Conference PaperPDF Available

Understanding and Preventing Delivery Disasters in Public Services


Abstract and Figures

Public service delivery disasters (PSDDs) involve cumulative harms or life-endangering mistakes for individuals or groups. They occur in unplanned ways at the implementation level and seem to be recurring and deeply entrenched features in public services, despite the growth of improved oversight and control mechanisms. We trace their roots in the nature of services and especially within compulsory consumption, which is fairly distinctive of government organized, financed or regulated services provision. PSDDs can also be compared and contrasted with other related concepts like policy disasters or fiascos. Two key issues involve ensuring oversight or the transmission of monitoring information, plus counteracting public agencies recurrent difficulties in repelling citizens’ or customers’ feedback and complaints. The final section considers potential remedies to avert public service delivery disasters, arguing the reinstating conventional checks and balances is unlikely to be effective, and that more hopeful prospects involve the second wave of ‘digital-era governance’ changes, no-blame methods for gathering information and lesson-drawing and strengthened citizen challenge and redress processes.
Content may be subject to copyright.
Understanding and Preventing
Delivery Disasters in Public Services
Patrick Dunleavy
with Jane Tinkler, Chris Gilson and Ed Towers
LSE Public Policy Group,
London School of Economics and Political Science
Abstract: Public service delivery disasters (PSDDs) involve cumulative harms or life-
endangering mistakes for individuals or groups. They occur in unplanned ways at the
implementation level and seem to be recurring and deeply entrenched features in
public services, despite the growth of improved oversight and control mechanisms.
We trace their roots in the nature of services and especially within compulsory
consumption, which is fairly distinctive of government organized, financed or
regulated services provision. PSDDs can also be compared and contrasted with other
related concepts like policy disasters or fiascos. Two key issues involve ensuring
oversight or the transmission of monitoring information, plus counteracting public
agencies recurrent difficulties in repelling citizens’ or customers’ feedback and
complaints. The final section considers potential remedies to avert public service
delivery disasters, arguing the reinstating conventional checks and balances is
unlikely to be effective, and that more hopeful prospects involve the second wave of
‘digital-era governance’ changes, no-blame methods for gathering information and
lesson-drawing and strengthened citizen challenge and redress processes.
© LSE Public Policy Group
This paper arise from ongoing LSE Public Policy Group research on
‘Citizen Redress’, funded by the Nuffield Foundation
Paper to the Political Studies Association Conference, Edinburgh, 30 March 2010
Service delivery disasters can be defined as:
(a) a series of implementation-level failures, defects, mistakes or fiascos
occurring in a service-providing organization (whether a firm, a
government agency or an NGO); that
(b) lead to catastrophic harm (such as serious injury or death) for an
individual or group of citizens; or
(c) produce lesser harm for citizens, but on a large or massive scale; such
(d) the organizations responsible for providing the service and for the
failure suffer a substantial loss of legitimacy and public credibility.
Conditions (a) and (d) are necessary to recognize a service delivery disaster (hereafter
SDD), plus either one (or both) of (b) and (c).
A public service delivery disaster is an SDD that occurs in aservice that is
directly organized by a government department, agency or authority; or that is
provided on behalf of and financed by government; or that takes place in a context
that is closely and specifically regulated by government. Most public service delivery
disasters (or PSDDs) will share two or more of these essential characteristics –
namely government organization, public financing or subsidies, and government
regulation. Given the recent growth of government outsourcing both to contractors
and to NGOs and third sector bodies, Figure 1 shows that there are many possible
combinations of services that meet the requisite ‘public’ involvement.
In this paper we first explore service delivery disasters in general and then
look at what makes them distinctive in government organized, financed or regulated
services provision. We particularly compare PSDDs with other related concepts like
policy disasters or routine failures. The next sections explore two qualities of services
provided to citizens by public agencies that contribute to disasters occuring:
- the problem of ensuring oversight or information transmission; and
- some public agencies face recurrent difficulties in responding to citizens’
or customers’ feedback and complaints.
The fourth section considers potential remedies to avert public service delivery
Figure 1: Defining public services
Notes: Public services are defined as the inclusion set shaded above.
1. The distinctiveness of public service delivery disasters
Tracing the origins of PSDDs needs to recognize three key aspects. First, they arise in
services, namely ‘anything sold in trade that cannot be dropped on your foot’ (The
Economist, quoted Quinn, 1992, p.8). All service industries have some particular
vulnerabilities in their production and distribution when compared with those of
goods. Second, public services tend to cluster fairly intensively in two areas where the
potential for disasters to arise is inherently strong, namely the provision of services
affecting life-and-death outcomes and involving ‘compulsory consumption’. These
two aspects inter-twice closely in regulatory activities designed to deter or prevent
criminals and potential criminals doing harm to vulnerable individuals or
communities. Third, service delivery disasters are essentially unplanned or unintended
consequences of implementation-level decisions by lower-tier staff and mangers, even
where they are systematically or recurringly produced. In this sense they can clearly
to be distinguished from policy disasters (on which a large literature already exists),
All services
Government organized
Government financed
where mistakes occur as the direct consequence of poor intentional choices by top
decision-makers. W discuss these aspects in turn.
1a The distinctive features of services play an important role in creating some of the
base conditions for public service delivery disasters to arise. Compared with goods:
- services have a strong intangible aspect – we return to this in detail in
section 3;
- services are produced continuously in time and do not exist outside the
time when being provided, so explaining the necessity for there to be serial
mistakes for a service disaster to occur;
- everyone involved in a services organization in a sense co-produces its
outcomes, and from management’s perspective they should care about
service quality (Gronross, 2007);
- as a result within the organization a whole series of individual errors
should be combated or rectified or offset for the most part by good quality
service provision;
- where mistakes occur, there should be well set-up procedures for rapid
service recover (Tax and Brown, 200) and provision of redress to affected
customers (Gronross, 2007).
Focusing first on continuous production, although a lot of publicity tends to be
given to adverse cases attributed to a single highly adverse action (such as a mistake
in identifying what to give as an injection made by a nurse or doctor), genuinely
isolated cases do not qualify as ‘disasters’ in our specific sense above. Public service
delivery disasters do not occur as the result of isolated or one-off failures, nor do they
fit an anomic or ‘random walk’ pattern of failure. Failures that are not serial and
repeated many times do not cumulate. Random failures (especially with a high
incidence) may also erode legitimacy in a ‘permanently failing organization’ (Mayer
and Zucker, 1989) that none the less manages to survive in an organizationally
protected niche. AN example is the Child Support Agency in the UK which was
maintained in being for more than two decades for public finance reasons, although
delivering a continuously poor level of service from its inception (NAO, 2006a). But
the de-legitimizing effects of stochastic failures (even on a largish scale) is rather
diffuse and not focused. No group jeopardy situation clearly arises, and the harm
focused on one individual may not extend over a long time or may be corrected by
repeat encounters with the organization or by the individual successfully activating
some form of redress process despite repeated failures. By contrast PSDDs have to be
serial, clearly patterned and seen as disastrous. For the same reason, these kind of
failings can only be identified post hoc. A problem or a pattern that is (as yet)
invisible or undiscovered or unconfirmed or hard-to-ascertain cannot qualify as a
PSDD, even if serious harm results. To take a recent example, econometric work by
the Centre for Economic Performance at LSE has argued that where hospitals have
difficulty in recruiting full-time nursing staff and have to rely instead on temporary
staff or agency nurses, there is a clearly adverse impact on mortality rates in hospital
care. This is especially clear in London, where high living and property costs make
recruiting full-time staff on NHS rates especially difficult (Propper and Van Reenan,
2008). Scaling this across the UK’s regions the analysts identify a significant impact
on patients’ life chances through reliance on agency nurses with less knowledge of
and commitment to a given hospital. Although some of the themes involved have
been taken up the Department of Health and the National Audit Office (2006b) in
trying to restrict trusts’ relying on agency nurses, this is mainly for financial reasons.
The life-and-death consequences of using agency nurses have not yet lead to this
intractable problem being seen as delivery disaster.
1b The coercive or ‘compulsory consumption’ character of some public services
arises chiefly from their role in regulating social evils intentionally or inadvertently
created by particular sections of the community. Some key differences arise in such
contexts from the classic, dyadic relationships between customers and firms portrayed
in service-delivery textbooks or conventional micro-economics. In practice, the
incidence of ‘classic pattern’ market exchanges is probably far less than is commonly
portrayed, since services consumption often involves decision-making by groups,
partially coerced purchases and certainly significantly constrained choices (if only for
social approval and positional goods reasons). A strong role is also played by
intermediaries, such as the media and other consumers. None the less a strong client-
firm dyadic linkage with an essentially voluntary or quasi-voluntary character at some
level can usually be identified.
In compulsory or coerced consumption, however, there are always at least
three-way, or four-way or multi-way relationships (shown in Figure 2) amongst:
(i) the immediate primary client of the service, its subject, the person on
whom compulsory consumption is practiced or focused, sometimes in a
way that cuts across or is opposed to their choices, intentions or
immediate preferences;
(ii) the secondary clients of the service, who are located in various places
and different orientations within the orbit of the primary client – e.g they
may be quasi-primary clients (like ill-treating parents of abused children)
or intermediaries (such as the family members of elderly people with
(iii) the community’s delegates, sometimes formally defined, as with local
councillors, and sometimes less formally defined, as with the
‘community leaders’ of ethnic or religious groupings;
(iv) the wider community, appearing in different guises and relationships
from one service context to another, as the immediate users or
beneficiaries of services, as taxpayers implicated in the funding of
provision or as the wider community involved solely via public interest
considerations and debates; and
(v) service regulators, more authoritative and usually more generalist
service controllers or regulators located at higher tiers of government.
All in all this makes six different sets of actors that grass-roots staff and organizations
directly providing services must potentially respond to.
1c Distinguishing service delivery disasters from policy disasters is not quite as
straightforward as may at first appear. Because they arise at implementation levels,
service delivery disasters are quite separate conceptually from policy disasters and
policy fiascos, where the origins of harms can be traced to explicit decisions made by
top-level political authorities. However, the line needs to be drawn carefully because
policy disasters (although intentionally willed from the top) will almost always
become manifest as a series of ‘unfortunate’ or unexpected implementation-level
failures, that may be criticized by top decision-makers in a ‘blame evading’ way.
Figure 2: The complex involvements of actors in ‘compulsory consumption’
A clear example of when we can see this effect but dismiss any PSDD
characterization is the almost certainly illegal invasion of Iraq by the USA and UK in
March 2003. This case clearly meets criteria (b), (c) and (d) in our initial definition of
PSDDs above, but not the crucial first criterion. In this case the primary cause of the
multiple deaths and severe harm to both Iraqis and ‘coalition’ armed forces was
clearly the policy decision to invade another sovereign country. Once triggered the
commitment to inaugurate war inevitably entailed hundreds of implementation-level
incidents in which deaths occurred, some due to mistakes made by coalition forces,
some deliberate as a result of the implementation of known and accepted standard
operating procedures, and some due to the unavoidable consequences of
concatenating foreign invading forces with large civilian populations. But these
incidents are not themselves service delivery disasters, since they followed closely
and directly from the primary decision.
Public service
(i) Primary client
or subject (ii) Secondary clients
or quasi-clients
(iv) The wider
(iii) Community
(v) External
service delivery flows redress or feedback flows
By contrast, the implementation-level decisions and policies that create pure
PSDDs are almost always ‘unintended’ in some sense by policy-makers, and hence
they must be inconsistent at some level with official policy. In this sense SDDs must
arise in ways that are independent of or in contradiction with what policy-makers (say
they) wanted or anticipated. The difficulty here is that in many circumstances policy-
makers have willed particular ends but then do not will the commensurate means. A
widespread potential for lesser public service delivery failures is commonly created
by policy-makers who systematically over-legislate or over-specify mutually
incompatible or at best orthogonal aims for grass-roots organizations and staffs,
without providing either sufficient resources or clear enough rules that would allow
multiple objectives to be consistently or simultaneously realized by implementing
organizations or staff. One Home Office official put this point to us as the issue of
whether a policeman exiting the police station turns right or left to start a beat or a
patrol. The discretion which consequently must be exercised by delivery
organizations or ‘street-level’ bureaucrats and professionals is a recurringly present
element in the seed-bed of root causes for public service delivery disasters (Lipsky,
1983). But unlike policy disasters or fiascos, there is no clear policy-level decision
that willed a disastrous outcome.
Perhaps the key area of overlap, however, occurs where PSDDs are a vehicle
by which the failures inherent in a policy disaster first emerge clearly. This overlap
almost always occurs because top-level policy decisions are not robust when
implemented, but they are liable only to probabilistic (and not inevitable) failures. The
promulgated policies may be just attainable or just realizable simultaneously if very
fine tolerances in service delivery are maintained. But these tolerances are not in fact
practically or universally sustainable. It follows that after a certain passage of time
(depending on how ‘lucky’ the policy-makers responsible are) an event will occur
where the required fine tolerances cannot be realized, at which point a policy’s
potential for ‘catastrophic failure’ will become manifest.
The Hillsborough football stadium disaster is a clear example of a catastrophe
whose immediate or proximate cause was a service delivery disaster (poor policing
decisions), but one that clearly exposed a wider policy disaster that was just waiting to
happen somewhere. The roots of the catastrophe was a policy decision pushed
through in the late 1980s by PM Margaret Thatcher that all English football grounds
should be fenced in with steel cages to prevent pitch invasions by spectators and other
forms of hooliganism. The PM’s highly political interventions destroyed the previous
delicate balance that had existed in a low-key policy community between preventing
hooliganism on the one hand and ensuring crowd safety in football stadia on the other.
Once steel cages were created for all fans, it was only a matter of time before a crowd
surge into an enclosed pen would produce some deaths by crushing. However, for
some years the UK was lucky that careful policing and an absence of triggering
incidents meant that this policy disaster did not become manifest. Then in a 1989 cup-
tie in Sheffield a series of major control mistakes were made by supervising police,
who directed a late-arriving crowd into an already full steel-fenced pen just as a goal
was scored – creating a crowd surge of extraordinary force and endurance.
The scale of the disaster was then hugely magnified by police commanders’
misunderstanding of the situation, their consequent extraordinarily slow responses to
alleviate the crush, and a series of avoidable decisions that prevented ambulances
reaching injured fans. In the end the cumulation of these mistakes produced an
appalling death toll of 99 dead fans and many hundreds of injuries. The disaster lead
to intense controversy over the implementation-level mistakes made, that continues to
reverberate twenty years after the events. But it also triggered an immediate
recognition across the UK that the steel cages policy was a catastrophically
unsustainable one. Within a few months all the fences around football pitches in
England were completely removed, and the previous policy community dominated by
implementation-level actors was restored to its consensual and balanced mode of
operating. By once again carefully setting crowd safety concerns against hooliganism
worries, and by more sophisticated training, better police tactics, and improved
surveillance and intelligence, the UK has had no further major football hooliganism
There may be other continuities and linkages from policy disasters to the
analysis of public service delivery disasters, especially in the so far limited but still
suggestive interpretations developed about the causes of policy disasters. One
tradition of analysis strongly associated with the historian Barbara Tuchman (1984)
blames ‘wooden-headedness’ for the creation of foreseeable ‘policy fiascoes’ (like the
Vietnam War or the 2003 Iraq invasion), where policy-makers were warned
repeatedly of dangers but none the less walked open-eyed into them. The bottom-line
here is Schiller’s lament that: ‘Against stupidity, the Gods themselves contend in
vain..’. A similar refrain often recurs in the reports of official committees of inquiry
documenting the hard-to-understand behaviour of implementation-level staffs in not
taking precautions, not processing evident information and ignoring multiple tripwires
or obvious warning signs not to proceed. But the potential for such behaviours is what
accounts for the existence of organizational checks and balances. In a well-run
services organizations ‘multiple locks’ should need to be opened before serial
mistakes can occur. Incidentally it is exactly these checks that are commonly absent at
the top policy level associated with policy disasters (Roberts, 2008), especially in the
UK’s still highly centralized polity (Dunleavy. 1995).
A second tradition consequently believes that ‘stupidity’ only flourishes when
it is allowed to do so. It therefore seeks answers in the organizational procedures,
culture and psychology that could allow policy disasters to be willingly entered into.
The ‘groupthink’ tradition associated in a monist way with Janis (1982), and in a
more pluralist way with t’Haart (1994; see also Bovens and t’Haart, 1998) essentially
blames the concatenation of a strong leader with an insulated top elite group – often
one that has overly high morale as a result of previous large-scale or hard-to-
accomplish achievements. Strong groupthink produces a marginalization of outsider
views or critical comments and a dynamic of completely marginalizing non-
conforming information, judged by the ‘insider’ group to be destructive of their group
morale and ambition (perceived as a vital ingredient of securing changes). This line of
analysis seems clearly inapplicable to the many public service delivery disasters
associated with ‘permanently failing organizations’ - such as mistakes made by over-
worked local government social service departments that are below complement on
professional staff, operating with swamping caseloads and with rock-bottom staff
morale. Clearly any groupthink going on here is chiefly of what Hood (1998) terms
the ‘fatalist’ variety. However, some PSDDs have clearly arisen when overly strong
management teams, over-committed to achieving financial or administrative targets,
feel so strongly supported by external regulators and central government pressures
that they do commit classic ‘groupthink’ errors of over-ambition and insider-
groupline thinking (see below for the Maidstone and Tonbridge Wells hospital trust
A third (rather fundamentalist) tradition is that initiated by Friedrich Hayek’s
critique of the necessarily dictatorial potential of compulsory consumption in
particular, but also inherent in the extension of any politicized discretionary choice
into the determination of fundamentally important consumption. For neo-liberal
authors the elimination of choice and distortion of market processes by governments
directly organizing services, or providing subsidies, or just regulating markets,
automatically stores up a potential for political and administrative power to aggregate
in uncontrollable ways in the hands of officials. After each incident or problem
exposed a limitation of past procedures, the overlaying of more regulation displaces
and masks this potential but cannot eliminate it (see Dryzek and Dunleavy, 2009,
Ch5). This Cassandran warning is rather too high level to be useful in analysing
particular cases. But it does find some echoes in the modern applied literature –
notably in Hood’s (1998) stress on the essential pluralism of administrative control
levers and in the advocacy of quasi-markets as correctives to hierarchist modes of
organizing. A useful (albeit still high-level) link from the Hayekain tradition to the
groupthink approach is made by James Scott (1988) in his critique of the necessarily
imperfect information available to policy-makers engaged in non-piecemeal social
engineering. He argues that the most ambitious schemes ‘to improve the human
condition’ are necessarily premised on ‘seeing like a state’. Policy-makers have
radically inadequate information about what he terms ‘metzis’, the highly complex
grass-roots compendium of practices, arts, skills, expectations, different ethoses and
cultures by which different areas of social life are recurringly organized, reorganized
and made to work in practical and joined-up ways. The link to PSDDs here clearly
focuses on the unanticipated impacts of centralized or upper-tier target-setting or
policy-directions in creating favourable conditions for delivery disasters to occur.
From a completely different tack, and designed to negate the neo-liberal
critique, there has been a near-continuous over-time push for initial treatment and
redress rights in public services to be more closely specified in legal terms and linked
to freedom of information and ‘open-book government’. These ‘voice’ correctives
were designed to counter the potential for unconstrained use of administrative
discretion and have acquired new resonance in the post-NPM era.
2. Poor monitoring information or oversight
The proximate roots of every service delivery disaster are always some kind of failure
in oversight or lack of information on the part of decision-makers about what is
happening at the implementation level, the grass-roots interface where the service is
actually being delivered to users, clients or citizens – or forced upon reluctant
consumers of compulsory consumption. At first sight the issue seem to be just a
traditional organizational control problem, reflecting the age-old tendency of agents to
want to shirk or displace organizational priorities into pathways that cut their
immediate transaction costs, making life that little bit more comfortable or less
challenging for the agent, but at the client’s expense. So if a social worker calls to see
a child who is potentially at risk and then is fobbed off by the parents involved
without getting any contact, it is easy to see that the road most travelled will be to file
a report and defer action until next time. But when police, health services personnel
and social workers have 62 similarly inconclusive contacts with the same family
about the same child in a short period (as happened in the Baby P case in Haringey), it
becomes a good deal harder to understand why no organizational or inter-
organizational tripwires were triggered and why more direct action was not taken by
someone at some stage. Three key dimensions are involved here beyond the internal
failings of weakly controlled delivery organizations. These are the complex nature of
principal-agent relationships in long delivery chains and inter-organizational settings;
the multiplication of regulatory systems that followed the new public management
(NPM) era; and the impact of more ICT-based ‘network control’ mechanisms in the
current post-NPM era.
2a Principal-agent conflicts in public service delivery chains have attracted
increased attention from a wide variety of scholars in recent years, ranging from the
‘soft’ economics approach to social policy of Julian Le Grand (2003) through to more
sustained efforts to bring together public choice and organizational analysis
perspectives (Brehm and Gates, 1999). Le Grand presents perhaps the clearest but
rather over-simplified picture in which the key questions revolve around whether the
agents involved in public service production are ‘knights’ with an autonomous push to
seek the public welfare, or ‘knaves’ always liable to divert activities in ways that suit
their self-interest, while treating service users themselves as ‘pawns’ to be organized
at the staff’s discretion and largely for their convenience.
The problem here is that this is too simple a set of alternatives, as more recent
economic analysis has recognized. For instance, Besley and Ghatak (2005) argue that
‘mission-committed’ public servants (such as the knightly image assumed by
progressive public administration) in fact offer a fundamental alternative to the
traditionally hostile view of agents’ motivations offered by micro-economic theory.
They point out that mission-committed agents select for particular jobs and roles,
derive welfare gains from working in the ‘right’ kind of organization, often accept
lower wage rates and in many cases will also foster a counterpart selection of users.
At its best, this can produce ‘matching equilibria’ with mutual gains for staff and
clients in successful organizations (but possibly adverse selection or club effects
Stemming from their economic origins, principal-agent models have also
historically assumed that the principal is always right. The principal is always
authoritatively and legitimately able to decide on what the organization should be
doing, always best informed about the ways in which their goals can be met, and
always essentially benevolent in their intentions towards the organizations they lead
(Perrow, 2006). Hence principal-agent problems are characterized in this literature in
a thoroughly one-sided fashion as the product of agency deviations from the
principal’s well-meaning and far-sighted intentions. Yet there is no theoretical or
other basis for such far-reaching assumptions. Top managements may well run
organizations so as to maximize their own personal welfare at the organization’s
expense. Think of the Rover buy-out directors who (gifted with government
subsidies) managed to extract significant private fortunes totalling £42 million in pay
and pensions from selling off the firm’s assets over three years before the company
went bankrupt in 2005, throwing its workforce onto the dole and its creditors into
problems (see the government inspector’s report at Inside the public sector, there is no
particular reason why we should expect service-delivery organizations to be run by
‘knightly’ managements battling ‘knavish’ agents. Instead both groups may behave in
both ways. In addition, even ‘knightly’ behaviours may be more ambiguous. They
may reflect an accurate and genuine effort to foster public interest goals; or they may
instead be based on overly strong ‘mission-commitments’ that deviate from the public
interest for non-knavish reasons, such as the biases of ‘zealots’ analysed by Anthony
Downs (1967; and see Dunleavy, 1991, Ch.6).
Following a conventional dichotomy in Le Grand mould, Figure 2 shows how
the interactions of differently motivated principals and agents produce four possible
outcomes – of which only the ‘convergent’ outcome is likely to result in smoothly
administered service implementation. If both sides behave opportunistically, then
conflict is likely to be maximized, with staff seeking to short-change the agency and
managers to exploit their workforce. Classic agency problems only arise if the
principal is clearly public interested, for instance, ensuring that agents are not
overloaded, are properly trained and equipped, and have feasible workloads that
would allow them to deliver services as intended. However, opportunistic
organizational leaderships (both administrative and political leaders) may commonly
not make these provisions, instead seeking to secure service delivery on the cheap, or
loading implementation-level staff with infeasible workloads and foreseeable stress.
Figure 3: Conventional ways of characterizing principal-agent problems
(Cell entries show the kind of organization resulting from the principal-agent
Yet in fact service delivery problems seem to most commonly occur when in
addition to knightly or knavish behaviour, either principals or agents behave
passively. Figure 3 shows how this possibility creates five additional organizational
conditions. If both groups are passive, the organization may be rudderless and its
operating practices dominated by a ‘fatalist’ culture in Hood’s (1998) terms – here the
accumulation of mistakes inherent in service delivery disasters can easily acquire
momentum over the long term. If the management is passive and the agents
opportunistic, then systematic rent-seeking by staff can create significant problems.
But equally if managers are opportunistic and staff are passive, much the same
outcomes can accrue. The last two outcomes present fewer problems, where managers
Staff (agents’) behaviour isTop management
Agency problemWelfare-
maximizing or
maximizing or
mission committed
Staff (agents’) behaviour isTop management
Agency problemWelfare-
maximizing or
maximizing or
mission committed
Figure 4: An expanded view of principal-agent problem combinations
(Cell entries show the kind of organization resulting from the principal-agent
are public-interested and staff are passively compliant, or where the management is
passive but a mission-committed staff still operates like an orchestra without a
conductor. Yet there are dangers inherent here too, especially in professionally
dominated organizations with their own flaws and blind spots. Here the absence of
active and effective management, and the minimal middle-line management and
strong work autonomy characteristic of professional bureaucracies (Mintzberg, 1992),
all mean that mistakes shielded by professional codes and working practices can
multiply or extend unacknowledged over long periods.
The overall context in the public sector is additionally shaped by the pervasive
appearance of multiple principals and agents, and by groups who play one role in
some of their interactions and a different role in others, as Figure 5 shows. In theory,
voters are the ultimate principals for the whole state apparatus, electing the legislature
which (at least in parliamentary theory) then becomes the principal selecting the
executive as agents. National level ministers then play the role of principals in
appointing top bureaucrats in ministries as agents. Top officials also act as principals,
however, in creating and running quasi-autonomously departments or agencies
composed of lower-ranked officials, or in mandating and funding lower tier agencies
to undertake implementation. Finally though, these lower-rank officials themselves
Staff (agents’) behaviour isTop
maximizing or
dominated agency
or fatalist
Acute rent
Exploited or
or mission-
Staff (agents’) behaviour isTop
maximizing or
dominated agency
or fatalist
Acute rent
Exploited or
or mission-
Figure 5: Principal-agent chains in the governance of public services
now appear as principals in their dealings with citizens (who are now treated as
agents), the transition being especially well-marked in compulsory consumption
Complex public service delivery chains create well-known effects, multiplying
the number of principals and blurring control systems. The experience of acting both
as principal in some roles and as agent in others is also almost calculated to sow the
seeds of potential confusions in such delivery chains – with the consequence that
agents consciously or otherwise act as a quasi-principal in an inappropriate context
where they should not do so. This problem is especially acute at the implementation
level, the end of the chain where agents meet citizens in the guise of principals
deciding for individuals or groups (albeit on behalf of citizens as a whole) what
treatments or services they can receive.
Of course, all public service delivery agencies are overseen internally and
externally (see below), although the arrangements operating often vary strongly
across three key contexts:
- for professions, where delegated and ‘mutuality’ controls are prevalent;
or TOP-
- for semi-professions, where legislators or governments often try to more
closely specify what services are to be developed; and
- for straightforwardly bureaucratic contexts, where legislators directly
mandate rules and behaviours, standard operating procedures seek to
minimize officials’ or implementers’ discretion, and staff are treated as
‘robotic’ non-professionals.
However, many developments in the new public management period worked to de-
privilege professions, in the process adversely affecting mission commitments
amongst increasingly externally regulated staffs.
2b The multiplication of regulatory systems was always a problem within the public
sector under traditional ‘progressive public administration’ approaches that dominated
the early post-war period, or under earlier Weberian model administrative processes.
But at least their respective emphases, upon professionalism and old fashioned citizen
redress (for PPA) and on top-down control systems internal to large organizations (for
Weberian machine bureaucracies), both meant that complexities were intra-
professional or intra-organizational. By contrast, new public management approaches
from the 1980s onwards stressed competition and disaggregation in organizational
design (including purchaser–provider separation), the displacement of professionalism
by ‘incentivization’ strategies that are fundamentally pecuniary-based, and stronger
corporate management in each public service organization. The combination of these
three changes all created obvious risks of non-conforming behaviour. And as these
additional sources of risk were actualized they triggered backlashes in the UK’s
strongly centralized governance systems, helping to spark an ‘audit explosion’
(Power, 1994).
The organizational fragmentation engendered by NPM practices to
disaggregate large hierarchies and separate out roles and spheres of activity under
stronger corporate managements, had a specific counterpart in regulatory
multiplication, where separate aspects of the same essential processes were
externalized to small, specialist regulatory agencies. Decision-makers influenced by
NPM always favoured clarity and focused expertise over coherence, especially in the
UK from 1985 to around 2005. An acute illustration was the situation for English
NHS hospital trusts, which by 2003 were tasked with reporting to no less than 17
different Whitehall or quasi-government agencies acting as narrow regulators of
issues like estates, legal cases, patient safety, healthcare quality and commissioning
practices – in addition to pre-existing professional controls. Greater institutional
complexity at the top made overall monitoring or oversight of implementation-level
agencies more difficult, as well as creating multiple difficulties in partnership working
between connected public services responding to different complaints systems appeals
systems, regulators and ombudsmen. Together with principal-agent chains that often
originated contradictory signals, regulatory disaggregation exacerbated difficulties for
grass-roots organizations and staffs in exercising their discretion about how to provide
services, already subject to legislative over-specification and built-in priority
2c Networked over-sight and modern widened spans of control have also
contributed to reducing implementation-level oversight in NPM systems, but in a
complex dialectical kind of way. The potential nature of the problems here were well
dramatized in Paul Greengrass’s 2006 film of United 93, where many scenes showed
the national civilian air traffic control headquarters in the USA. Although kitted out
with sophisticated ICTs for monitoring routine traffic events, the highly professional
staff were completely unable to process or summon up relevant information for the
unusual circumstances of the 9/11 terrorist incident. How could it be, the film asked,
that the ordinary passengers on the eponymous flight, using no technology beyond
cell phone calls to their families, were able to work out what was happening - while
civilian and military control rooms were a chaos of unsighted expert staff, awash with
rumours and confuser stories, with very weak connections to the military commands,
and liaising largely by shouting at each other across roomfuls of expensive and partly
useless IT equipment?
A key theoretical perspective is provided by Garicano, van Reenan and
colleagues who argue that modern ICT developments essentially involve two partly
contradictory trends (Bloom et al, 2009). On the one hand the creation of more and
more sophisticated ICT monitoring networks with radically increased channel
capacity and near-instantaneous response times have been powerful impulses behind
the de-layering of middle-line managements in business, and more recently in the
government sector. Of course, there have been a range of technological shifts here,
not just ICT capabilities but related changes in business processes over long periods,
for instance to create multiple outputs and outcomes indicators that did not previously
exist. But networks expansions have had a decisive influence in unlocking
organizational re-structuring. Spans of control in organizations and public service
delivery chains have widened, so that a single superior can now use ICT-based and
KPI information to regularly monitor the behaviour of much larger numbers of
subordinate personnel or units. Fundamentally this is a centralizing effect, gathering
more controls in the hands of top decision-makers.
At the same time, the authors argue that the development of more and more
powerful and effective databases tends to empower grass-roots workers and
organizations. Handling much larger volumes of information, much more quickly and
intelligently, means that implementation-level staffs have less need to refer issues
upwards to superiors or sideways to specialist units for explication or clarification. So
the ability for grass-roots staff to resolve previously difficult issues has been greatly
boosted. Database development is thus fundamentally decentralizing in its impacts,
tending to make more relevant knowledge easily accessible to the service deliverers
themselves, and reducing the need for centralized advice/control resources.
In certain circumstances, these apparently orthogonal trends might combine to
work very effectively in curbing some significant implementation-level problems. For
instance, the original hopes for the large National Programme for IT (NPfit) covering
hospital IT and databases in the UK were that it would cumulate in a situation where
some sources of mistakes could be systematically excised. There are around 3,000
deaths a year in the NHS, arising from patients who are clearly documented as being
allergic to penicillin none the less being prescribed the drug by doctors. If a central
data registry of patient records can in a sense ‘look over the shoulder’of all prescribers
then in theory many potentially serious mistakes like this could be nipped in the bud.
Here the ICT system would first warn doctors to check for contra-indicated
behaviours, then perhaps require co-authorization before proceeding, or perhaps not
allow the potentially fatal choice to proceed at all. Of course, at this stage the actual
implementation of NPfit has not in fact reached the level of development or
nationwide spread of the technologies needed to deliver clearly improved results. But
a range of other related changes have made some useful progress – e.g. the
development of patient testimonials on the NHS Choices website has begun in a
limited way to improve the flow of information from one patient to another, creating
an important discipline also on hospital managers’ responsiveness to the public
exercising their expanded ‘voice’ and exit options.
Yet given the imperfect and uneven state of ICT progress, and difficulties in
articulating it with established professional and organizational methods of working, it
is possible for improvements in regular or routine monitoring to coexist with serious
failings in coverage on exactly the kind of exceptional cases that tend to trigger
service delivery disasters. Information overload may be generated by floods of
unprocessed or de-prioritized data where separating the wheat from the chaff becomes
more difficult as data volumes grow exponentially. The combination of network
centralization with both Weberian bureaucracy’s ‘safety first’ precautionary stance,
and with NPM’s stress on stronger corporate management, seems to have greatly
contributed in the UK to the development of a ‘surveillance state’ complex, and the
accumulation of ‘needle in a haystack’ problems in areas like intelligence about or the
monitoring of child abusers (as the Soham double murder case revealed in one serious
3. Ignoring complaints and feedback
Public administration systems in western liberal democracies have been in place in
substantially their modern form now for more than a century in all countries, and in a
few countries for much more than that. In this period the evolution and setting of their
fundamental procedures and orientations has had multiple occasions to change and
develop in ways that are now deeply embedded. All long-lived public service
organizations have also evolved strong legitimizing narratives of what it is that they
do and in particular of how they relate to citizens, customers and clients. Modern
controls and checks and balances have included the shift to articulating ‘public value’
components more explicitly, freedom of information requirements, civil and human
rights laws, and modernized oversight and redress systems using e-government
processes. They might all have been expected to constantly improve organizations’
responsiveness to their customers. And yet, service delivery disasters along with other
large-scale mistakes and scandals still seem to occur with depressing regularity. We
trace the roots of these problems to first the rather suppressed ‘dark-side’ of
progressive public administration, and second the strongly and distinctively anti-
responsiveness elements in new public management doctrines that dominated debates
from 1980 to 2005.
3a The dark side of progressive public administration (PPA) systems was always
the abuse of professional power at the expense of the ‘customers’ or ‘clients’ of public
services, itself founded on a kind of disdain the knowledge and choice capacities of
ordinary people. A profession is a strong in-group, with its own potential for
‘groupthink’ problems and a strong occupational community that operates in a
language that is often largely inaccessible to its users. In public administration
contexts, it was easy to both recognize the need to follow public opinion and to look
down on ordinary people’s capabilities. Woodrow Wilson (1886) wrote:
[T]he reformer is bewildered by the fact that the sovereign’s mind has no
definite locality, but is contained in a voting majority of several million heads;
and embarrassed by the fact that the mind of this sovereign also is under the
influence of favorites, who are none the less favorites in a good old-fashioned
sense of the word because they are not persons by preconceived opinions; i.e.,
prejudices which are not to be reasoned with because they are not the children
of reason.
Wherever regard for public opinion is a first principle of government,
practical reform must be slow and all reform must be full of compromises.
The compounding of these strong professional impulses with state patronage
of the professions largely founded on the compulsory consumption by citizens (often
reluctant citizens) of services that the state judges beneficial, all created historically
strong conditions for the flourishing of a professionally based paternalism (Johnson,
1972). Most of the errors and deviations that resulted were undertaken for reasons that
always had some strong public interest or mission-committed rationale, and could
then be sustained by a culture of official and professional secrecy that kept knowledge
of controversial working practices to a minimum. Yet the essentially coercive
character of ‘doing good’ on behalf of people whose own preferences and voices were
heavily discounted in producing solutions was evident in field as different as the
provision of mass housing as high rise flats (Dunleavy, 1981; Ravetz, 19982); the
period of highly treatment of childbirth in medicine (Cahill, 2001); and the
progressive upwards revision of scientific standards for risks to human health from
nuclear radiation and polluting materials such as asbestos (Bartrip, 2002). A good
micro-example of the same tendency that endured into the twenty first century was
the decisions made by hospital surgeons across the UK for many decades to
accumulate large stocks of body parts taken from patients who had died, without any
authorization to do so from the families involved, and apparently with little oversight
by the governance structures in place supposed to control such behaviours (Bentham,
We are also now in a better position to appreciate the large scale and lengthy
endurance of some of the ‘dark side’ practices of progressive public administration
from the 1900s through to the 1960s. Unacceptable practices involving strong and
repeated abuses of human rights from this period include the long-run poor treatment
of many mentally ill and mentally handicapped people in large institutions; quasi-
eugenics policies pursued in the UK, USA and many other liberal democracies over
many decades, including the sterilization of persons judged unfit; the state-organized
exporting of working class children from the UK to Australia and New Zealand; and
the masking of long-run abuses in the treatment of children and elderly people by a
minority of staff in state-run institutions. None of these incrementally occurring but
cumulatively massive scandals created service delivery disasters in our terms, yet they
did decisively undermine the legitimacy of many long-established institutions set up
on PPA lines, and help usher in the period of NPM’s hegemony.
3b The false dawn of new public management’s ‘customer’ orientation is much
less well-recognized still, despite more than two decades of experience. In its early
manifestations NPM included some substantial rhetoric about listening more closely
to customers and moving away from the professional hegemonies of the PPA period
(Savoie, 1995). Part of the allegedly more ‘businesslike’ orientation of early NPM
included paying more attention to customers’ views captured in regular satisfaction
surveys. But this phase died out within a few years, as NPM instead emphasized the
importance of forcing through change whatever ‘customers’ might say, fostering a
kind of Leninist ‘choice’ agenda in which consumers would be forced to be free,
whether they wanted to retain old-style pubic services or not. Each of the three big
NPM themes created its own perverse effects on implementation-level (Dunleavy et
al, 2006; Dunleavy et al, 2008, Chs. 4 and 9):
(i) Disaggregation, the breaking up of large hierarchies into component parts,
promoted purchaser-provider separation, the growth of quasi-government
agencies, regulatory fragmentation, privatizations and the push to create
micro-local agencies. Key adverse consequences for service delivery disasters
followed from the growing institutional complexity of the public service
world. With more cracks between multiple ‘boutique bureaucracies’ involved
in service provision, so the potential grew for the least assertive and
knowledgeable customers to fall between the cracks, as in the spiralling of
homelessness in major cities during the 1980s and ’90s.
(ii) Competition to some extent compensated for greater fragmentation by
improving customer information in some choice areas (such as schools
provision) and imposing new market disciplines on providers there. But again,
competition fostered unchecked organizational fragmentation, increased
complexity and created a range of ‘club’ effects, some managerially enforced
and others pushed by user representatives. Competition also undermined
mission-commitment amongst staffs, contributed to target- or KPI-focused
policy making and priority setting and eroded professional regulatory
mechanisms without really providing strong compensating mechanisms to
prevent delivery mistakes, especially in the NHS. Competition also
encouraged a particular kind of client-orientated bureau-shaping in which
local implementation organizations sought to pass the buck for difficult clients
to other bodies. By the end of the 1990s ‘client-shaping’ strategies between
rival or separately managed providers accounted for a large part of the
delivery disasters in areas like childrens’ abuse cases, failed care of the elderly
and homelessness – in all of which ‘difficult lead agency problems’
(iii) Incentivization strategies promoted by NPM also contributed strongly to
such trends by stressing that only economic or pecuniary incentives can work
to produce public-interested behaviour. Professional de-privileging pushed and
generalized the apparent assumption that all public servants are ‘knavish’
agents seeking to divert public resources to boosting their own welfare unless
constantly checked and regulated. But once this corrosive meme seemed to be
officially countenanced and endorsed in multiple control practices, it elicited
its own adverse or imitative effects, contributing strongly to the reduction in
salience of the ‘public service ethic’ and the erosion of mission commitment
amongst many different agencies’ staffs.
The cumulative impacts of these sustained changes over almost a quarter of a
century are well captured in a 2009 report to the Department of Health on the
National Health Service, prepared by the Massachusetts-based Institute for Healthcare
Improvements. It identified the neglect of patients as a serious obstacle to improving
the NHS, arguing that ‘The patient doesn’t seem to be in the picture’:
‘The lack of a prominent focus on patients’ interests and needs ... represents a
significant barrier to shifting the trajectory of quality improvement in the
NHS… We were struck by the virtual absence of mention of patients and
families ... whether we were discussing aims and ambition for improvement,
measurement of progress or any other topic relevant to quality… Most targets
and standards appear to be defined in professional, organisational and political
terms, not in terms of patients’ experience of care’ (Sunday Times, 7 March
Underlying this transition, however, there were a whole series of NPM-
influenced changes introduced by successive Tory and Labour governments, that
marginalized patient involvement in health services provision. Especially
notable here were the complete abolition of the Community Health Councils in
England (which were well-working independent, consumer-run bodies); and their
replacement by a Patient Advisory and Liaison Service (PALS) run instead by
hospital trusts, with few resources and whose minimal staffs actually function as a
kind of public relations ‘smoothers’, persuading disgruntled patients not to complain.
These already substantial NPM pressures to ignore or marginalize customers were
then compounded by central target-setting and financial penalties for non-complying
NHS authorities.
There have also been specific linkage from some major NHS delivery
disasters to management teams pursuing strongly NPM-influenced agendas. For
instance, the Maidstone and Tonbridge Wells hospital trust in 2006 was seeking to
become a foundation trusts (that is, a health service QGA with enhanced financial and
business autonomy), developing a major Private Finance Initiative for new buildings,
and regrading its consultants – a huge agenda of change to handle at once (Healthcare
Commission, 2007a) . So when its infection-control consultant left and the post was
left unfilled for many months, this was low on the priority list of its top managers.
Over the course of a year the trust had two major outbreaks of the hospital-acquired
infection c-difficile, the first of which they did not even recognize had occurred. For
months the trust repeatedly denied to patients and their relatives that it had any
problems, until eventually more than 1,000 patients had fallen ill from the infection,
of whom around 95 died. Not until many months after patients’ relatives began
lobbying for regulatory action were the full facts uncovered by a regulatory inquiry,
prompting the resignation of the chief executive and belated action to reinstate a full
infection prevention service.
Similarly in the East Staffordshire hospital trust, patient care standards were
widely and routinely neglected by a management obsessed with financial target-
keeping, who also enforced a climate of non-discussion, non-disclosure and
falsification of information amongst medical staff. Managers routinely brought in the
police to remove community protestors seeking redress after their relatives had fallen
ill or died through very explicit quality-shading of basic care procedures. Here again,
after three years of systematic efforts by the trust to suppress complaints, regulatory
intervention uncovered an excessive death toll linked to poor cleanliness and the
neglect of many elderly and frail patients, causing the avoidable or premature deaths
of at least 400 people over the period, and perhaps three times as many (Healthcare
Commission, 2009).
4. Preventing service delivery disasters
Delivery disasters in public services have complex roots and their recurrence in fields
as diverse as the examples discussed above strongly suggests that there are no easy
solutions to preventing them in the future. We review three main options currently
open to public authorities seeking corrective responses to implementation-level
failures in service provision – turning back the clock to reintroduce more conventional
internal checks and balances; using favourable ICT trends more intelligently than in
the past to foster more agile and more decentralized responses to service failures; and
introducing new ‘challenge’ and systematized redress opportunities.
4a Reinstating conventional checks and balances has continued to be the
instinctive response of UK politicians and top administrators to new crises or
scandals, with ‘crack downs’ on poor practices often following on months or years
afterwards because of the need for a lengthy regulatory or public inquiry process,
examining what went wrong in triggering a delivery disaster. An influential strand of
structuralist pessimism in public administration theory, well represented by
Christopher Hood’s The Art of the State (1998) concurs from a cultural theory basis
that there can only be an endless cycle of remedies for past failings. In this view it is
important not to restrict the menu of historically given solutions, and to continually
permute them in new combinations to counter the latest delivery disaster virus, while
also being aware that new practices may in turn often compound difficulties, by
producing new variant problems.
Yet there are strong grounds for believing that turning the clock back from a
post-NPM setting so as to try and interpose Weberian controls or modernized forms
of PPA-style regulation will rarely work to curb delivery disasters in public services.
In the UK especially contemporary risk-based systems of administration are mostly
too fragile and too fine-tuned already to be able to muster the extra finance, staff or
management attention needed to reinstate the Weberian virtues of constant checks, or
the PPA virtues of collegial professional delegation. Sure there can be a post-crisis
response for a few years, in which the extra resources needed to shut the stable door
on the escaped horse are delivered. But a ceaseless push for efficiency savings will
often dull the edge of internal regulation tightening after a few years, with strong
pressures to cutback activities that seem simply supernumary.
Equally, while Hood is sanguine about adopting ‘mix and match’ strategies in
terms of control mechanisms, importing additional diversity into NPM-influenced
administrative cultures is not straightforward. The multiplication of control
mechanisms ineluctably tends to overload implementation-level staff and create
additional grass-roots discretion. In a post-disaster climate staff will also often adopt
risk-minimizing responses with their own substantive dangers. Thus successive efforts
to respond to child abuse cases in Britain over several decades have greatly tightened
up computerized audit and paperwork controls on social workers. But a key
consequence has been the perverse effect of inducing social workers to spend ever
larger proportions of their time on desk-based paperwork, constructing an audit trail
that puts them in the clear in the event of future disasters, rather than actively
engaging with clients’ families (Munro, 1996). The Baby P case in 2009 also
produced a very large precautionary year-on-year increase in the numbers of children
removed from their parents and into foster care or childrens’ homes. Successive
disasters and criticisms, backed by imposing new bureaucratic checks, have also
worsened the recruitment of new social workers and reduced the profession’s morale
and legitimacy in dealing with all their clients.
In sharp contrast are the non-conventional strategies being copied in the NHS
from well-established ‘safety bureaucracies’ (such as the airline industry, where
safety levels have dramatically improved). With 50,000 accidental and potentially
avoidable deaths a year in NHS hospitals in view, the Department of Health has
successfully pursued ‘no-blame reporting’ policies, designed first and foremost to
generate accurate information of the extent and nature of accidents, so that more
focused lesson-drawing and corrective actions can be developed and integrated into
existing established mechanisms, such as improving clinical audit.
4b Fostering database-lead decentralization also offers better prospects for
boosting the feedback and control loops that can help nip in the bud those service
delivery disasters that involve the cumulative repetition of mistakes and that grow
over time to adversely affect large groups of people. The decentralizing and grass-
roots empowering character of improved databases stressed by Bloom et al (2009) of
course requires business process changes, and often also organizational culture
changes, before they can foster radically improved responsiveness. In professional
contexts like medicine, for instance, some authors have argued fot he importance of
relatively simple organizational changes that have proved important in air safety
improvements – including the systematic use of checklists in all simple and complex
operations, and the empowering of all team members in monitoring the behaviours of
leading professionals, on the aircrew safety model (Gawande, 2010).
The digital era governance (DEG) model stresses two key requisite
organizational and cultural changes for digitalization to be effective in boosting
service quality (Dunleavy et al, 2006 and 2008):
(i) The reintegration of provision into more coherent, post-NPM patterns through
- regovernmentalizing key risks and functions;
- some recentralization, especially creating a more intelligent centre that
realizes public information goods for complex delivery chains as a whole,
as with initiatives like NICE in assessing health technologies;
- strong partnership working across the myriad organizational boundaries
created by NPM, eventually building into a case-by-case mergers and
acquisitions process already extensively underway in UK public services
at local level and many related ‘shared services’ developments (Dunleavy,
(ii) Needs-based holism in the design of administrative structures, in particular
- creating client-focused government structures, such as the main internal
sections of the Department of Work and Pensions (DWP) focusing on
working age and elderly client groups; and integrated childrens’ services
departments at local government level, replicated centrally in the
Department of Children, Schools and Families (DCSF);
- developing integrated public service one stop shops and one-stop
windows, backed by business process changes, such as the ‘Tell us once’
initiative and the fundamental pooling of previously separated databases
into useable integrated information (Varney, 2006);
- empowering new choice and voice options, such as the direct customer
testimonial capacity of NHS Choices, which has a potential to contribute
to preventing the recurrence of problems such as those in Maidstone and
East Staffordshire hospital trusts; and
- a push towards the increasing co-production of complex outcomes jointly
with citizens, using ‘nudge’ and other behaviourally-focused techniques
that preserve citizens choices but set up choices in ways that foster
developmental changes in areas like changing diets and alcohol
consumption, fostering healthier lifestyles and involving families and
communities more effectively in the networked care of the frail elderly,
long-term ill patients, or mentally handicapped people (Thaler and
Sunstein, 2009).
The second wave of digital era governance changes have already begun to
create a ‘DEG 2’ paradigm for rapid organizational change where organizations not
only ‘become their web site’ but also develop the new technologies and social web
potential of ‘Web 2.0’ changes so as to produce quantum increases in the amount of
distributed information accessible to civil society. The dramatic advances of ICT-
assisted ‘open book government’ in the UK since the Freedom of Information Act
(including its application to Parliament especially triggering the expenses scandal of
2009-10) have been especially critical in ushering in an improved prospect for citizen
redress and citizen co-production of services. Allied with powerful database learning
effects that empower implementation staff across related organizations to provide
more joined-up services, a wide range of DEG 2 developments have a significant
capacity to help prevent service delivery disasters occurring, or growing once they
have occurred.
4c Effective challenge processes and improved redress mechanisms are none the
less still the changes most needed to control for groupthink or principal/agent
interaction problems in public sector service delivery organizations. Controls by
internal managers, or reporting to top-tier or specialist regulators are not enough to
secure the necessary cultural changes to underpin agile and genuine responsiveness to
citizens and users at the implementation-level. Here the challenge function is best
undertaken by citizens themselves, using greatly simplified, joined-up and re-
weighted citizen redress mechanisms. Currently the UK has a baffling range of
redress mechanisms in public services, based on a complaints versus appeals
dichotomy that is sacrosanct for government officials and agencies, but has little
meaning for ordinary people (Dunleavy et al, 2005). Yet redress systems also involve
a bewildering range of regulators, mediators, and multiple varieties of ombudsmen, as
well as legal challenges and the constituency service roles of MPs, creating an overall
‘redress industry’ costing some £2 billion annually and processing millions of cases,
albeit slowly and often poorly (Dunleavy et al, 2010). Radically simplifying and
integrating redress systems, and making them more agile and effective, could go a
long way to building the capacity of citizens to challenge poor levels of service
provision at source, and to speed up the way that even NPM managements are forced
to react.
Delivery disasters in public services can be defined in ways that parallel but are
clearly distinct from policy fiascos and disasters, primarily in terms of arising at the
implementation level and in unplanned or unintended ways. Service delivery disasters
have complex origins partly grounded in the nature of services and the focusing of a
good many public services on compulsory consumption. All delivery disasters involve
poor oversight or information systems on the one hand, but also tendencies built into
both PPA and NPM public services that tend to marginalize complaints and feedback.
No single strategy is likely to be effective in preventing PSDDs, but equally there is
no reason to retreat into a quasi-structuralist pessimism that accepts poor performance
‘to endless years the same’. Regressing to Weberian control mechanisms or PPA-style
reliance on professionalism are unlikely to be feasible longer term in modern, risk-
based systems of administration. But systems fostering no-blame reporting and linked
to the decentralizing potential of databases hold better prospects of moving public
service delivery systems towards sustainable improvements paralleling ‘safety
bureaucracy’ developments elsewhere in the economy. Essential ingredients for
improvement are also likely to be improved challenge functions and more simplified
and integrated citizen redress mechanisms, fitting within a broader pattern of ‘digital
era governance’ changes towards reintegration, needs-based holism and the strong
development of second-phase digitilization changes.
* This paper draws on work for the LSE Public Policy Group’s ‘Citizen Redress’
project, funded by the Nuffield Foundation (2009-10) and drawing also on earlier
work for the National Audit Office (2004-8). We would like to thank both bodies for
their help and support, but the arguments and conclusions made here are those of the
research team alone. We thank Martin Loughlin, Patricia Bartholomeou, Rosie
Campbell, Francoise Boucek and Tobias Escher who have worked with us on aspects
of the Citizen Redress project across this period. We are deeply indebted to our co-
researcher Helen Margetts from the Oxford Internet Institute fro many ideas and
1. Formally, let us label serial implementation level failure as F, catastrophic harm
to someone as H, large-scale but perhaps lesser harm as S, and legitimacy loss as L:
then an SDD = F + L + (H S).
Bartholomeou, P., Campbell, R., Dunleavy, P., Margetts, H. and Tinkler, J. (2005)
Report on Four Focus Groups conducted for the Citizen Redress NAO report.
Available from:
Bartip, P. W. (2002) The Way from Dusty Death: Turner and Newall and the
Regulation of Occupational Health in the British Asbestos Industry, 1890-1970
(Athlone Press).
Bentham, M. (2001)‘ “Half of hospitals” involved in body parts scandal’, Daily
Telegraph, 28 January.
Besley, T. and Ghatak, M. (2005) Competition and Incentives with Motivated Agents’,
American Economic Review, Vol. 95, No. 3, pp. 616-636.
Bloom, N., Garicano, L., Sadun, R. and Van Reenen, J. (2009) ‘The Distinct Effects
of Information Technology and Communication Technology on Firm Organization’
(London: London School of Economics, Centre for Economic Performance),
Discussion Paper No 927.
Bovens, M. and and t’Hart, P. (1998) Understanding Policy Fiascos (New Brunswick:
Brehm, J. O. and Gates, S. (1999) Working, Shirking and Sabotage: Bureaucratic
Response to a Democratic Public (Ann Arbor: Michigan University Press).
Brewer, Brian (2007) ‘Citizen or customer?: Complaints handling in the public
sector’, 73 International Review of Administrative Sciences 549-556.
Cahill H. A. (2001) ‘Male appropriation and medicalisation of childbirth: an historical
analysis’, Journal of Advanced Nursing. Vol. 33, pp. 334–42.
Crerar, L (2007) Independent Review of Regulation, Audit, Inspection and Complaints
Handling of Public Services in Scotland. (Edinburgh: Scottish Government).
Dean, H (2004) ‘Losing appeal? The changing face of redress’, 12 Benefits 3-7.
Department for Constitutional Affairs (DCA) (2004) Transforming Public Services:
Complaints, Redress and Tribunals, Cm 6243. (London: HMSO).
Department for Health (2004) Learning from Complaints: Consultation on Changes
to the Social Services Complaints Procedure for Adults. (Leeds: DoH).
Department of Work and Pensions. (2008) Performance from the Customer’s View
(London: DWP Customer Insight).
Downs, A. (1967) Inside Bureaucracy (Boston: Little Brown).
Dryzek, J. and Dunleavy, P. (2009) Theories of the Democratic State (Basingstoke:
Dunleavy, P. (1981) The Politics of Mass Housing in Britain, 1945-75: A Study of
Corporate Power and Professional Influence in the Welfare State (Oxford: Clarendon
Dunleavy, P. (1991) Democracy, Bureaucracy and Public Choice:Economic
Explanations in Political Science (Hemel Hempstead: Harvester-Wheatsheaf, 1991).
Dunleavy P. J. (1995) 'Policy disasters: explaining the UK's record', Public Policy and
Administration, vol. 10, no. 2, pp. 52-70.
Dunleavy, P. (2010) ‘The Future of Joined-up Services’, (Paper for the ESRC and
2020 Public Services Commission).
Dunleavy, P., Margetts, H., Bastow, S. and Boucek, F. (2005) Citizen Redress: What
citizens can do if things go wrong with public services. (London: The Stationary
Office). National Audit Office value for money report, HC 21 Session 2004-2005.
Also Web published at and at
Dunelavy, P., Margetts, H., Bastow, S. and Tinkler, J. (2006) ‘New Public
Management is dead. Long live digital-era governance’, Journal of Public
Administration Research and Theory, vol.16, no.3, pp. 467-94.
Dunleavy, P, Margetts, H, Bastow, S and Tinkler, J (2008) Digital Era Governance:
IT Corporations, the State and e-Government. (Oxford: Oxford University Press),
revised edition.
Dunleavy, P, Margetts, H, Bastow, S, Escher, T and Tinkler, J (2007) Government on
the internet: progress in delivering information and services online. (London: The
Stationary Office). HC 529 Session 2006-2007, National Audit Office value for
money report. Also Web published at and at
Dunleavy, P., Margetts, H,. Raraty, D., Tinkler, J. (and with Dorrell, D., Escher, T.,
Goldchluk, S., Hinds, L., Khan, M.K., and Reissfelder, S.) (2009) Department for
Work and Pensions: communicating with customers (London: The Stationary Office).
HC 421 Session 2008-9. NAO ‘Value for money’ report. Also Web published at and at
Dunleavy, P., Bastow, S., and Tinkler, J., with Gilson, C., Goldchluk, S. and Towers,
E. (2010) ‘Joining up citizen redress in UK
central government’, in M. Adler (ed)
Administrative Justice in Context (London: Hart), Chapter 17, pp. 421-56.
Gawande, A. (2010) The Checklist Manifesto (London: Profile Books).
Greengrass, P. (2006) United 93 (Los Angeles: Universal Pictures), feature film.
Gronross, Christian (2007) Service Management and Marketing: Customer
Management in Service. (Chichester: Wiley).
Healthcare Commission (2007a) Investigation into outbreaks of Clostridium difficile
at Maidstone and Tunbridge Wells NHS Trust. (London: Commission for Healthcare
Audit and Inspection).
Healthcare Commission (2007b) Is Anyone Listening? A report on complaints
handling in the NHS. (London: Commission for Healthcare Audit and Inspection).
Healthcare Commission (2009) Investigation into Mid Staffordshire NHS Foundation
Trust. (London: Commission for Healthcare Audit and Inspection).
Hood C. (1998) The Art of the State: Culture, Rhetoric and Public Management.
Oxford: Oxford University Press.
Janis, I. (1982) Groupthink: Psychological Studies of Policy Decisions and
Fiascoes (Houghton Mifflin).
Johnson, T. (1972) Professions and Power (London: Macmillan).
Laming Review (2009) The Protection of Children in England: A Progress Report.
(London: The Stationary Office).
Le Grand, J. (2003) Motivation, Agency and Public Policy: Of Knights and Knaves,
Pawns and Queens (New York: Oxford University Press).
Lipsky, M. (1983) Street-level Bureaucracy (New York: Russell Sage Foundation).
Mayer, M.W. and Zucker, L.G. (1989) Permanently Failing Organizations(London:
Ministry of Justice (2007). Transforming tribunals: Implementing part 1 of the
Tribunals, Courts and Enforcement Act 2007. London: MoJ.
Mintzberg, H. (1983) Structure in Fives: Designing Effective Organizations
(Englewood Cliffs, NJ: Prentice Hall).
Mulcahy, L. (2003) Disputing Doctors: the socio-legal dynamics of complaints about
medical care. (Maidenhead: Open University Press).
Munro, E. (1996) Avoidable and unavoidable mistakes in child protection work.
British journal of social work, 26 (6). pp. 793-808.
National Audit Office (2006a) Child Support Agency implementation of the child
support reforms. HC 1174 Session 2005-06. (London: The Stationary Office).
National Audit Office (2006b) Improving the use of temporary nursing staff in NHS
acute and foundation trusts. HC 1176 Session 2005-06. (London: The Stationary
National Audit Office (2008) Department for Work and Pensions: Handling
Customer Complaints. HC 995 Session 2007-08. (London: The Stationary Office).
National Ombudsman of the Netherlands (2008) Annual Report Summary. (The
Hague: National Ombudsman).
PASC, Public Administration Select Committee (2003) ‘PASC attacks “disturbing”
evidence of government administrative failure’. Session 2002-3, Pres Notice No.8.
Available at:
Parliamentary Commissioner for Administration (2007b) Principles of Good
Administration. (London: The Stationary Office).
Parliamentary Commissioner for Administration (2007c) Principles for Remedy.
(London: The Stationary Office).
Perrow, C. (1986) Complex Organizations: A Critical Essay (McGraw Hill).
Power, M. (1994) The Audit Explosion (London: Demos).
Carol Propper, C and Van Reenen, J. (2008) ‘Can pay regulation kill? Evidence from
English hospital trusts’, Centre for Economic Performance discussion paper (London:
London School of Economics, CEP, 30 January).
Public Administration Select Committee (2008) When Citizens Complain. Fifth
Report of Session 2007-08. (London: The Stationary Office).
Quinn, B. (1992) Intelligent Enterprise: A Knowledge and Service Based Paradigm
for Industry (New York: Free Press).
Ravetz, A. (1982) Remaking Cities (London: Croom Helm).
Roberts, A. (2008) The Collapse of Fortress Bush : The Crisis of Authority in
American Government (New York : New York University Press)
Sampson, F. and de Silva, N. (2001) Police Conduct, Complaints and Efficiency.
(London: Blackstone Press).
Savoie, D. (1995) 'What is Wrong with the New Public Management?', Canadian
Public Administration, vol. 38, no.1, pp. 112–121.
Scott J. C. (1998) Seeing Like a State: How Certain Schemes to Improve the Human
Condition have Failed. New Haven: Yale UP.
Sunday Times, ‘Labour hid ugly truth about National Health Service (NHS)’, 7
March, 2010
Tax, SS and Brown, SW (2000) ‘Service recovery, research insights and practices’, in
TA Swartz and D Iacobucci (eds), Handbook of Services Marketing and Management,
Thousand Oaks, CA: Sage Publications, 271-285.
Thaler, R. and Sunstein, C. (2009) Nudge: Improving Decisions About Health, Wealth
and Happiness (London: Penguin).
t’Hart, P. (1994) Groupthink in Government: A Study of Small Groups and
Policy Failure (John Hopkins University Press, 1994)
Tuchman B. W. (1984) The March of Folly: from Troy to Vietnam. London: Michael
Varney, David. (2006) Service transformation: A better service for citizens and
businesses, a better deal for the taxpayer. (London: The Stationary Office).
Wilson, W. (1886) ‘The study of administration: an essay’. Accessed at, 25 March 2010.
... Regardless of cause, the need to address challenges to joined up service provision and find solutions remains pressing. The literature mostly focuses on "service delivery disasters" particularly in relation to children (Frost, 2005;Dunleavy, 2010;Dunleavy et al., 2010) with relatively little discussion of everyday challenges for workers on the coalface. ...
Full-text available
Purpose – This paper aims to describe coal-face challenges to making services in the UK work to ensure the mental and physical health, safety and wellbeing of children. Design/methodology/approach – After briefly referring to some challenges to effective joined-up service provision, it describes examples from the first author's experience of problems, during 30+ years as an NHS clinical child psychologist, and some solutions. It then describes two challenges that underpin many of these problems: lack of understanding of, or training in, evaluating evidence for interventions and a more general lack of knowledge about effective behaviour change principles. Findings – The paper concludes with recommendations about how to achieve effective joined-up services. Common themes emerging from the research are discussed, including choosing evidence-based programmes, providing adequate training to staff, and increasing people's understanding of behavioural principles. Originality/value – Having effective joined-up services would mean better services for parents and their children, and would be more cost-effective for the NHS. The ideas presented in this paper could also be applied to other services within the NHS.
... Despite the prevalence of such topics in practitioner discussions for many years now, 1 since at least the White Paper on Modernising Government (Cabinet Office, 1999), there is actually only a small research literature that more thoughtfully discusses these issues. 2 Most of these works focus on particular services or particular client groups where problems of 'service delivery disasters' have been most acute, as with children's services (Frost, 2005;Dunleavy et al, 2010). ...
Full-text available
Griffiths, Kippin and Stoker bring together many of the country’s leading academic and policy experts to explore the long-term challenges facing public services, and ask what the role of government, citizens and society should be in addressing them. The book sets out a new reform agenda, exploring possibilities for the future design and delivery of public services in the UK and beyond. Public Services: A New Reform Agenda is an important new contribution to the debate that will be invaluable for policymakers, practitioners and academics.
... In the UK, there has been a push to replace top-down national regulatory controls over local delivery in hospitals, schools and local governments with online customer feedback mechanisms (such as the large UK website NHS Choices) and transparency initiatives that open up expenditure data to public scrutiny. The hope is that instant patient and family feedback on hospitals will substitute for previous long-winded and after the fact regulatory investigations of problems, which conspicuously failed to prevent service delivery disasters [48]. Early Web science studies of such initiatives suggest that customer testimonials accurately identify hospitals with patient care problems [49], paralleling findings in the USA that Google searches for flu symptoms provide accurate advance indicators of the regional and local spread of flu cases, beating federal Centers for Disease Control and Prevention monitoring at least on timeliness [50]. ...
Full-text available
Widespread use of the Internet and the Web has transformed the public management 'quasi-paradigm' in advanced industrial countries. The toolkit for public management reform has shifted away from a 'new public management' (NPM) approach stressing fragmentation, competition and incentivization and towards a 'digital-era governance' (DEG) one, focusing on reintegrating services, providing holistic services for citizens and implementing thoroughgoing digital changes in administration. We review the current status of NPM and DEG approaches, showing how the development of the social Web has already helped trigger a 'second wave' of DEG(2) changes. Web science and organizational studies are converging swiftly in public management and public services, opening up an extensive agenda for future redesign of state organization and interventions. So far, DEG changes have survived austerity pressures well, whereas key NPM elements have been rolled back.
This paper develops a theory of the relationship between policy disasters and political institutions. Policy disasters, defined as avoidable, unintended extreme negative policy outcomes, are important political, and historical events above that receive relatively little attention from political scientists and scholars of public policy. Using the predictions of punctuated equilibrium theory, I argue that systems with higher error accumulation will experience more policy disasters. Systems with more veto players and weaker information flows will experience more policy disasters, but information flows will have a stronger impact than veto players. I test this theory using data on financial crises and natural and technological disasters across 70 countries over 60 years. I find strong evidence that systems with weaker information flows and more veto players tend to have greater policy disaster risk.
Conference Paper
Full-text available
Much research about large cities has focused on policy-level action for concerns such as infrastructure, basic amenities and education, treating the citizens as a collective of millions. We suggest an alternative, drawing on recent moves towards ‘digital era governance.’ We argue for and develop a foundation for the design and modeling of services that focus on individuals. Drawing on and extending prior work in service modeling and public-sector governance, we develop a formalism for modeling citizen-centric services, illustrate its application, and extract principles that underlie this effort. The paper concludes with pointers to other aspects of the larger iCity project, aimed at building smart cities in the world’s rapidly growing regions.
Full-text available
Policy-induced waste and forseeable fiascoes are serious and endemic problems of the British political and administrative systems. They can be traced to - the large-scale of British policy making; - a core executive preoccupation with speedy policy-making, giving inadequate scope for deliberation and legislative scrutiny; - a strong form of political hyper-activism, induced by the UK’s particular system of party competition; - the over-confidence of a talented but inexpert and overly centralizing administrative elite; and - the periodic failure of internal checks and balances inside the core executive.
When the Bush presidency began to collapse, pundits were quick to tell a tale of the "imperial presidency" gone awry, a story of secretive, power-hungry ideologues who guided an arrogant president down the road to ruin. But the inside story of the failures of the Bush administration is both much more complex and alarming, says leading policy analyst Alasdair Roberts. In the most comprehensive, balanced view of the Bush presidency to date, Roberts portrays a surprisingly weak president, hamstrung by bureaucratic, constitutional, cultural and economic barriers and strikingly unable to wield authority even within his own executive branch. The Collapse of Fortress Bush shows how the president fought-and lost-key battles with the defense and intelligence communities. From Homeland Security to Katrina, Bush could not coordinate agencies to meet domestic threats or disasters. Either the Bush administration refused to exercise authority, was thwarted in the attempt to exercise authority, or wielded authority but could not meet the test of legitimacy needed to enact their goals. Ultimately, the vaunted White House discipline gave way to public recriminations among key advisers. Condemned for secretiveness, the Bush administration became one of the most closely scrutinized presidencies in the modern era. Roberts links the collapse of the Bush presidency to deeper currents in American politics and culture, especially a new militarism and the supremacy of the Reagan-era consensus on low taxes, limited government, and free markets. Only in this setting was it possible to have a "total war on terrorism" in which taxes were reduced, private consumption was encouraged, and businesses were lightly regulated. A balanced, incisive account by a skilled observer of U.S. government, The Collapse of Fortress Bush turns the spotlight from the powerful cabal that launched the war in Iraq to tell a much more disturbing story about American power and the failure of executive leadership.
Service recovery: Research insights and practices Service recovery is now recognized as a significant driver of customer satisfaction and loyalty and an important component of a quality management strategy (Fornell and Wernerfelt 1987; Rust, Zahorik, and Keiningham 1996; Smith, Bolton, and Wagner 1998; Tax and Brown 1998). Performing very well in recovery can overcome disappointment and enhance relationships, whereas performing poorly can severely damage satisfaction, trust, and commitment and lead to customers switching service providers (Keaveney 1995; Smith and Bolton 1988; Tax, Brown, and Chandrashekaran 1998). This realization has led some firms (e.g., FedEx, Hampton Inn) to treat service recovery as an investment and deploy considerable resources in programs (e.g., service guarantees, employee training) and assets (e.g., customer call centers) to improve recovery efforts. This contemporary view of service recovery differs dramatically from the perspective held only a short time ago, and still practiced by many firms: that resolving complaints ...