Developments in UK executive agencies: Re-examining the ‘disaggregation–reaggregation’ thesis

Introduction

Following the high-profile programme of agencification which the UK government implemented from the late 1980s, the executive agency model has more recently sat outside of the political and academic limelight. With the completion of the ‘Next Steps’ agency creations in 1998; the cessation of the annual Next Steps reports; the abolition of the ‘quinquennial’ review process; and without any government-wide review since 2002, the agency format has, as Gains (2003: 17) predicted, lost the ‘distinct and celebrated identity’ it formerly enjoyed. In academia, compared with the torrent of research pertaining to the first decade of agencification (for instance Gains, 1999; Greer, 1994; Hogwood, 1995; James, 2003), recent work has been rather scant. Latterly, agencies most often form an aside in broader discussions of arm’s-length government (Flinders, 2008; Gash et al., 2010), thereby leaving many questions unanswered. In particular, the long-term evolution of agencies and the agency landscape remain largely uncharted, and as a result, the implications for broader – paradigmatic – developments in UK public administration are poorly understood.

As the Coalition Government embarks upon its programme of fiscal consolidation and democratic renewal, political interest in the agency model has reawakened. Within a year of the coalition taking office, seven new agencies were announced (Jenkins and Gold, 2011), including four in the agency-barren Department for Education. The Public Bodies Review, aimed at reducing the number of non-departmental public bodies (NDPBs), has, in addition, transferred many smaller quangos into existing agencies (Cabinet Office, 2011). The new government sees executive agencies as the ‘default delivery option’ for central government services which require some ‘operational independence’ (Minister for the Cabinet Office, 2011: 22), and it is committed to reviewing their accountability and effectiveness as part of a wider roll-out of ‘autonomous’ organisational forms (HM Government, 2011: 41–2).

Research interest also is beginning to revive (James et al., 2011; James et al., forthcoming). However, at present, the gap in the UK’s agency story has not received sufficient attention, meaning that the incipient academic and political renaissances currently lack context. There is a general consensus that some rollback has occurred, although, in the absence of any systematic review, its magnitude and character remain contested. One influential paper, for instance, argues that ‘from the large-scale disaggregation of the early 1990s policy seems to have almost completely reversed itself, in practice if not in rhetoric’ (Talbot and Johnson, 2007: 55), while another reports only a ‘modest “de-agencification” trend’ (James et al., forthcoming: 10). Accordingly, this article examines changes to agencies and the agency landscape between 1988 and 2010, in order to re-evaluate existing accounts of the UK’s now prolonged courtship of this mode of government administration.

After reviewing the original Next Steps programme, the paper outlines the ‘disaggregation–reaggregation’ thesis. Drawn from the dominant existing account of the ‘post-Next Steps’ period, this posits a substantial policy reversal, evidenced primarily by the return to larger organisational forms. To explore the underlying complexity and multidimensionality of this trend, the article proceeds to explore agency developments along two aspects: ‘structural’, relating to the separation of organisational entities; and ‘functional’, relating to the department–agency task division. The conclusion, in brief, is that the disaggregation–reaggregation thesis has considerable merit when viewed within the structural–functional framework, provided that ‘reaggregation’ is not equated with ‘de-agencification’ and a relapse to a ‘pre-managerialist’ condition. State bureaucracies continue to be heavily ‘agencified’, and it is therefore the evolving manner of this agencification which, going forward, should attract considerable attention.

Executive agencies

The rise of executive agencies has been well documented (Gains, 1999; Goldsworthy, 1991). In the main, the model owes its introduction to the Thatcher Government’s concern that repeated attempts at civil service reform were largely failing to instil a managerial culture into Whitehall. In 1986, the Efficiency Unit examined what further measures were necessary. Its now famous Next Steps Report (Efficiency Unit, 1988) concluded that fundamental changes were needed to the incumbent Whitehall structure, which, perversely, had required officials to deal with policy and parliamentary matters as well as manage complex service delivery. Political pressures invariably led to the subordination of the latter, and thus the report recommended that ‘agencies’ be established as separate delivery organisations, headed by personally accountable chief executives charged with efficient and effective service delivery. Responsibilities and delegations would be formalised in ‘framework documents’, with a presumption in favour of significant operational autonomy (Efficiency Unit, 1988: 9).

The 1990s saw rapid progress towards achieving the agency system envisioned by the Efficiency Unit. By the last days of the Major Government there were more than 130 agencies (James, 2003: 56). It was then announced that the Next Steps programme would end (Chancellor of the Duchy of Lancaster, 1997). Thereafter, the benefits of agencification were broadly accepted by the Labour Government elected in 1997, although a refocusing on collaborative working was sought (Minister for the Cabinet Office, 1999). Thus, while agencies continued under New Labour, a contrasting narrative developed on their role within government (Gains, 2003; see also Schick, 2002).

The ‘disaggregation–reaggregation’ thesis

Talbot and Johnson (2007) argue that the trend of the 1990s towards disaggregating public organisations into multiple smaller units has latterly undergone a significant reversal. This represents a ‘cyclic’ fashion swing back towards ‘big government’. Although they acknowledge that ‘the new big government differs in significant ways from the old’ (2007: 53), the suggestion is nonetheless that some reversal has taken place. Executive agencies are just one instance of this. Several examples of reaggregations are cited as undermining the original agency concept, including: the merger of the Employment Service and Benefits Agency into Jobcentre Plus – a single, much larger agency with questionable autonomy; the development of a single, department-wide framework document for the social security agencies; and the move away from operating on Next Steps lines in the UK tax authorities. Moreover, at a macro level, reductions in the agency population and in the proportion of the civil service in agencies were noted (Talbot and Johnson, 2007: 56). Collectively, these developments signal both a return to larger organisational forms and a departure from the agency model.

Despite its relatively modest empirical basis, this research has proved influential. It remains the leading account of ‘post-Next Steps’ developments, cited by Pollitt (2007: 532) when describing how ‘the fashion has swung towards and then away from the semi-autonomous agency format’, and Halligan (2007: 227), who writes of ‘the reversal of the agencification trend internationally, now most apparent in Britain’. In an earlier paper, Dunleavy et al. (2006: 470) also noted the rollback of agencification. This, they argue, is evidence of a paradigm shift away from NPM. As they explain, contrary to initial expectations, agency coverage stabilised at just over half of the civil service, reflecting a general questioning of the performance of agencies and the advisability of ‘decoupling’ policy systems.

The existing literature thus portrays the evolution of UK agencification as a tale of expansion and decline. Such ‘cycles’ have long been observed in public administration, particularly regarding the perennial centralisation–decentralisation debate (Pollitt, 2005). Nevertheless, until now, little attempt has been made to go beyond high-level headcounts and select examples in order to test more fully the disaggregation–reaggregation thesis.

Before proceeding to such an evaluation here, it is necessary to formulate some conceptual distinctions. Specifically, differentiation is needed between a narrative of agencification–de-agencification on the one hand, and disaggregation–reaggregation on the other. Strictly speaking, the latter refers to machinery-of-government changes that can be achieved through a variety of reorganisation strategies (including de-agencification and agency-to-agency mergers), while the former denotes a specific process whereby functions previously removed from unitary bureaucracies (agencification) are returned to them (de-agencification). This interpretation aligns with the official definition of de-agencification as ‘[the] return of the agency and its functions to within its sponsoring department’ (Cabinet Office, 2006: 8). De-agencification and disaggregation are, therefore, related yet distinct concepts. Any link between them requires empirical verification. Moreover, moving away from purely structural concerns, formal de-agencification is not a requisite for fundamentally altering the interaction of policy and delivery units within the central government policymaking process. Changes to the department–agency task division can modify the process of government without resorting to changes to its structural composition. Overall, then, equivalence between the disaggregation–reaggregation and agencification–de-agencification theses cannot be assumed, but rather is a matter for empirical investigation.

Investigating patterns of disaggregation

In the original Next Steps proposals, the concept of disaggregation featured at two main levels, identified here as the ‘structural’ and ‘functional’ dimensions. The ideas embodied by these two dimensions are, moreover, key to the wider NPM movement. By defining reaggregation along either dimension as observable developments in the opposite direction to that specified for disaggregation, the structural and functional dimensions together form a dynamic framework through which the disaggregation–reaggregation thesis can be tested systematically.

Structural disaggregation and reaggregation

Structural disaggregation involves breaking apart organisational entities, primarily by removing executive functions from unitary departments. In order to chart developments along this dimension, Table 1 presents a chronology of UK agency landscape changes between 1988 and 2010. Yearly figures are provided for the number of agencies established, along with the number of closures. Drawing partly on official literature (Cabinet Office, 2006) and existing research (Massey and Pyper, 2005: 90), dissolutions are further subdivided between four categories: privatisation (by organisational sale or wholesale outsourcing); merger (at the agency level); reintegration (into a sponsoring body); and miscellaneous. The annual net change and year-end total are also detailed.

OPEN IN VIEWER

Table 1.

UK agency landscape ^a

Year	Agency Launches b	Agency Dissolutions					Net Change	Year-end Total
		Privatised	Merged c	Reintegrated	Misc. d	Total Closures
1988	3					0	+3	3
1989	6					0	+6	9
1990	25					0	+25	34
1991	23					0	+23	57
1992	19					0	+19	76
1993	16	1				1	+15	91
1994	12				1	1	+11	102
1995	14	3	4			7	+7	109
1996	31	8	1		2	11	+20	129
1997	12	3				3	+9	138
1998	4		2	1	1	4	0	138
1999	6		6	1	2	9	−3	135
2000	5		3	6		9	−4	131
2001	6		4	3	2	9	−3	128
2002	3		2	2	1	5	−2	126
2003	10		6	1	2	9	+1	127
2004	3		2	4		6	−3	124
2005	6		1	6	3	10	−4	120
2006	5		4	5	1	10	−5	115
2007	6		8	8	2	18	−12	103
2008	6		8	6	3	17	−11	92
2009	3		3	4	1	8	−5	87
2010	2			2	3	5	−3	84
Totals	226	15	54	49	24	142

Several cautionary remarks about the data are needed. First, reflecting the frequently ‘messy’ character of machinery-of-government changes, closure classifications are made according to the dominant direction of transfer for an agency’s functions. For example the majority of HM Stationery Office was sold in 1996, though a small part was retained. This is listed simply as a privatisation. Accordingly, the chronology should be read as an abstraction, useful for analytic purposes, rather than as a definitive register of events. Second, in the absence of any official agency registry for 1988–2010, the chronology’s absolute coverage cannot be guaranteed, although the numbers presented in Table 1 are internally consistent.

Overview

The UK agency population (including the territories and, latterly, devolved administrations) grew in the decade after 1988 to peak at 138 in 1997 and 1998. By the end of 2010, the population fell to 84, with the largest net reductions in 2007 and 2008 (Figure 1). Since 1997, 2003 was the only year to see a net increase (of one agency). James (2003: 57) reports that agency employment (excluding the Northern Ireland Civil Service (NICS) and departments adopting Next Steps principles) reached 285,000 in 1997 – 60% of the Home Civil Service. Adopting the same exclusion criteria, the number of full-time equivalent (FTE) civil servants in agencies in 2009–10 totalled approximately 303,000 – 62% of the Home Civil Service. ¹ If the NICS is included, the figure is 315,050. Together, the five largest agencies employed nearly 207,000. The smallest agency has just 38 FTE.

OPEN IN VIEWER

Figure 1.

UK agency population, 1988-2010.

Scotland

Closely paralleling political events, Scotland saw no major rationalisation until 2008. In 2001, the Scottish Executive proposed reducing the number of NDPBs by returning functions to either the core government or an executive agency (Scottish Executive, 2001: 6). Agencies were painted favourably since they allow ‘a strong emphasis on the delivery of outputs’ while remaining ‘an integral part of the Executive’ (Scottish Executive, 2001: 18–19). Across 2001–06, the number of Scottish agencies duly rose to 18 (Table 2). However, with the change of administration in 2007 came renewed concern that Scotland had too many disparate public bodies. A ‘Simplification Programme’ was launched, and agencies were no longer immune to the cull. Indeed, a stated aim was to ‘[enhance] links between policy development and delivery’ by bringing these together in government (Scottish Government, 2009: 12). Six agencies were reabsorbed in 2008 and 2009, although two new agencies were also designated. By the end of 2010, Scotland had 13 agencies, indicating that the extent of de-agencification remained limited.

OPEN IN VIEWER

Table 2.

Agencies in Scotland

	1990	1991	1992	1993	1994	1995	1996	1997	1998	1999	2000	2001	2002	2003	2004	2005	2006	2007	2008	2009	2010	Total
Launches	1	2	1	3	1	1	0	1	0	0	0	2	1	1	1	2	1	0	1	1	0	20
Reintegrations																			3	3		6
Misc.																					1	1
Net Change	+1	+2	+1	+3	+1	+1	0	+1	0	0	0	+2	+1	+1	+1	+2	+1	0	−2	−2	−1
Total Agencies	1	3	4	7	8	9	9	10	10	10	10	12	13	14	15	17	18	18	16	14	13

Wales

Wales stood apart during Next Steps, only gaining one agency of its own. Three further Welsh agencies were launched after devolution, although by 2005, three had also been lost (Table 3). Despite the rollback in Wales predating that in Scotland, their agency stories display several similarities. In particular, no agency-to-agency mergers have occurred, making de-agencification the dominant reaggregation mechanism. Furthermore, in both countries, the changes occurred against a wider backdrop of ‘quangocracy’ reforms (for details of this in Wales, see Thomas, 2004).

OPEN IN VIEWER

Table 3.

Agencies in Wales

	1991	1992	1993	1994	1995	1996	1997	1998	1999	2000	2001	2002	2003	2004	2005	2006	2007	2008	2009	2010	Total
Launches	1	0	0	0	0	0	0	0	0	1	0	0	2	0	0	0	0	0	0	0	4
Reintegrations													1	1	1						3
Net Change	+1	0	0	0	0	0	0	0	0	+1	0	0	+1	−1	−1	0	0	0	0	0
Total Agencies	1	1	1	1	1	1	1	1	1	2	2	2	3	2	1	1	1	1	1	1

Northern Ireland

Agencification in Northern Ireland was conditioned by the region’s socio-historical and political particularities, including the existence of a separate civil service (NICS). By 1998, 25 agencies had been established: three for the UK government’s Northern Ireland Office; and 22 as part of the NICS (Minister for the Cabinet Office, 1999: 252). Unlike in Scotland and Wales, net agency reductions began soon after 1999, though the decline was not dramatic (Table 4). Between 2003 and 2005, the population stabilised at 22, before a second period of reaggregation followed the Review of Public Administration (RPA, 2006). This came in response to accusations that Northern Ireland was vastly over-administered (Knox and Carmichael, 2006), and resulted in a major rationalisation through a combination of mergers and reintegrations, leaving 13 agencies by the end of 2010.

OPEN IN VIEWER

Table 4.

Agencies in Northern Ireland

	1990	1991	1992	1993	1994	1995	1996	1997	1998	1999	2000	2001	2002	2003	2004	2005	2006	2007	2008	2009	2010	Total
Launches	1	2	3	3	0	5	10	0	1	0	0	0	0	1	0	0	0	2	0	0	0	28
Mergers																		4	2			6
Reintegrations											1		2				2		1	1		7
Misc.													1					1				2
Net Change	+1	+2	+3	+3	0	+5	+10	0	+1	0	−1	0	−3	+1	0	0	−2	−3	−3	−1	0
Total Agencies	1	3	6	9	9	14	24	24	25	25	24	24	21	22	22	22	20	17	14	13	13

UK central government

As well as highlighting the different evolutionary patterns in Scotland and Wales (which opted for de-agencification) and Northern Ireland (which implemented a mixture of de-agencification and mergers), the removal of the territorial agencies clarifies the picture for UK central government (Table 5). Closure by merger is still the most prominent category, accounting for 48 dissolutions since 1988, 43 of which came after 1997. By contrast, the number of reabsorptions falls to 33, reflecting the prevalence of de-agencification elsewhere. Miscellaneous closures sum to 21.

OPEN IN VIEWER

Table 5.

Agencies in UK central government

	1988	1989	1990	1991	1992	1993	1994	1995	1996	1997	1998	1999	2000	2001	2002	2003	2004	2005	2006	2007	2008	2009	2010	Total
Launches	3	6	23	18	15	10	11	8	21	11	3	6	4	4	2	6	2	4	4	4	5	2	2	174
Privatisations						1		3	8	3														15
Mergers								4	1		2	6	3	4	2	6	2	1	4	4	6	3		48
Reintegrations											1	1	5	3			3	5	3	8	2		2	33
Misc.							1		2		1	2		2		2		3	1	1	3	1	2	21
Net Change	+3	+6	+23	+18	+15	+9	+10	+1	+10	+8	−1	−3	−4	−5	0	−2	−3	−5	−4	−9	−6	−2	−2
Total Agencies	3	9	32	50	65	74	84	85	95	103	102	99	95	90	90	88	85	80	76	67	61	59	57

Language evoking the ‘joined-up government’ agenda has often rationalised central government mergers. The broad idea, which surfaced early in New Labour’s term of office, was that coordination in public services needed to be restored following the fragmentation engendered by managerialist policies (Cabinet Office, 1999a; see Bogdanor, 2005). For agencies in particular, it was suggested that ‘An issue for all … to consider is whether there is scope for improving performance by cooperation with bodies beyond the agency boundary’ (Minister for the Cabinet Office, 1999: v–vi). Different tools for improving horizontal (agency-to-agency) coordination require different degrees of organisational change. For example, James (2003: 136–40) distinguishes between ‘sporadic links’, based on voluntary agreements between organisations, and ‘networks’, which involve ‘more enduring partnerships’ and which, taken to the extreme, result in mergers. Similarly, Dunleavy (2010: 16–19) posits a seven-stage model for joining up services, the last stage being ‘mergers, take-overs or integration’. Thus, changes along the structural dimension, as posited here, represent more dramatic forms of coordination-improvement mechanisms.

Although there has been no central policy seeking mergers specifically on this basis, ideas of ‘joining-up’ often accompanied such developments. The formation of the Maritime and Coastguard Agency, for example, allowed ‘synergies’ between the separate Coastguard Agency and Marine Safety Agency to be exploited (Minister for the Cabinet Office, 1999: 141). More recently, the amalgamation of the Pensions Service and the Disability and Carers Service was said to ‘[provide] more holistic services to those receiving pension age benefits’ (Disability and Carers Service, 2008: 4). It thus appears that, in attempting to mitigate the fragmentation brought about by the agency system in the 1990s, considerable structural upheaval to the agency landscape was achieved – along with diminished task-specificity – without, however, reversing the broad policy of agencification.

Of the 33 departmental reintegrations, 24 were in the Ministry of Defence (MoD). During the 1990s, this department thoroughly embraced the agency model (Hogwood, 1995). Throughout the following decade, however, the population steadily declined – initially through mergers, but increasingly through reintegrations – from its high point of 44 agencies, to just nine by the end of 2010. Questioned in 2006, an MoD official offered a quasi-transaction cost rationale for this pronounced de-agencification, reporting a concern with ‘the overheads that go with a very decentralised structure’, and a realisation that ‘the decentralising tendency in the Department went perhaps a little too far in the 1990s’ (see House of Commons Defence Committee, 2006: 19).

Looking to other departments, there have been no similarly coherent programmes of agency rollback. Most non-MoD reintegrations took place before 2001 and involved low-profile bodies from various departments. Therefore, in contrast to Scotland and Wales, ‘de-agencification’ in Whitehall has been restricted to a particularly agencified department and some tidying in the wake of the rapid initial roll-out of agencies.

As for miscellaneous closures, many of these involved agencies attaining other arm’s-length classifications, either individually (entailing continued disaggregation), or as part of a merger (sub-departmental reaggregation). These changes do not constitute de-agencification proper; their functions remain at ‘arm’s length’ – or even at greater arm’s length – so that, while diminishing the agency population, the effect on the overall sphere of delegated governance is minimal.

Functional disaggregation and reaggregation

In turning to the second component of the analytic framework, the evaluation of the disaggregation–reaggregation thesis proceeds beyond the terms of the argument expounded by its original authors, looking not only at organisational boundaries, but also at the nature of the department–agency task division. Although, officially, Next Steps sought the institutional separation of policymaking and operations, existing research has highlighted the practical difficulties experienced in trying to achieve this (Greer, 1994; Hogwood, 1995). A review in the mid-1990s, for example, found that ‘half of agencies have policy functions either resulting from their framework document … or de facto’ (Trosa, 1994: 3). This calls into question the extent to which functional disaggregation was originally achieved, and thus whether any subsequent reversal can now be measured. Nevertheless, the Labour Government’s early discourse on agencies focused largely on the problems engendered by a policy-implementation dichotomy in government. In particular, the ‘vertical’ aspects to the joined-up agenda were predicated on the idea that a detrimental schism had arisen which sat awkwardly with the new, holistic vision of policymaking:

‘Whilst organisational and management changes over the past decade have emphasised the separation of policy making and policy implementation, “modernised” policy making demands that they be re-integrated into a single, seamless, flexible process’ (Cabinet Office, 1999b: 9).

Labour’s Agency Policy Review similarly found that ‘the gulf between policy and delivery is considered by most to have widened’ (OPSR and HM Treasury, 2002: 5). Thus, although not achieved universally, functional disaggregation appears to have been implemented sufficiently well during the Next Steps era to have subsequently concerned advocates of joined-up, ‘seamless’ policymaking. Moreover, Labour’s search for this new policymaking ideal within its broader ‘modernisation’ programme (Cabinet Office, 1999a, 1999b) implies that functional reaggregation was actively sought during their term of office.

In order examine the extent of functional (dis-/re-)aggregation under Labour, agency framework documents have been examined to establish how agencies are involved formally in the policymaking process. Although these are, at best, only indicative of practice, they provide a useful account of the changing ‘official’ image of agency policy activities.

Given the political nature of New Labour’s ‘modernisation’ project, the sample is confined to UK central government. The most recent frameworks for 47 of the 58 agencies operating at the end of the New Labour Government in 2010 have been examined and classified according to the level of policy responsibilities devolved to the agency (Table 6). Most documents were authored by Labour but, reflecting the declining salience of agencies, some predate the 1997 election. Of those agencies that were not analysed, five belong to the Department for Transport’s Motoring and Freight Services Group, which now operates a single framework. The ‘grouping’ of agencies in this manner was actually intended to ‘strengthen the role and position of the agencies, ensuring they are fully engaged in shaping as well as delivering policy’ (Department for Transport, n.d.) – in itself, a clear attempt at functional (and quasi-structural) reaggregation. However, this multi-agency framework proved unsuitable for inclusion in the main analysis. For the remaining six unclassified agencies, appropriate frameworks could not be sourced.

OPEN IN VIEWER

Table 6.

Formal policy responsibilities of UK central government agencies

Agency undertakes significant levels of policy formulation and development work		Agency undertakes periodic policy development work, either assisting the department with the formulation of ministerial policy advice, or being consulted on operational aspects of policy	Agency has no formal policy responsibilities
Agency leads on policy and/or legislative matters	Agency and department share policy duties (agency may lead in certain areas and/or is a key ministerial advisor)
FERA (2009) Identity and Passport Service (2009) National Fraud Strategic Authority (2008) National Measurement Office (2010) Office of the Public Guardian (2009) UK Intellectual Property Office (2008)	Defence Storage and Distribution Agency (2003) Defence Support Group (2008) Defence Vetting Agency (2006) Dstl (2006) Forest Research (2003) MHRA (2004) NOMS (2008) Ordnance Survey (2004) Service Children’s Education (1996) Skills Funding Agency (2010) The National Archives (2009) UK Border Agency (2009) UK Debt Management Office (2005) UK Hydrographic Office (2007) Valuation Office (2009) Veterinary Medicines Directorate (1995)	Asset Protection Agency (2009) Department for Work and Pensions (2008) (encompassing Jobcentre Plus and Pension, Disability and Carers Service) Highways Agency (2009) HM Courts Service (2008) Planning Inspectorate (1998) Rural Payments Agency (2005) Service Personnel and Veterans Agency (2007) Treasury Solicitor’s Department (2008)	Central Office of Information (1998) CEFAS (2008) Companies House (2009) Criminal Records Bureau (2003) FCO Services (2008) Forest Enterprise England (2003) Land Registry (2008) Met Office (2007) Ministry of Defence Police and Guarding Agency (2006) National Savings and Investments (2009) People, Pay and Pensions Agency (2008) Queen Elizabeth II Conference Centre (2003) The Royal Parks (1993) Tribunals Service (n.d.) Veterinary Laboratories Agency (1995) Wilton Park (2009)

In-depth textual analyses were performed, leading to the formulation of initial categories of agency policy responsibilities. These were then refined through repeat readings. Throughout the process, consistency in applying the high-level criteria was paramount. Nevertheless, the contestable nature of language interpretation, combined with the evolving criteria, meant that some documents occupied different positions before the final classification was reached. Table 6 thus represents but one possible taxonomy, reflecting as much the author’s subjective judgement as the ‘facts’ presented in the official literature.

The taxonomy posits three hierarchically ordered groups. The first contains framework documents reporting a significant, regular policymaking role for the agency. These are further subdivided between agencies leading in policy development, and agencies sharing responsibilities with the sponsor. In the former category is the Identity and Passport Service, whose ‘role extends beyond those normally associated with an executive delivery agency’ (Home Office, 2009: 6). It is ‘the primary source of advice to the Home Secretary on policy issues relating to the National Identity Service’, with the chief executive ‘representing the Home Office on safeguarding identity policy’. The second subset is illustrated by the National Offender Management Service (NOMS). Here, the sponsor ensures ‘the NOMS Agency is fully participative and influential in the process for setting national priorities and strategic policies under which it must operate’ (Ministry of Justice, 2008: 15). The agency takes the lead in certain policy areas, thus suggesting, overall, a sharing of policy formulation duties.

The second major category includes agencies whose policy participation is periodic, occurring more explicitly at the discretion of departments. Agencies may be consulted on policy matters, or assist departments in formulating their advice for ministers. HM Courts Service (2008: 21), for example, is consulted by the Ministry of Justice ‘on all policy and legislative proposals which it is developing that may have an operational impact on the courts’.

The final category lists framework documents which contain no details of a devolved policy role. This is not to say that these agencies do not, in practice, contribute substantially to policymaking. However, for the purposes of exploring the ‘official’ image of department–agency relations, these bodies have no formal policy responsibilities.

Conclusions

Up to the early years of the New Labour Government, the UK’s agency story is a familiar one. It has been widely retold as the embodiment of NPM-inspired reform. Latterly, despite limited empirical testing, the ‘disaggregation–reaggregation’ thesis has gained acceptance as a narrative of the longer term trend in central government administration. By developing and operationalising an analytic framework built from the original Next Steps discourse and wider NPM literature, this article has found considerable evidence to support this thesis, although not entirely in the terms in which it has often been cited. Specifically, the concept of ‘disaggregation’ has been problematised, with the chief argument being that it should not be treated as a synonym of ‘de-agencification’, or else confined solely to structural elements. In making this case for greater conceptual clarity, the evaluation of the structural and functional aspects of ‘reaggregation’ has moved beyond simple, unidirectional classifications of ‘expansion’ and ‘decline’ to explore the multidimensionality of recent developments.

The UK agency landscape has indeed evolved considerably from the immediate ‘post-Next Steps’ position. De-agencification has played a relatively minor role in this, although the analysis revealed some differentiation across the devolved administrations. Moreover, accompanying the dominant trend towards reaggregation, the last decade has also witnessed further structural disaggregation in some policy areas, thus partially qualifying the dominant narrative. Figures suggest that by 2009–10, direct employment in agencies was actually greater than in 1997. The fact that the agency cohort contracted considerably over this time only serves to underline the multidimensional nature of agency evolution, the importance of conceptual clarity when describing this, and the pressing need for commentators to explore how the executive agency model is itself evolving. For scholars and policymakers alike, the account presented above poses some key challenges going forward, three of which are outlined below.

First, there is the perennial issue of definition. In seeking a transnational formulation of the agency idea, Schick (2002: 35) suggested that ‘the only common element is that agencies are not departments, that is, they are not conglomerations of multiple activities’. The post-New Labour situation in the UK poses a significant challenge to this characterisation, since, through agency-level mergers and, in some cases, the devolution of policy capacity to agencies, the ‘role purity’ associated with high managerialism has been considerably eroded. The theoretical and practical implications of this require further investigation, including on the benefits and costs of expanding the delivery remit of agencies, and ‘outsourcing’ policy advice to them.

Second, the limitations of research based on ‘official’ accounts need to be recognised. Given the developments reported above, it is necessary to ask how department–agency dependencies and relationships are evolving. Gains (2003: 14) has already suggested that, under New Labour, agencies were ‘reined back into departments’. This shortening of the arm’s-length relationship, and the commensurate loss of autonomy, represents a facet of the structural dimension which the analysis above only touches upon. Hence, there is a need for fieldwork to explore beneath official accounts in order to understand contemporary agency practice.

Finally, the paradigmatic implications of the continued use of agency-type structures requires reconsideration. Some quarters now assert that the NPM paradigm is ‘dead’, evidenced in part by the popular expansion–decline narrative for UK agencification (Dunleavy et al., 2006). The situation outlined above, however, reveals a less clear-cut scenario. Although structurally and functionally modified, agencies live on in central government as key infrastructure from the NPM era. This suggests, perhaps, that a ‘post-NPM’ narrative – which purports the adjustment rather than replacement of NPM (Christensen and Lægreid, 2006b: 361) – is more empirically sustainable, at least for central government administration. This remains, however, a preliminary conclusion, requiring further empirical and conceptual work if it is to gain authority.

In terms of current policy, despite the Coalition Government’s commitment to the agency model, recent events suggest that, going forward, developments will retain the complex, multidimensional character observed under New Labour. For instance as well as the promised further wave of agencification, several existing agencies have been merged since the Coalition took office. Moreover, the removal of agency status from the social security bodies in October 2011 (DWP, 2011) constitutes a major retrenchment, and thus sits uncomfortably with the assertion that agencies should be the ‘default’ mode of service delivery. At present, therefore, there is little to suggest that the Coalition will oversee a significant increase in the agency population. Nonetheless, given the expansion of agencies into the education sector, the apparent re-centralising of policy capacity, and the potential impact of the Public Bodies Review, it seems appropriate to predict an era of further continuity and change for agencies in the UK.