing the line, helping them to develop appropriate responses to recruitment problems or in their efforts to improve labour flexibility in the wards, and secondly in cascading trust level HR policies so that a more formalized HR perspective was brought to bear on the HR activities of line managers. As a result, the distinction between line and HR initiatives was very often blurred in reality. The experience of the HR Director of the second acute trust particularly illustrates this point. How she exercized influence in a highly decentralized system was informative. First, and most importantly, initiatives were always identified as emanating from senior line management regardless of her own input. Second, the personnel officers in each of the directorates acted as agents of central HR. Third, the senior management team was persuaded to introduce what was in effect a “management of change” reporting system to avoid embarrassing inconsistencies across the directorates, and if she was unhappy with anything, she could take it to the management committee. Fourth, she also persuaded the management team to fund a full-time union convenor post, which meant that she had access to a staff voice on most personnel issues before any contentious points could be reached. Several other senior HR managers also argued that a closer integration of HR and line management activity had emerged in recent years, demonstrated by the role the HR function was given in the management of major change. As a result, operational changes in services were more likely to be associated with flexible working time initiatives such as annual hours, part-time work or changes to shift arrangements. In Acute 2, for example, the HR Director assumed responsibility for commissioning a £60 million investment in a new facility. As an HR-led initiative, there was a direct advisory role for HR in the arrangement of services and work organization in the new building. Likewise, in Acute 3, the HR function had a wide-ranging responsibility in change management, in partnership with Finance, of a large-scale “hospital process redesign” (HPR) programme. The aim of HPR is to critically review current processes, such as Accident and Emergency admittance or procurement procedures, with a view to streamlining operations for “patient-focused” care. The HPR initiative in Acute 3, ongoing for a year at the time of the research, was said to have helped make clinicians reconsider how work is organized, facilitating the development and implementation of the trust’s HR agenda. As the HR “Information and Workforce Planning” manager explained:

HPR is turning out to be the catalyst of change within HR itself... when you are going through a process you might be looking at things like annualized hours in Accident and Emergency or in XYZ department. So HPR should be the drive behind a lot of the HR initiatives and be the excuse for a lot of these initiatives to actually be taken up in the different areas... HPR will certainly involve HR implementation, the implementation of a lot of HR initiatives.

Conclusions and Implications

A major objective underlying the creation of NHS trusts was to devolve responsibility for HRM, especially in pay matters, to local level. The failure of local pay was due to contingencies outside of the control of local trust managers relating to the financial and contracting environment and the perpetuation of the national industrial relations frameworks. Senior managers were also consumed with issues such as trust mergers and generally did not want to provoke confrontation with the main staff groups, partly perhaps because of schooling in the traditions of the “model employer” (Kelliher 1998). An important point, however, which has largely been overlooked in the literature on HRM in the NHS, is that these constraints did not necessarily lead to the marginalization of HR within the trusts. Nor did it preclude significant change from emerging, particularly over working-time initiatives in which the role of local HR was enhanced in more subtle ways by developing relationships with line management. Furthermore, these changes were diffused throughout the trusts as a result of HR “benchmarking” and managerial networks.[7] In this sense, the process of isomorphism acted to subvert the top-down initiative of local pay, but tended to disseminate the bottom-up initiatives over working time. This has implications for policy and future research that are relevant beyond the immediate context of the study’s sector and country focus.

The first concerns the focus of academic research and the role of the HR function. A common view of public sector HRM is that it is subservient, cost-driven and slow to introduce change because of a politicized environment that lacks a market dynamic (Oswick and Grant 1996; Boyne, Jenkins and Poole 1999). Our results suggest that this might only be true if the focus is on formal collective bargaining. Trusts were able to develop local initiatives over working time, in particular over shift patterns, flexible working and part-time working, through the collaboration of line management and HR and in direct consultation with staff. These were often developed at ward level and established incrementally. Research might therefore benefit from a wider view of “strategic choice,” which does not necessarily involve the formulation and subsequent directive implementation of “first order” decisions (Purcell 1989). A “processual” perspective takes account of emerging strategy that is not necessarily pre-defined and articulated but develops as managers try to make sense of complex and uncertain situations (Mintzberg and Walters 1985; Pettigrew 1997). This approach could be useful in developing a “firm-in-sector” understanding of the constraints and possibilities offered by decentralization within large organizations.

The second implication relates to policy. Decentralization does not solve problems in itself (Hales 1999). In fact, in pay terms, there is a particular logic in having national rather than local arrangements in a large public-service organization such as the NHS. The reality is that the labour markets for the key professional groups are shaped by the policies of national government rather than local employers. Trade unions are well established and, far from being fragmented by local bargaining, became united by their opposition to local pay. There are economies of scale and fewer transaction costs in having national pay determination arrangements. There is a political desire to avoid an emerging “two-tier” service of local winners and losers. From a managerial point of view, a (reformed) national pay process also has the potential advantage of removing one of the most complex and contentious issues from the local bargaining agenda, freeing managers to focus on meeting the requirements of the service (Sisson 1987). All of this is implicitly recognized in the latest U.K. government proposals on reforming NHS pay, which advocates a simplified national framework (Department of Health 1999). It is a lesson learned the hard way in the U.K. that might be especially useful to policymakers and industrial relations practitioners elsewhere. At the same time, even with the distraction of local pay, decentralization served to stimulate broader change over work organization and working time, in part through new roles for the HR function.

Parties annexes