ChapterPDF Available

CH 5 Growing the Productivity of Government Services] Chapter 5 How productivity can remain unchanged despite major investments


Abstract and Figures

Government agencies can achieve very rapid progress and productivity growth as this chapter shows by analysing the UK Customs agency, which introduced timely IT for handling imports and exports transactions and was able to handle sharply increased trade volumes with constant staff levels as a result.
Content may be subject to copyright.
5. How productivity can remain
unchanged despite major
investments – social security
Modern social security systems are the largest distributive counterpart
of the government’s capacity to raise taxes. Any welfare state is at root
a system in which resources are requisitioned from those in work or on
higher incomes, as well as from companies and from the well off, and then
redistributed to the elderly, sick and disabled people, those unlucky in
seeking work, and families or children in low- income households. Much of
this redistribution is achieved through public services, supplied to all citi-
zens by government agencies or by private organizations in ways that are
funded, regulated and shaped by government. However, most social secu-
rity transfer payments across the world are mediated in much less complex
ways. Especially in simpler and more ‘statist’ welfare systems (such as that
in the UK), monies are moved from the taxation department to a social
security department, whose officials then allocate state benefits directly
to eligible individuals or households. In more complex European welfare
states such direct government transfer payments are often smaller because
the state essentially supplements the social insurance that individuals have
themselves taken out with other voluntary or quasi- private organizations,
such as insurance funds, trade unions or social housing providers. But
even here, the non- governmental providers are almost always organized in
government- regulated schemes that are also underpinned in financial and
risk- assurance terms by taxpayers. In the USA, welfare provision remains
stubbornly partial and fragmented, dashing early post- war expectations of
a ‘positive service state’ (Roberts, 2010).
We begin by considering how the direct delivery of social security ben-
efits is characteristically managed in advanced industrial societies, in a
distinctive administrative format that allows societies to pursue national
conceptions of ‘social justice’ while still maintaining the rule of law. The
second section looks at the particular set up for social security in the UK,
especially three key changes made since 2000 – bringing together benefits
administration with the provision of employment services; the rebuilding
of services delivery around a phone- based model of communication with
M3049 - DUNLEAVY 9780857934987 PRINT.indd 120M3049 - DUNLEAVY 9780857934987 PRINT.indd 120 17/12/2012 09:1017/12/2012 09:10
Patrick Dunleavy and Leandro Carrera - 9780857934994
Downloaded from PubFactory at 07/28/2022 01:29:55PM
via free access
Social security 121
customers; and the transition to much more capital- intensive administra-
tive processes. Section 5.3 examines the record of productivity change in
UK social security, looking back two decades. We show that virtually no
growth has been achieved, despite a belated but otherwise quite exten-
sive modernization of services. Finally we seek to track down the factors
underlying this (basically static) outcome, linking it to the main invest-
ment and managerial changes pursued in the recent period.
In The Road to Serfdom (1944) the economic philosopher Friedrich Hayek
famously argued that vesting the ability to determine people’s incomes and
their feasible ways of life in a government bureaucracy must inherently
undermine the rule of law. It would create a slippery slope leading to an
over- powerful (even totalitarian) government, intrusively demanding full
information about the life choices and behaviours of a state- dependent
population. These people would have to be a compliant population,
because their economic and social welfare would now be determined by
the decisions (whims) of officials exercising power in discretionary ways.
Hayek’s (1944) conclusion was that any state attempt to operationalize a
concept of ‘social justice’ must end in undermining autonomous citizen-
ship and the rule of law, and with it any viable concept of democracy. In
every advanced industrial country, this conviction has remained an influ-
ential but minority position through to the present day. But the Hayekian
critique of administrative discretion has generally been rejected as exag-
gerated or over- fundamentalist. In most OECD countries convincing
electoral majorities have been built by political parties (stretching from
the political left through to the centre- right) committed to first construct
and then maintain and defend an extended social security system. Each
country has developed their own strong vision of the type of social justice
and welfare state that government should pursue.
In every welfare state, however, the Hayekian critique has strongly
shaped how social security administration has been designed, structured
and regulated. To minimize or eliminate officials’ discretionary power
there has been a dominant emphasis upon establishing clear systems of
rules governing legal entitlement to assistance, and on enforcing a highly
egalitarian, impartial and non- discretionary pattern of implementation.
National politicians have retained the powers to set benefit rates and
eligibility rules directly through legislation, and social security depart-
ments are normally regular ‘line’ ministries under full political control.
M3049 - DUNLEAVY 9780857934987 PRINT.indd 121M3049 - DUNLEAVY 9780857934987 PRINT.indd 121 17/12/2012 09:1017/12/2012 09:10
Patrick Dunleavy and Leandro Carrera - 9780857934994
Downloaded from PubFactory at 07/28/2022 01:29:55PM
via free access
122 Growing the productivity of government services
But implementation has been extensively delegated to bureaucracies of
various kinds.
There are partial variations around this pattern, especially where social
security arrangements are set up as long- term ‘trust funds’, as with the
US Social Security Administration (SSA) – its funds cannot be diverted
or used for other purposes by the presidential executive. But at the same
time, SSA operates under close congressional control. In some European
systems too, where state funding inter- leaves with the funded models of
voluntary sector social insurance providers, the social security apparatus
may operate with a greater degree of independence. But most welfare
states are still set up on a ‘pay as you go’ (PAYGO) basis, where the taxes
of current workers and companies are used to fund the current outgoings
– that is, the pension entitlements of previous generations of workers, and
income maintenance for people who are currently unemployed, ill or disa-
bled. If PAYGO commitments are allowed to rise then so must tax pay-
ments, and often on a literally massive scale – hence the general political
centrality of social security decision- making.
At the same time the ideal of impartial rule of law administration under
liberal democracy has meant that politicians are not allowed to intervene
directly or individually in the administration of benefits, beyond the
normal constituency role of legislators in assisting their voters. Instead
a voluminous set of legislation and administrative regulation has been
established that seeks to set out in precise detail how people are or are
not eligible for benefit, and how much they should consequently be paid.
Because social security systems aim to provide comprehensive support for
people in need, whose circumstances are very varied, the accompanying
rule books quickly mushroom in complexity, with many special condi-
tions and exemptions seeking to cope with apparent anomalies in previous
rules. But there is no social security counterpart to the enhanced occupa-
tional autonomy of doctors, teachers or even social workers. There is no
legitimate space where the individual professional discretion of an official
can displace the rule book’s provisions. The embedding of complex rules
in ICT systems increasingly fixes the complete legal specification of pay-
ments in machine code. Indeed numerous efforts have been made to con-
struct ‘expert systems’ to administer benefits and to minimize the need for
human judgement or intervention – so far without much success.
The push for impartiality creates a risk of creating a machine- like ben-
efits system, one that operates in ways that treat often fragile people in
considerable need without clear regard to their personal circumstances
and situations. Yet officials seeking for information may make mistakes.
And like anyone else, benefits agency employees may let their judgements
of the claimants they encounter colour the ways in which they respond,
M3049 - DUNLEAVY 9780857934987 PRINT.indd 122M3049 - DUNLEAVY 9780857934987 PRINT.indd 122 17/12/2012 09:1017/12/2012 09:10
Patrick Dunleavy and Leandro Carrera - 9780857934994
Downloaded from PubFactory at 07/28/2022 01:29:55PM
via free access
Social security 123
with consequent scope for systemic biases to emerge in the treatment of
different kinds of people.
Countries have pursued three basic strategies (that complement each
other) in order to counter this effect:
1. ‘Humanizing’ strategies for social security were established early on
in the 1930s and 1940s. They focus on socializing agency staff into
a strong ‘public service’ ethos, in which officials’ values and internal
reward systems centre in large part on doing a good job for the citi-
zens they deal with. A strong management overview of how staff treat
people, comprehensive training in treating claimants fairly and an
organizational culture that stresses a degree of empathy for clients’
needs are all characteristic of social security organizations. These ele-
ments serve to blur and soften the edges of their legal implementation
culture. Compared with tax agencies, for instance, social security staff
often interact with households frequently (instead of the tax agency
annual interactions). And they also may have smaller individual
caseloads to handle. Social security agencies all recognize that poor
or disadvantaged households are highly dependent on their decisions
for money if they are to survive from day to day: so they recognize an
obligation to ensure that their services facilitate solutions instead of
compounding clients’ problems.
2. Appeal, redress and legal challenge processes give a second chance
to claimants unhappy with the agency’s decision on their case. They
have a chance to ask for a second review by a manager, and after
that if they are still unhappy to appeal what they believe to be wrong
decisions to higher appeal bodies or to the law courts. In Europe the
emphasis tends to be on administrative tribunals, which operate less
formal independent review processes where ordinary citizens can
present their own cases. In the USA the solutions embodied in the
Administrative Procedure Act (1946) focused on providing internal
review by higher- tier officials across federal government, including
within the Social Security Administration. However, US practice has
also rather quickly and strongly developed from the 1960s towards the
pervasive legalization of social security disputes in more adversarial
ways, with agencies facing suits or damages claims in the event of legal
3. Improving ‘point- of- service’ standards to approximate those in private
sector service organizations has been an important development
of the public service ethos under the influence of the ‘new public
management’ (NPM) approach that dominated changes in public
administration in many OECD countries from the later 1980s to
M3049 - DUNLEAVY 9780857934987 PRINT.indd 123M3049 - DUNLEAVY 9780857934987 PRINT.indd 123 17/12/2012 09:1017/12/2012 09:10
Patrick Dunleavy and Leandro Carrera - 9780857934994
Downloaded from PubFactory at 07/28/2022 01:29:55PM
via free access
124 Growing the productivity of government services
the mid- 2000s. The ‘managerialist’ focus here was on moving away
from rundown buildings; getting rid of block bookings of face- to-
face customers with long queues and wait times; radically modern-
izing old- fashioned phone- handling procedures; and minimizing the
most expensive interactions (face- to- face interviews). Instead service
premises could be modernized to private- sector- like standards; clients
would be booked only to individual appointment times, with queuing
kept to a minimum; modern, high- capacity and web- enabled call
centres would be established to handle much larger volumes of phone
traffic more efficiently; and the need for face- to- face interactions with
‘clients’ would be cut to a minimum.
In comparative public policy terms the British welfare state has long been
something of a halfway house between the more laissez faire minimal-
ism of the USA and the generous income- replacement levels provided
by social security in the older EU member states. The traditional UK
approach to social security is statist, integrated, universalistic and highly
centralized (Dunleavy, 1989a, 1989b). It is also characteristically mean in
comparison with the level of income replacement made by welfare benefits
in other European countries. For many decades the UK old age pension
has typically paid no more than a third of median weekly earnings, com-
pared with levels around 60 per cent to two- thirds in Germany, France or
Scandinavian countries. Similarly, the UK system provides newly unem-
ployed people with a benefit rate that is just 13 per cent of the average
weekly earnings, compared to replacement rates that are more than four
times as generous in the same set of European countries.
On the other hand, the UK system still (just) retains the attempted com-
prehensiveness of the original Beveridge Report dating from 1944, with its
aspiration of providing ‘cradle to the grave’ coverage against ill luck and
the adverse contingencies of unemployment, illness, disability and old age.
Thus the UK welfare state provides a far more multi- pronged system of
social protection than the minimal insurance against unemployment, short
illness and the basic State Pension that the Social Security Administration
in the USA delivers. Table 5.1 shows that the British system is far more
comprehensive than the US welfare state in recognizing an obligation on
government to provide money benefits to prevent people from becoming
homeless, going hungry or coping with illness, disability or disastrous
personal life choices. As a result, Table 5.1 also shows that the annual
M3049 - DUNLEAVY 9780857934987 PRINT.indd 124M3049 - DUNLEAVY 9780857934987 PRINT.indd 124 17/12/2012 09:1017/12/2012 09:10
Patrick Dunleavy and Leandro Carrera - 9780857934994
Downloaded from PubFactory at 07/28/2022 01:29:55PM
via free access
Social security 125
Table 5.1 The main social security benefits in the UK welfare state
(andthose covered for the productivity analysis in section 5.3)
Category Annual
Spending in
(£ billions)
(%) of 2008
Specific Names of
Related UK Benefits
Involved Included in
Productivity Analysis
Benefits for unemployed
people with insurance
contributions (i.e.,
National Insurance in UK)
1867 1.5 Unemployment Benefit to
1996, and contributory
Jobseeker’s Allowance
Benefits for unemployed
people without insurance
8759 6.8 Income Support
for the unemployed
(from 1988 to
1996), thereafter
non- contributory
Jobseeker’s Allowance
Benefits for working age
people who are long-
term ill
6575 5.1 Incapacity Benefit
(before 1995, data
include Invalidity and
Sickness Benefit)
Other benefits for working
age people
932 0.7 Other benefits: Maternity
Allowance and Widow
Loans to assist working
age people on welfare
to purchase consumer
durables etc.
2399 1.9 Social Fund Grants
and Loans (introduced
in 1993)
Sub- total: Payments for
working age people not
in work
20 532 16.0
Old age pension based on
57 366 44.7 Basic State Pension
State Earnings Related
Pension Scheme
War Pensions (paid
separately by the
Ministry of Defence)
Assistance for elderly
people with low household
7227 5.6 From 1993 to 2000
Income Support
for people aged 60
and over; Minimum
Income Guarantee up
to 2003; from 2003
onwards, Pension
M3049 - DUNLEAVY 9780857934987 PRINT.indd 125M3049 - DUNLEAVY 9780857934987 PRINT.indd 125 17/12/2012 09:1017/12/2012 09:10
Patrick Dunleavy and Leandro Carrera - 9780857934994
Downloaded from PubFactory at 07/28/2022 01:29:55PM
via free access
126 Growing the productivity of government services
value of all UK social benefits paid to claimants in 2008 amounted to
£129 billion, and total government social security expenditure amounted
to almost £137 billion, that is, 12 per cent of the country’s gross domestic
product (HM Treasury, 2009). The rows in italics towards the bottom of
Table 5.1 (continued)
Category Annual
Spending in
(£ billions)
(%) of 2008
Specific Names of
Related UK Benefits
Involved Included in
Productivity Analysis
Additional assistance to
elderly people needing
help to get around or
cope with daily living
4440 3.5 Attendance Allowance
Assistance for disabled
elderly people
9809 7.6 Disability Living
Allowance (from 1993)
Assistance for people
looking after the elderly
or disabled
2173 1.7 Invalid Carers’
Allowance, later
Carers’ Allowance
International Pension
200 0.2 Paid to UK citizens
living overseas
Sub- total: Payments to
elderly people and the
81 215 63.2
Assistance with meeting
local housing rents
15 745 12.3 Housing Benefit
(administered by local
authorities, repaid
from DWP budget)
Assistance with meeting
council tax payments
4124 3.2 Council Tax Benefit
Sub- total: Payments for
anyone eligible on welfare
19 869 15.5
Future Pension Forecasts 345 0.3 Future Pension
Tax- free payment for each
6508 5.1 Child Benefit (paid by
Sub- total: Benefits paid and
services given to all
eligible people, whether on
welfare rolls or not
6853 5.3
Total benefits spending 128 469 100.0
Note: The italicized benefits are not included in the analysis below.
M3049 - DUNLEAVY 9780857934987 PRINT.indd 126M3049 - DUNLEAVY 9780857934987 PRINT.indd 126 17/12/2012 09:1017/12/2012 09:10
Patrick Dunleavy and Leandro Carrera - 9780857934994
Downloaded from PubFactory at 07/28/2022 01:29:55PM
via free access
Social security 127
Table 5.1 show that there has also been a recent growth of tax credits for
working families and for children, adding to the substantial payment of
universal Child Benefit, until 2012 paid to all households with children
(whatever their income levels). All tax credit payments are undertaken by
HM Revenue & Customs (HMRC), with credits especially closely linked
to the tax system (see Chapter 4). But this still accounts for less than a fifth
of UK welfare state spending. The remaining monies (more than four-
fifths) route through another integrated large ministry at central govern-
ment level, the Department for Work and Pensions (DWP). It is on these
that this chapter concentrates.
The department is a classic ‘transfer agency’ in terms of the ‘bureau-
shaping’ typology (Dunleavy, 1991, Ch. 7). The total administrative
costs for social security in 2008 were £7.5 billion, making this the largest
administrative cost borne by any government sector. (The tax depart-
ment HMRC came second at £5 billion.) Personnel numbers in the DWP
peaked at over 131 000 in 2002, more than a quarter of the whole central
government civil service. They then fell back to a low of just over 105 000
staff by 2008 – before substantially expanding again in 2009, reflecting the
impact of recession in raising unemployment levels and benefit claimant
numbers (see page 145). Although this is a massive administrative under-
taking, it is worth stressing that running costs in the DWP accounted for
only 4.4 per cent of its total budget in 2008, closely in line with bureau-
shaping expectations for a welfare state transfer agency.
The policy and organizational context for delivering social security has
changed extensively in the period since 1988, when useful data begin to
be available. The timeline in Figure 5.1 summarizes some key influences,
including the following:
A rapid succession of legislative changes (32 separate Acts in 20 years).
A frequent turnover of top ministers (12 times in the period) despite
one single change of government in 1997 from Conservative to
Labour (even for the UK, this is an extreme case of short- termism).
– Main policy programme changes in this area always centred on the
introduction of new benefits and the phasing out of old ones, usually
by new ministers.
Changes in the macro- organizational architecture for administering
social security benefits, especially the merger of two previously sepa-
rate networks of offices to form a single Department for Work and
The extensive use of ICTs and contractors formed a strong dynamic
in the administrative development of social security, with a major
outsourcing in 1992, renewed in more comprehensive ways in 1998.
M3049 - DUNLEAVY 9780857934987 PRINT.indd 127M3049 - DUNLEAVY 9780857934987 PRINT.indd 127 17/12/2012 09:1017/12/2012 09:10
Patrick Dunleavy and Leandro Carrera - 9780857934994
Downloaded from PubFactory at 07/28/2022 01:29:55PM
via free access
DHSS splits
4 new Exec Agencies 9000
new staff (12% increase)
AA, FC, IS, Child
5 Social Security Acts
Child Support / JSA
ES to DfEE
Pensions Act
Incapacity benefit
AFFINITY strategic partnership with EDS 10 yrs
EDS absorbs 2000 ITSA staff
Growth in contact centres to 6000 staff
NI Contributions to IR
Appeals Service
New Contract for Welfare
Green and White Papers
New Deal/Sure start Future New Deal for Young and Disabled
Further expansion of Sure Start centres
Simplify Child Support
ES + DSS = DWP established
BA + ES = JCP (£2.2 bn)
Gershon DWP targets
Cut 30 000 staff (14 000 JCP)
Efficiency savings £960 m per annum
Pension credit
DSS/DWP e-strategy CSA IT crisis
Pension Service/DCS
5 Social Security Acts
Welfare Reforms and Pensions Act
SERPS crisis £2.5 bn
Reform Act
CS, Pensions and Soc Sec Act
CSA axed-
new NDPB
PRIME PFI contract for DSS estates (target to cut by 17% m2)
DSS computer centre
outsourced to EDS DSS data
centre contracts
ITSA staff to EDS
ITSA market test
Pension credit IT problems
ES created
Dip in % of working age
people in work from mid to
low 70%
18–24 yr unemployment drops
by 75% to 50 000
Income Support claimants around 3.9 m Income Support down to 2.1 m
SEMA BPO on medical assessments
10 Social Security Acts
3 Employment Acts
2 Child Support/Maintenance Acts
Figure 5.1 Main developments in the administration of UK social protection, 1988–2008
Note: In Figure 5.1, the abbreviations used are: ACCORD consortium of computer firms; AA Attendance Allowance; BA Benefits Agency;
BAPOCL Benefits Agency/Post Office Counters Limited, joint programme; BPO business process outsourcing; CSA Child Support Agency;
DfEE Department for Education and Employment; DHSS Department of Health and Social Security; DCS Disability and Carers’ Service; DLA
Disability Living Allowance; DSS Department of Social Security; EDS major computer firm; ES Employment Service; IS Income Support; ITSA
Information Technology Services Agency (within DSS); JCP Job Centre Plus; JSA Job Seekers’ Allowance; MIG Minimum Income Guarantee;
NDPB non- departmental public body; NI National Insurance; NICO National Insurance Contribution Office; NIRS National Insurance Register
System; OPRA Occupational Pensions Regulatory Authority PFI Private Finance Initiative; PRIME property sale and re- lease deal; Sema major
computer firm; SERPs State Earnings Related Pension Scheme, a top- up pension; STEPS acronym for property transfer consortium.
M3049 - DUNLEAVY 9780857934987 PRINT.indd 128M3049 - DUNLEAVY 9780857934987 PRINT.indd 128 17/12/2012 09:1017/12/2012 09:10
Patrick Dunleavy and Leandro Carrera - 9780857934994
Downloaded from PubFactory at 07/28/2022 01:29:55PM
via free access
Social security 129
Increased modernization spending by Labour ministers on PFI new
building contracts (around 5 to 7 per cent of administrative costs per
year) and consultants (around 2 per cent of administrative costs per
year), responding to the previous chronic under- investment under
Conservative governments.
An attempted transition of DWP administrative processes from
paper- based forms and face- to- face contacts and towards phone-
based customer contacts, all underpinned by a massive new invest-
ment in IT systems and in new buildings.
The Department of Work and Pensions was itself created in April 2002
by merging together two pre- existing large bureaucracies: the Department
of Social Security, a cabinet ministry whose role consisted solely of assess-
ing and paying out claims for welfare benefits; and the Employment
Services Agency, a large but discrete part of a separate Whitehall depart-
ment, covering education and employment policies. We discuss each of
these ‘parent’ organizations in turn, before looking at how DWP itself
developed after its creation.
The Department of Social Security
The Department of Social Security (or DSS) was formed in 1988 from
the de- merger of a previous ‘super- ministry’ that had linked together
Health and Social Security for two decades. Under the ‘Next Steps’
reorganization of Whitehall begun in 1988 (see Chapter 4), DSS was
internally separated into four main agencies separated on functional
The largest component was the Benefits Agency, whose job was
to pay out social security benefits payments, especially pensions,
support for ill and disabled people and ‘safety net’ support for
unemployed people not entitled to unemployment pay (run by a
separate ministry, see below). Essentially the agency was the ‘oper-
ating core’ of DSS in Mintzberg’s (1983) terms (i.e., the part of the
organization that really set its identity and carries out its key mission
functions). It maintained the department’s local office network. At
its peak size it employed nearly 70 000 staff.
The Contributions Agency had the job of monitoring National
Insurance (NI) payments by employees (the UK version of a social
security tax), and then passing the right information to Inland
Revenue, who actually organized the collection of NI contributions.
In 1999, after a series of embarrassing IT contract mistakes, and a
M3049 - DUNLEAVY 9780857934987 PRINT.indd 129M3049 - DUNLEAVY 9780857934987 PRINT.indd 129 17/12/2012 09:1017/12/2012 09:10
Patrick Dunleavy and Leandro Carrera - 9780857934994
Downloaded from PubFactory at 07/28/2022 01:29:55PM
via free access
130 Growing the productivity of government services
failure of service covering several months, the agency and its 10 000
staff were completely transferred away from DSS to Inland Revenue.
The third main element of DSS was ITSA, the Information
Technology Services Agency, which ran the large DSS computer
centres and national IT systems. Most of its staff were transferred
to private contractors when the department’s IT functions were
outsourced in 1999.
Finally, the badly misnamed Child Support Agency (CSA) was
created as an executive agency in 1993. It started life as a part of
DSS in 1981 when Margaret Thatcher’s government began chasing
divorced fathers with obligations to make maintenance payments to
dependent children. Operating very much as a semi- detached part of
the department, from the outset its effectiveness was controversial,
with administration costs commonly absorbing more than half of
the sums recovered from fathers, and with repeated instances of IT
and administrative crises. Efforts were made to refashion CSA’s
legislative basis and method of operating, and its IT systems were
renewed, without much improvement. In 2008 the organization
ceased to be an executive agency, converting into the delivery arm
of a shortlived body (the Child Maintenance and Enforcement
Commission), which was abolished in 2012 and its functions trans-
ferred to the Child Maintenance Group (CMG) within DWP itself.
CSA lives on in a hollowed- out form (with no autonomy) as the
body doing implementation for CMG.
Most of the early DSS period was taken up with the development of its
‘Operational Strategy’ to computerize social security records and office
processes, moving away from a wholly manual set up with paper files and
records, where the administration of income support for the unemployed
and sickness/disability benefits was handled in hundreds of local offices
(Margetts, 1991; Margetts, 1998, Ch. 3). More centralized structures were
created for benefits administration, with computer records taking over
from paper files. Local offices and paper forms were retained for inter-
acting with customers, but the recording of decisions was moved onto
computers. The Department chose as its contractor, EDS, the dominant
system integrator firm in the UK at this period. Later the same company
supplied the IT system installed for the Child Support Agency.
The strategy was planned to install computer networks and terminals
and to last 15 years, but the whole idea of a 15- year computerization strat-
egy was vitiated by the rapid development of IT, including the launch of
PC- based systems (Margetts, 1998). Cost overruns also occurred and the
efficacy of the new systems was queried by a number of Parliamentary
M3049 - DUNLEAVY 9780857934987 PRINT.indd 130M3049 - DUNLEAVY 9780857934987 PRINT.indd 130 17/12/2012 09:1017/12/2012 09:10
Patrick Dunleavy and Leandro Carrera - 9780857934994
Downloaded from PubFactory at 07/28/2022 01:29:55PM
via free access
Social security 131
select committee hearings and National Audit Office (NAO) reports. The
Operational Strategy was more or less halted in 1999, after the expendi-
ture of some £1.75 billion over 17 years. Its achievement was clearly to
accomplish the long- delayed computerization of the benefits systems, but
its legacy in terms of locking the department into outdated solutions and
technologies (endorsed by both senior officials and their main contractors)
is also hard to underestimate.
When we visited a series of Benefits Agency offices in 1999 on an NAO
investigation, its 68 000 staff were equipped with just eight PCs capable
of accessing the internet. None of its over 400 local offices had any access
to PCs, or any knowledge of websites (Dunleavy et al., 1999, Part 2). The
Department did have a small ‘posterboard’ website (of very low function-
ality), which was actually built by a couple of low- ranked officials from a
Yorkshire office in their spare time. One of their main tasks in 1999 was
to receive and collate the more than 200 e- mail queries a month that DSS
received from UK pensioners overseas, print them out and then send them
to the appropriate paper forms office. The relevant NAO report concluded
that the department’s websites ‘have been relatively static and under-
developed, cautiously funded and under- resourced, providing information
within conservative and unimaginative designs’ (Dunleavy et al., 1999,
paragraph 2.31).
The working model of the department varied across its main benefits
areas. The administration of pensions was generally a low- cost activity,
costing on average around £0.5 to £0.6 per client per week in the late
1990s (Margetts, 1999). This was unsurprising since many pensioners were
paid the same basic pension for many years, and rarely moved address.
However, paying benefits to working age people who were unemployed,
sick or disabled then cost around £5.50 per person per week. These clients
had far more changes of circumstances, switches of address and a much
higher incidence of new claims to continuing claims, all of which greatly
raised costs. Maintaining a comprehensive network of local benefits
offices to see clients in person was also costly, but DSS officials (and min-
isters) saw it as an essential check on who was receiving the higher- risk
benefits, like Income Support.
Throughout its period of operation, under the Conservative govern-
ments of Margaret Thatcher and John Major, the political salience of
the DSS was generally rising, albeit from a low ebb. Influenced by US
developments, Tory ministers embraced a shift away from the automatic
granting of benefits to unemployed people and towards more of a ‘work-
fare’ orientation, in which jobless people were expected to demonstrate
that they were active in seeking work. Conceivably benefits could be
withdrawn if jobless individuals refused to accept viable work that was
M3049 - DUNLEAVY 9780857934987 PRINT.indd 131M3049 - DUNLEAVY 9780857934987 PRINT.indd 131 17/12/2012 09:1017/12/2012 09:10
Patrick Dunleavy and Leandro Carrera - 9780857934994
Downloaded from PubFactory at 07/28/2022 01:29:55PM
via free access
132 Growing the productivity of government services
offered to them. Initially this ‘stricter benefits regime’ stance was criti-
cized by Labour and the trade unions. But in the later 1990s Labour too
swung towards a strong endorsement of a humanized workfare look- alike
approach – one that promised job seekers much stronger assistance and
retraining in looking for work, in return for a responsibility on their part
to move off benefits as soon as they could do so. The Blair government
was influenced here by the active labour market policies of the Swedish
welfare state. But the Labour vision also could not be achieved simply
within the confines of the DSS brief, because the new vision was centrally
linked to employment policies.
The Employment Services Agency
The Employment Services Agency (ESA) in 1988 formed a rather separate
and discrete part of the Department of Employment. Its origins can be
traced back to the 1909 decision to create Labour Exchanges where local
jobs would be advertised by employers and the government would try to
bring a degree of order to chaotic local labour markets. Labour exchanges
helped jobless people to more quickly and easily find work by broadening
their options and widening their job searches, with staff assistance. Later
renamed Employment Exchanges and run by the Ministry of Labour,
these local offices also took on the payment of unemployment benefits for
jobless people looking for work and who had made the requisite contri-
butions when in work to be entitled to receive benefits – called National
Later on the parent department was renamed the Department of
Employment, and in 1973 its network of local offices were renamed Job
Centres. In 1988 the Employment Services Agency was created as part of
the ‘Next Steps’ agencification process (see Chapter 4), where staff were
moved out of main Whitehall ministries and into executive agencies. It was
now run on more corporate lines, and operated Job Centres in most towns
and cities. Finally in 1995, the Conservative government decided to merge
the parent department of the Job Centre service with Education in a short-
lived Department for Education and Employment (DfEE, which lasted in
this form for just six years).
The Job Centre network was widely rated as a success until the advent
of the 1981–85 recession, when it came under acute pressure because of the
growth of mass unemployment. The department did succeed in somewhat
updating its services and assisting job seekers more, but the linkage with
DSS services was always problematic. Unemployment benefits (based on
past National Insurance contributions) were paid for up to the first year
that people were out of work. Thereafter a jobless person would need
M3049 - DUNLEAVY 9780857934987 PRINT.indd 132M3049 - DUNLEAVY 9780857934987 PRINT.indd 132 17/12/2012 09:1017/12/2012 09:10
Patrick Dunleavy and Leandro Carrera - 9780857934994
Downloaded from PubFactory at 07/28/2022 01:29:55PM
via free access
Social security 133
to move onto a different benefit (called Income Support) paid by DSS,
transferring across from one local office to another, often several miles
apart. Long- term unemployed people, or those who did not have the right
National Insurance contributions, would not need to visit Job Centres to
receive benefits, and so might not need any assistance with job seeking.
Nor was it easy for DSS staff to check if job seekers were actively looking
for work. Expertise in assisting with job searches and knowledge of local
labour markets also lay with the ESA, and not with DSS staff.
In 1997 the Labour government under Tony Blair came to power
pledged to introduce active labour market policies with a much stronger
administrative push to get non- working people into employment (UK
Government, 1998). Integrating service delivery across the DSS and DfEE
boundary line became essential to the government’s ‘New Deal’ pro-
grammes for young unemployed people, for lone parents, the long- term
unemployed and the disabled – all of which stressed New Labour’s ‘rights
and responsibilities’ agenda. The ‘stricter’ Jobseeker’s Allowance (JSA)
regime of the Conservatives was now supplemented by ‘guaranteed’ inten-
sive employment assistance after a particular duration of unemployment
(Willetts et al., 2003).
There were two possible ways of resolving this conundrum, however:
(1) to move the administration of all social security benefits for working
age people across from the DSS and into the employment ministry DfEE,
thereby downgrading DSS into something like the old Ministry of Pensions
that existed in the 1960s; or (2) to move the Employment Services Agency
across from DfEE into DSS, thereby creating a beefed up Job Centre
operation that would now include not just Income Support for the jobless,
but also all the job search assistance and administration of contributory
Jobseeker’s Allowance. Naturally enough ministers in DfEE fought hard
for option 1, while DSS ministers argued strongly for option 2.
In 1999 the civil service began to trial joint workings between the two
departments at the local level. Called the ‘One’ pilot, this experiment
rapidly spread to more regions and localities and was quickly judged so
successful that the pilot was never formally evaluated, but a decision was
made to merge the operations of ESA and DSS for working people. A
Whitehall inter- departmental committee spent two years comprehensively
examining the whys and wherefores of creating a single new department
bringing together all benefits and advice services for working age people in
a service to be called Jobcentre Plus (JCP). The new agency would break
away from the old, traditional images of ‘labour exchanges’ and provide
a radically improved standard of customer service. A small working party
was set up to create a blueprint for Jobcentre Plus, chaired by the Treasury
second in command and including DSS and DfEE ministers and senior
M3049 - DUNLEAVY 9780857934987 PRINT.indd 133M3049 - DUNLEAVY 9780857934987 PRINT.indd 133 17/12/2012 09:1017/12/2012 09:10
Patrick Dunleavy and Leandro Carrera - 9780857934994
Downloaded from PubFactory at 07/28/2022 01:29:55PM
via free access
134 Growing the productivity of government services
officials. Eventually it recommended that JCP should be created within
DfEE (option 1 above). However, the Prime Minister deferred a final
decision until after the 2001 general election (White and Dunleavy, 2010).
After Labour won, the detailed dynamics of his cabinet- making led Blair
to decide that option (2) should prevail. ESA would switch across from
DfEE (which was now actually abolished in its old form) to DSS, so as to
create a new Department for Work and Pensions.
The Department for Work and Pensions
The Department for Work and Pensions (DWP) formally began work
in April 2002, but in fact it had already existed in proto- form for more
than a year. It was perhaps the best prepared, longest incubated and
most expensive departmental merger in UK central government, ever.
The centrepiece of the reorganization was the creation of Jobcentre Plus
(JCP) services offering unemployed, disabled and sick people of working
age a more intensive regime of employment assistance. There were to be
frequent checks to ensure that claimants were actively seeking work, or
were using training opportunities, and that they were not simply languish-
ing on benefits. Single JCP offices in town centres and accessible locations
were created where all claimants had to register in person that they were
seeking work and check in every two weeks for follow- on interviews, if
they wanted to continue receiving benefits. Once jobless people reached six
months and then a year out of work, more intensive interviews and assist-
ance packages were mandated.
All the Job Centre buildings were remodelled nationwide from scratch,
getting rid of queuing systems, screens and rundown premises. They were
quite rapidly replaced with modern- looking, open plan offices where all
claimants had to turn up for precisely timed interviews, and with security
guards on hand to protect staff and to ensure that no one was admitted
before their appointed time. The old- fashioned cards advertising vacancies
were replaced by computerized ‘kiosks’ showing vacancies on screens and
allowing some details to be printed. There were some dedicated phones
for ringing up potential employers, but not a single PC was available for
use by unemployed people across the 850 local offices, a situation that
persisted until 2010 (Dunleavy et al., 2009).
The mature JCP model, which applied throughout the later 2000s, still
had very complex processes to administer the benefit, described in detail
in Dunleavy et al. (2009, Part 2). Jobless people claiming benefits had to
ring up a DWP call centre for around 40 minutes and explain their situ-
ation and details to the contact person. They then ended up with a local
JCP appointment, where they needed to show up at the Job Centre with
M3049 - DUNLEAVY 9780857934987 PRINT.indd 134M3049 - DUNLEAVY 9780857934987 PRINT.indd 134 17/12/2012 09:1017/12/2012 09:10
Patrick Dunleavy and Leandro Carrera - 9780857934994
Downloaded from PubFactory at 07/28/2022 01:29:55PM
via free access
Social security 135
their documentation to establish their identity, and be given advice by a
JCP local worker on looking for work. However, the person seeing them
face to face did not make a decision on their benefits – instead the JCP
staff only completed a computerized form that was next sent on to one of
77 Benefit Decision Centres organized at (small) regional levels. Here the
information relayed by the call centre and the local JCP officials was used
to make a decision. Information was lost and had to be repeated at each of
these stages, and the efficacy of the DWP computer systems and networks
was so poor that 40 per cent of the time information did not transfer in a
satisfactory or timely way from the call centre to the local JCP offices, or
from JCP to the remote decision centres (Dunleavy et al., 2009, Part 2).
In addition, the computer kiosks that seemed such a good idea in
2002 were pretty much obsolescent by 2009. On our visits to Job Centres
during late 2008 they were being little used. By contrast in the equivalent
Australian service (called Centrelink) there were hundreds of proper PCs
available to jobless people for them to search for jobs and e- mail applica-
tions. In the UK there were none at all: unemployed people had to be sent
to local libraries or other online access centres to do such tasks, making
it very unlikely that JCP officials could monitor whether they did or not.
Similarly, in the Access Canada service set up in the later 2000s there was
a strong emphasis on clients or customers sitting on the same side of the
desk as those helping them and looking at the same information on com-
puter screens. But in every JCP office we visited during 2008, staff sat on
the other side of desks from claimants, and could not turn their screens to
show clients what jobs were on offer or how to operate web- based appli-
cations. Yet by this time Britain’s largest employer (the NHS) was only
accepting e- mail applications for any of its positions.
A large part of the estimated direct costs of integrating the ESA
employment services into the new department arose from pay differentials
between the two departments, which cost a minimum of £143 million to
equalize over the years after 2002 (White and Dunleavy, 2010). In addi-
tion, the convulsion of JCP services clearly produced a productivity dip,
with staff and managers focusing some of their attention on protecting
and enhancing their own positions in the complete reorganization of
services for working age people. Using a metric derived from private
sector mergers and acquisitions, White and Dunleavy estimate the short-
term costs of lower productivity at £166 million. The separate and long-
standing Employment Services Agency organizational culture was also
imperfectly absorbed in the new hybrid DWP, causing some staff to leave,
and others to retire early.
The other two sections of DWP covered pensions and services for the
elderly and those for the disabled and carers. They were also affected by
M3049 - DUNLEAVY 9780857934987 PRINT.indd 135M3049 - DUNLEAVY 9780857934987 PRINT.indd 135 17/12/2012 09:1017/12/2012 09:10
Patrick Dunleavy and Leandro Carrera - 9780857934994
Downloaded from PubFactory at 07/28/2022 01:29:55PM
via free access
136 Growing the productivity of government services
extensive change processes. A 2001 review of strategic options for the new
department concluded that it badly needed to modernize its services, and
that the key way to do so cost- effectively was to transition away from
expensive face- to- face services for elderly and disabled customers, and
also to move on from paper form applications. Instead, the DWP’s whole
ways of communicating with its elderly and disabled customers would
be revolutionized, by using phone- based administrative processes. Huge,
modern call centres would be built and phone- based applications brought
in, especially for modernized benefits.
The new benefit, Pension Credit, was launched in 2003 to replace
Income Support and supplementary provision for elderly people with poor
National Insurance records and hence only small State Pensions. The new
credit aimed to guarantee anyone aged over 60 a defined weekly income.
To register, people phoned up a call centre and went through a long- ish
conversation with an advisor who filled in a complex form on computer
as they spoke. Staff aimed to give callers a decision on their entitlement
more or less then and there, at the end of the conversation. Yet when we
visited several Pension Credit call centres in 2008 we found that barely one
in four of the lowest- income applicants for the new benefit could be given
an immediate decision, called ‘sunny day’ cases by the staff (Dunleavy
et al., 2009, Part 3). The vast bulk of cases involved applicants submit-
ting paper documents and additional information about their housing,
employment status or bank accounts, which always greatly extended the
decision- making process, often for four to six weeks, and caused many
applications to lapse and have to be restarted. This was mainly due to
the supplementation characteristic of this benefit, which required that
a claimant’s earnings and savings were first assessed to determine how
much extra income they should get. The DWP computer systems set up to
optimize phone applications also worked in a very cumbersome way with
conventional paper applications – yet a third of elderly people preferred
to start their application off in this way, printing off forms from the inter-
net or getting copies from charities. IT limitations also meant that paper
forms were no quicker to process than the long call centre conversations.
Experienced DWP grassroots staff often evaded the newer, screen- based
IT systems that took so long to fill in, in order to complete applications
more quickly directly on the older mainframe computers (dating from the
Operational Strategy era).
Meanwhile in some other areas of DWP, our visits in 2009 found some
administrative functions operating essentially unchanged from 1970s’
processes. DWP paid an important benefit for sick or constrained elderly
people who needed extra help with day- to- day household activities like
washing, dressing, cooking and going to bed. This was called (mislead-
M3049 - DUNLEAVY 9780857934987 PRINT.indd 136M3049 - DUNLEAVY 9780857934987 PRINT.indd 136 17/12/2012 09:1017/12/2012 09:10
Patrick Dunleavy and Leandro Carrera - 9780857934994
Downloaded from PubFactory at 07/28/2022 01:29:55PM
via free access
Social security 137
ingly) Attendance Allowance (AA). In a 2003 study of ‘difficult forms’
for the NAO this form was by far the longest, most complex and most
complained about government form (Dunleavy et al., 2003). As result of
the NAO study, the AA form was eventually simplified and the number
of questions greatly reduced, but our 2009 study (Dunleavy et al., 2009)
found that it still took elderly people at least three hours to complete, and
was almost impossible for them to fill in without extensive assistance from
families, carers or case workers. All AA applications were on paper, even
though very sensitive and hard- to- explain matters were involved, and
processing and decisions took several weeks.
The DWP decision to remodel the department around phone- based
processes was based on analysis of survey responses in 2001–02 that showed
that far fewer DWP customers were using PCs or had internet access than
in the population at large. As a result, throughout the noughties there were
no developments at all undertaken to put core DWP transaction services
online. There were two rather separate exceptions, a long way removed
from benefits- based transactions. First, the old ESA labour market services
had begun to move online in the late 1990s, before the formation of DWP,
because there was already an evident demand from businesses and employ-
ers to have a web- based job vacancies service. This ESA- initiated project
developed rapidly in the noughties. By 2007 the now JCP labour market
systems were carrying details of 40 per cent of all job vacancies across the
country. At first vying with private sector services, and often criticized by
major private competitors at that time, the DWP provision was eventually
guaranteed and stabilized against being outsourced by an EU ruling that
all member states must deliver web- based employment services for their
populations. Later the key JCP site developed fruitful information sharing
of huge blocks of vacancies with its main private sector rivals, enhancing
their mutual effectiveness (Dunleavy et al., 2007, p. 6).
Second, the DWP’s provision of web- based information about benefits
slowly improved, from its ineffectual 1999 beginnings (discussed above).
A 2002 study by NAO following up on government departments’ website
provision found that DWP was still lagging behind in terms of its depart-
mental website, with incomplete, inconsistent information presented in
complex and inaccessible ways (Dunleavy et al., 2002). The 2003 study
of ‘difficult forms’ followed up by demonstrating that DWP websites still
offered a poor access route for citizens seeking information about their
eligibility for benefits.
A cabinet sub- committee reviewed the cross- departmental picture of poor
website development and in 2007 concluded that a new strategy was needed,
focusing on creating two new government ‘super- sites’. The ambition here
was for the central government to replicate some of the BBC’s (British
M3049 - DUNLEAVY 9780857934987 PRINT.indd 137M3049 - DUNLEAVY 9780857934987 PRINT.indd 137 17/12/2012 09:1017/12/2012 09:10
Patrick Dunleavy and Leandro Carrera - 9780857934994
Downloaded from PubFactory at 07/28/2022 01:29:55PM
via free access
138 Growing the productivity of government services
Broadcasting Corporation) success in building up a well- presented store
of information, going beyond news and into information provision more
generally. One super- site (called Directgov) would be citizen- facing, and
the other (called Businesslink) would be company- and employer- facing.
Initially established under the Cabinet Office, the two sites later moved to
be sponsored by main departments, Directgov by DWP and Businesslink
by HMRC (which Chapter 4 shows is extensively business- facing).
Directgov focused on copying and migrating information for citizen
services across from departmental websites in a rewritten and re- presented
form. It achieved early success when the Transport Department began
using it for its forms for motor vehicles and driver licences (see Chapter
6). The rewriting and re- presenting of DWP’s benefits materials proved
much more complex and by 2009 was only partly accomplished, with
partly inconsistent wording still appearing on the DWP’s own websites
and on Directgov (Dunleavy et al., 2009, p. 7). By this time, though, DWP
assumed departmental responsibility for running Directgov, which subse-
quently somewhat speeded up the transfer across of information. With the
migration of service- related information supposed to be completed in 2011,
Directgov’s salience as a source of government information had markedly
increased over previous failed government portals. Experiments by the
Oxford Internet Institute in 2009 found that by using Directgov, nearly
seven in ten respondents (amongst internet users) could find out salient
details about eligibility for different DWP benefits (Dunleavy et al., 2009).
However, the government super- sites plan had expected Directgov to
develop portal- like ‘synergies’, where citizens looking for one piece of
information would also learn about other services and complete online
transactions. In fact, these behaviours did not grow much because the UK
government had completely failed to anticipate the dominant develop-
ment of Google and other search engines – where people go directly to
the relevant information page for their specific needs, rather than navi-
gating through the still crowded and complex Directgov opening screens
(Dunleavy et al., 2007). In 2011 the government announced the superses-
sion of Directgov by a more ambitious Government Digital Service.
Instead of just providing passive information sheets (as all the DWP sites
did), the new site aimed to actually help people to complete transactions
online, something completely infeasible across the board in the DWP
services until late 2009, and only slowly developed since then to the time
of writing (mid- 2012).
The development of DWP’s means of communicating with its custom-
ers up to and beyond the end of our study period are summarized in
Table5.2, developed from material gathered for a 2009 NAO report on the
department (Dunleavy et al., 2009). In the mid- noughties the department’s
M3049 - DUNLEAVY 9780857934987 PRINT.indd 138M3049 - DUNLEAVY 9780857934987 PRINT.indd 138 17/12/2012 09:1017/12/2012 09:10
Patrick Dunleavy and Leandro Carrera - 9780857934994
Downloaded from PubFactory at 07/28/2022 01:29:55PM
via free access
Social security 139
Table 5.2 The evolution of approaches to information exchange in the
Department for Work and Pensions
Key Periods Disseminating
Information about
Benefits to Potential
Applying for
Contacts with
before 1999
Paper leaflets
Face- to- face
explanations in
local offices
Media campaigns
and advertising
Mailed in paper
Paper forms
completed face
to face in local
Paper letters
Face- to- face
made from
2000 to 2008
DWP websites
developed and
online information
greatly increased
Extensive redesign
of leaflets and
improved risk
assurance on their
information being
up to date
Phone- based
strongly for all
Redesign of many
paper forms
Development of
electronic claims
Some claim forms
available online
Phone- based
increased with
all customers,
especially those
over 60
Local office access
for customers
over 60 removed
from 2009
Consolidate and
improve all
online benefits
information on
Directgov (closing
down DWP
citizen- facing sites)
Develop the first
two major
online benefit
applications for
JSA and State
Develop the first
systems for online
with customers
via e- mails, web
accounts etc.
Develop ‘Tell Us
Once’ procedures
long- run
(to 2017)
Further develop
online advertising
Develop Web 2.0
applications and
Grow the
of online
applications (to
40 per cent for
JSA by 2011)
Grow the
of online
Develop ‘self-
service’ and
online accounts
Source: Derived from materials gathered for Dunleavy et al. (2009).
M3049 - DUNLEAVY 9780857934987 PRINT.indd 139M3049 - DUNLEAVY 9780857934987 PRINT.indd 139 17/12/2012 09:1017/12/2012 09:10
Patrick Dunleavy and Leandro Carrera - 9780857934994
Downloaded from PubFactory at 07/28/2022 01:29:55PM
via free access
140 Growing the productivity of government services
expensive attempted transition to phone- based services was threatened
by an avalanche of almost 195 million customer phone calls a year, many
generated by the complexity of benefits rules and the opaque language in
which they were communicated, and others reflecting payment delays and
uncertainties. A quarter of phone calls were judged ‘not value- adding’
and DWP admitted to a Parliamentary select committee that its contact
centres were so overwhelmed that 44 per cent of calls in 2005 were going
unanswered. DWP’s increasingly effective top management team took
drastic action to try to cut the volume of calls – by making more use of
postal forms; redesigning paper forms, contact centre phone scripts and
web pages to try and get communication with customers right first time;
and using intelligent voice recognition (IVR) systems to automate phone
call handling (although IVR is much disliked in the UK). These radical
efforts succeeded in cutting phone calls by 40 per cent by 2008.
But just as this battle seemed to being won, in late 2008 DWP top man-
agers were shocked to discover from new research on benefits claimants
that 51 per cent were already online with broadband internet access. This
rate was substantially less than the UK population as a whole, but also a
world away from the assumptions that had led the department to make its
expensive transition to phone- based services. At the same time, the 2008
recession drastically increased the numbers of unemployed people regis-
tering for JSA, and also brought in a new wave of redundancies in sectors
of business where people were used to doing things online.
Responding to numerous requests, DWP created a simple online form
to pre- populate some parts of the form that call centres normally com-
pleted for applicants in a 40- minute phone call. This online facility was
a cheap ‘quick fix’ that was expected to be of minor significance. Yet by
December 2008 some 50 000 people a month were filling in and returning
the form. The high level of demand led to top managers bringing forward
to 2009 an online form designed to replace in full the initial contact centre
phone call (Dunleavy et al., 2009, p. 5). Online pension registration was
pushed back to later action in the UK – although it was successfully
introduced in the USA and Canada in 2009 and expanded rapidly in both
countries (Dunleavy and Rainford, 2011).
The DWP’s startling conservatism in the face of online services, lagging
more than a decade behind developments towards online transactions
for UK tax paying, reflected a conflict of values within its organizational
culture. The number one value for managers and most staff was maintain-
ing the security, accuracy and integrity of the existing, complex benefits
systems. The senior IT staff who were handling rules changes to DWP
systems were especially conservative, with a change notification process
that never took less than two years to implement any changes, small or
M3049 - DUNLEAVY 9780857934987 PRINT.indd 140M3049 - DUNLEAVY 9780857934987 PRINT.indd 140 17/12/2012 09:1017/12/2012 09:10
Patrick Dunleavy and Leandro Carrera - 9780857934994
Downloaded from PubFactory at 07/28/2022 01:29:55PM
via free access
Social security 141
large. Hundreds of minor system changes were under consultation or pro-
gressing towards implementation at any one time. This primary preoccu-
pation with maintaining complex systems’ basic stability was run close for
top managers by a drive to minimize running costs, creating savings in staff
numbers, improving efficiency and cutting fraud and error (NAO, 2008b
and 2010c). These latter motivations were partially shared and partially
rejected by staff, not anxious to see their job prospects worsened and work-
loads increased. Finally, a relatively strong ‘public service ethos’ was a core
value amongst grassroots staff. At the top management level it focused on
‘delivering the best feasible customer experience’ (Dunleavy et al., 2009,
p. 12). The huge modernization of DWP services, both in the smartened- up
Job Centres and in the building and business processes used for the newer
phone- based benefits like Pension Credit, testified to the strength of this
third impulse. But improving the customer experience was throughout
easily trumped by the drives to maintain benefits system integrity and keep
costs to a minimum, reflecting strongly conservative influences.
The key past constraint on organizational learning about productivity in
government has been the absence of reliable data on output measures. In
the UK this information deficit in social security only began to be offi-
cially addressed in the mid- 1990s. Our main analysis draws on valuable
work by DWP’s own analysts in cost- weighting outputs in sophisticated
ways, and thus covers the period 1998–2008, which provides a useful per-
spective. But since ten years is still a relatively restricted perspective, we
have also undertaken a longer- run 20- year analysis, covering the period
1988 to 2008, seeking to map productivity trends from somewhat cruder
data series, assembled in a consistent way from publicly available data
from the DWP and the ONS.
Table 5.3 shows that our analysis of the DWP and its predecessors’
productivity is based on looking at data for the full range of its services.
Because responsibility for the payment of some smaller ‘social protection’
benefits was transferred from the DSS/DWP to other government depart-
ments during the period under analysis, we excluded these benefits from
our calculations, to keep our output measure fully comparable over time.
Data were assembled on a financial year basis, starting in financial year
1997–98 and running forward for ten years. Again following the recom-
mendations of the Atkinson Review (2005b), our key measures for 14
main areas are the number of applications for that benefit registered in
M3049 - DUNLEAVY 9780857934987 PRINT.indd 141M3049 - DUNLEAVY 9780857934987 PRINT.indd 141 17/12/2012 09:1017/12/2012 09:10
Patrick Dunleavy and Leandro Carrera - 9780857934994
Downloaded from PubFactory at 07/28/2022 01:29:55PM
via free access
142 Growing the productivity of government services
a year (claims), together with the total number of ongoing payments for
that benefit per year (load). The logic here is that in terms of administra-
tive costs, new benefits claims are far more expensive to process than
ongoing load. Many of the transactions costs of administering benefits
lie in judging people’s eligibility, checking databases and documentation,
coping with missing information and assuring against fraud or error.
Especially once the DWP transitioned to using electronic payments to
bank and Post Office accounts, the main costs of ongoing payments arose
from processing changes of addresses and other altered circumstances.
To cost weight output volumes we again followed the methodology sug-
gested by the Atkinson Review. Output volumes were weighted according
to their share of total administrative costs for processing new applications
Table 5.3 Data and adjustments used for the measurement of productivity
in UK social security, 1998 to 2008
Variable Evidence Used, and Adjustments Made
Outputs for
processing of
Number of new claims, and number of ongoing
payments, processed for: Jobseeker’s Allowance and its
predecessor; Incapacity Benefit; Maternity Allowance;
Widow’s Benefit*; Social Fund grants and loans*; State
Pension; SERPS; Attendance Allowance; Disability
Living Allowance; Carers’ Allowance; International
Pension Credit*. Internal data provided by DWP
covering 1997–98 onwards
For our longer time series (1988 to 2008) we use DWP
published data on numbers receiving most of the
same benefits. However, we exclude the smaller
benefits marked * above, but because we use a ‘social
protection’ measure we must perforce include War
Pensions (not administered by the DWP, but by the
Ministry of Defence)
Cost- weighting of
Unit costs for each benefit above, provided by DWP
internal data
Inputs, for total
factor productivity
Deflated total labour and other administration
costs obtained from DWP statistical teams. Capital
consumption was also provided by DWP staff
Inputs for staff
Number of full- time equivalent (FTE) staff in social
security and employment assistance, obtained from
DWP annual reports and those of its predecessors.
Longer- term data also include staff administering War
M3049 - DUNLEAVY 9780857934987 PRINT.indd 142M3049 - DUNLEAVY 9780857934987 PRINT.indd 142 17/12/2012 09:1017/12/2012 09:10
Patrick Dunleavy and Leandro Carrera - 9780857934994
Downloaded from PubFactory at 07/28/2022 01:29:55PM
via free access
Social security 143
for benefits (claims) and for the maintenance of existing benefit caseloads
(loads). In this chapter we used weights calculated by the internal produc-
tivity unit at DWP, which are based on unit costs. The unit costs show
how much was spent to produce each benefit payment as a share of total
administrative expenditure. Again, in normal years the variations in costs
are mostly quite small. However, costs are often higher in the first years of
introducing a new benefit, because both the staff and claimants are unfa-
miliar with how it is supposed to work, and mistakes, operational failures
and other ‘teething problems’ more often occur. As the benefit’s opera-
tion become routinized, operating costs generally fall, unless an IT glitch
develops, or a vulnerability in the benefit processing procedures becomes
apparent (e.g., because of a new type of fraud spreading from one region to
another). For our longer series covering two decades, we cost weight more
crudely using an annual cost per 1000 benefit payments measure derived
from the public annual reports of DWP and its predecessor departments.
Are quality controls needed for social security series, and if so what
should they focus on? The Atkinson Review suggested that the extent of
fraud and error in paying benefits was the most appropriate element to
consider in quality- weighting. The National Audit Office has been quali-
fying the accounts for first the Benefits Agency and later the DWP since
the end of the 1980s, on the grounds that the overall rate of fraud and
error is too great to allow the accounts to be signed off. However, there
has been consistent action by successive top civil servants at the depart-
ment to counteract the problem. Rates of fraud have fallen by half in
absolute number terms in the noughties, and declined from 2.1 per cent of
benefits paid out in 2000 to less than 1 per cent by 2009 (Tinkler, 2010).
Customer error has shown trendless fluctuation in the same period at
between 0.6 and 0.9 per cent, and official error (mistakes made by staff)
has grown slightly from 0.4 to 0.6 per cent (albeit with peaks of 0.8 per
cent in 2004–05 and 2006–07, mainly around Pension Credit). These vari-
ations are so small that in our view they are well within the margin of error
in the underlying statistical systems producing them. We would have to
quality weight fraud and error very heavily for such variations to affect the
over- time productivity trends in any visible way, although levels are much
greater in some areas (such as Housing Benefit, not handled by DWP but
delivered by local authorities). An NAO study (2008b) (and see World
Bank, 2010) compared the fraud and error rates in the UK with those in
other advanced industrial societies, and concluded that they are unexcep-
tional in either direction – neither clearly better, nor worse.
A second dimension where quality- weighting could well be applied con-
cerns the quality of DWP customer services. These certainly were clearly
increasing in general modernity and point- of- service standards throughout
M3049 - DUNLEAVY 9780857934987 PRINT.indd 143M3049 - DUNLEAVY 9780857934987 PRINT.indd 143 17/12/2012 09:1017/12/2012 09:10
Patrick Dunleavy and Leandro Carrera - 9780857934994
Downloaded from PubFactory at 07/28/2022 01:29:55PM
via free access
144 Growing the productivity of government services
the noughties, contrasting strongly with the very static and depressing con-
dition of many DSS offices in the 1990s. Similarly, DWP call centre services
generally improved, with the exception of some problems around the intro-
duction of Pensions Credit (and to a lesser extent around earlier and later
benefits changes). For instance, Table 5.4 shows that in 2006 DWP received
nearly 60 times more complaints about its pensions services (mostly about
Pension Credit) than it did about Jobcentre Plus. However, the overall DWP
rate of complaints even in this peak year was less than two per 1000 customer
transactions, lower than that of its Australian counterpart, Centrelink.
Many other aspects of quality can only be guessed at – for instance, the peak
of missed calls in 2005 did not generate as many complaints.
After reviewing a wide range of evidence we conclude that quality-
weighting could enhance the accuracy of the main series below, but that
the available data on complaints and problem incidence is not good
enough to do this accurately or consistently. Equally, using the arguments
made in Chapter 2, we do not believe that much of what the department’s
top managers regard as quality improvements (such as premises mod-
ernization and better point- of- service standards) can be accepted as such.
DWP has no competitors and has a highly dependent customer base,
making it a monopoly supplier par excellence. Hence, in line with the argu-
ment in Chapter 2, we regard the belated modernization of DWP premises
and business systems in 2001–07 as largely a catch- up operation, bring-
ing point- of- service standards up to touching distance of private sector
service providers, but certainly doing no more than that. Important as
these changes were for managers, staff and the continuing public support
Table 5.4 The ratios of complaints to transactions for DWP agencies in
2006 compared with Centrelink, Australia
Name of Organization Total Number
of Transactions
Total Number
of Complaints
Complaints per
1000 Transactions
Jobcentre Plus 48 202 000 40 000 0.83
Pensions Service 8 240 000 41 000 4.97
Disability and Carers’
na 8900 na
For comparison:
Centrelink, Australia
9 870 000 39 300 3.98
Note: Transaction numbers in the first two columns are rounded to the nearest 1000.
Centrelink transactions are those listed as individual entitlements in the Centrelink Annual
Report 2006–07 (Table 1, p. 11).
Source: LSE Public Policy Group (2008, p. 28).
M3049 - DUNLEAVY 9780857934987 PRINT.indd 144M3049 - DUNLEAVY 9780857934987 PRINT.indd 144 17/12/2012 09:1017/12/2012 09:10
Patrick Dunleavy and Leandro Carrera - 9780857934994
Downloaded from PubFactory at 07/28/2022 01:29:55PM
via free access
Social security 145
for the service, it is hard to regard it as anything more than a late, forced
accommodation to modern service organization standards, already well
established elsewhere in the rest of the economy.
Turning to inputs, for the total factor productivity (TFP) series in the
most recent decade we were able to obtain from DWP statisticians good-
quality measures of ‘total relevant expenditure’, that is, the total annual
costs of staff salaries, intermediate outputs and contracting, and good-
quality numbers on capital stock depreciation. Despite the DSS to DWP
transition, data are available on a consistent basis from financial year
1999–2000 (that is, for nine years). They also include one- off investment
costs. All these costs were deflated according to specific and sophisticated
pay and capital deflators, again provided to us for each year by DWP.
For the labour productivity series, the DSS and DWP staff numbers
count is a well- established and reliable annual statistic in this policy area.
At various times, this one department accounted for nearly a quarter of
Home Civil Service numbers, so that its personnel numbers consequently
attracted significant Treasury and parliamentary scrutiny. Table 5.5
shows that the numbers of FTE staff in DWP peaked in 2002–03, some
15 000 staff higher than the combined opening numbers of staff in the DSS
and Employment Services Agency. They then fell by 25 000 by the end of
our study period.
Turning then to the substantive analyses, we consider first the main
study focus on the last decade, and then discuss the more tentative picture
that we can draw for the longer period for two decades after 1988. Third,
we compare the pictures offered by the two sets of productivity series.
Main Productivity Series for 1999 to 2008
The total factor productivity (TFP) series shown in Figure 5.2 was calcu-
lated by dividing the output index by an index of inputs based upon all
Table 5.5 Staff numbers in the Department for Work and Pensions, and
before 2001 in the Department of Social Security, in thousands
of FTEs (full- time equivalents)
Year 1997–
115.8 118.5 114.6 116.1 124.1 131.4 130.8 126.9 118.3 112.7 105.9
Source: Authors’ calculations assembled from data for DWP, DSS and relevant agencies.
M3049 - DUNLEAVY 9780857934987 PRINT.indd 145M3049 - DUNLEAVY 9780857934987 PRINT.indd 145 17/12/2012 09:1017/12/2012 09:10
Patrick Dunleavy and Leandro Carrera - 9780857934994
Downloaded from PubFactory at 07/28/2022 01:29:55PM
via free access
146 Growing the productivity of government services
staff salaries, other administration costs and capital depreciation costs,
deflated as set out above. The overall outputs curve here declined slightly
from 1999 to 2002, and then fluctuated around the new level for the rest
of the period. The total inputs costs curve increased to 2003–04, especially
sharply in the last two years with the DWP reorganization and launch of
new benefits. There was also significant extra spending on consultancy
and on new IT systems at this time: the combined expenditure for both
headings more than tripled from £94 million in 2001–02 to £306 million
in 2003–04 (see section 5.4 below for a fuller discussion). Input costs then
fell at a fairly steady rate for the rest of the period, largely under the influ-
ence of the departmental merger being consolidated, and DWP offering
up large staff reductions during the cross- government Gershon Review
(see Chapter 4).
Consequently the overall TFP productivity trend for social security
administration shows a decline from 1999 through to 2002, where it bot-
tomed out for two years. There were then steady improvements in the last
years of our period, with rises here responding to the decreases achieved in
input costs. DWP maintained a consistent quality standard in its services,
Volume (99/00 = 100)
97/98 98/99 99/00 00/01 01/02 02/03 03/04 04/05 05/06 06/07 07/08
Financial years
Index of output Total expenditure index Total factor productivity
Figure 5.2 Total factor productivity in UK social security, 1997 to 2008
Source: Authors’ calculations based on data provided by DWP.
M3049 - DUNLEAVY 9780857934987 PRINT.indd 146M3049 - DUNLEAVY 9780857934987 PRINT.indd 146 17/12/2012 09:1017/12/2012 09:10
Patrick Dunleavy and Leandro Carrera - 9780857934994
Downloaded from PubFactory at 07/28/2022 01:29:55PM
via free access
Social security 147
so these gains are solid. Given that outputs remained almost stable during
this later period, the increase in productivity was wholly attributable to the
reduction of overall administrative spending for social security.
Looking at staff productivity, we divided the total cost- weighted
outputs index for social security by an index of the number of FTE
staff employed. Figure 5.3 shows that the key feature here is the large
jump in staff in 2001–03, during and following the DWP merger. Given
the static nature of the overall outputs curve, the result is inevitably a
large apparent slump in staff productivity in the reorganization years,
also evident in the DWP’s own official analyses (DWP, 2008) and in a
separate analysis by the Office for National Statistics (ONS, 2008b).
However, a trend for falling productivity was evident before this acute
downwards blip.
Volume (99/00 = 100)
97/98 98/99 99/00 00/01 01/02 02/03 03/04 04/05 05/06 06/07 07/08
Financial years
Output index Staff productivityIndex of FTE staff
Figure 5.3 Labour productivity in UK social security, 1997 to 2008
Source: Authors’ calculations based on data from DWP.
M3049 - DUNLEAVY 9780857934987 PRINT.indd 147M3049 - DUNLEAVY 9780857934987 PRINT.indd 147 17/12/2012 09:1017/12/2012 09:10
Patrick Dunleavy and Leandro Carrera - 9780857934994
Downloaded from PubFactory at 07/28/2022 01:29:55PM
via free access
148 Growing the productivity of government services
The 2002–03 nadir in productivity coincided with the troubled introduc-
tion of Pension Credit and a range of other new DWP initiatives, which
reflected specific ministerial efforts to achieve greater policy effectiveness.
We would normally expect the conjunction of all these reorganization and
policy changes to have some significant negative impacts on staff produc-
tivity as new systems bedded in; benefits staff gained expertise in how to
operate novel benefit procedures; and customers (plus their families and
charities or care workers advising them) gained more experience of new
application processes. However, a substantial recovery of staff productiv-
ity levels occurred from 2004–05 onwards, with a somewhat steeper curve
than for total factor productivity.
Longer- term Productivity Trends, 1988 to 2008
To get a longer- term picture, we need to shift to data where far more
caveats about quality apply. The essential move here is to compute our
own cost weights for different main benefits outputs, and use them in a
consistent way across the whole two decades period. They are derived
from the data published annually by the Office for National Statistics in
the Abstract of National Statistics and in DSS/DWP departmental Annual
Reports. These estimates are clearly somewhat lower- quality data than the
ones used in the previous calculations, because the cost- weighting of ben-
efits activities is based on cruder annual allocations of costs across benefits
derived from the departmental reports’ costs attributions. In addition, the
‘social security’ outputs measure used here also includes some benefits not
paid by DWP (or DSS before it), most importantly, a diminishing amount
of veterans’ pensions administered by the Ministry of Defence (specifically
the Service Personnel and Veterans Agency within MOD). To maintain
consistency we have included the relevant MOD administrative expendi-
ture also in the inputs series used here.
Figure 5.4 shows an especially rapid (step) increase in output volumes
in the period 1991 to 1992, a period of economic recession, with a gentle
increase in other years before and after this up to 1995. After that date
output volumes fell back gradually. Total input costs on this basis grew
steadily from 1988 to 1995, increasing by more than 40 per cent in this
period. Costs were then pruned sharply for two years, before starting a
gradual rise in the run-up to the departmental reorganization. Trends in
these series for the decade since 1999 are generally very close to the pat-
terns discussed above.
The TFP curve thus shows two long- term and sizeable dips, plus two
recoveries. At the nadir of the first dip, in 1991–92, productivity levels
were down by more than a sixth. However, by the end of the first dip, in
M3049 - DUNLEAVY 9780857934987 PRINT.indd 148M3049 - DUNLEAVY 9780857934987 PRINT.indd 148 17/12/2012 09:1017/12/2012 09:10
Patrick Dunleavy and Leandro Carrera - 9780857934994
Downloaded from PubFactory at 07/28/2022 01:29:55PM
via free access
Social security 149
1996–97, productivity levels were somewhat higher than they had started
from in 1988. During the second dip, productivity levels were down by
three- tenths in the worst year, 2002–03. By the end of our study period, in
2007–08, total factor productivity levels were still somewhat worse than
they had been in 1988. So the net effect of a series of massive organiza-
tional changes across 20 years was to leave social security TFP almost
unchanged, in fact slightly worse than it was at the start of the period. This
is disappointing enough for a period of such high hopes and such energetic
reorganization and reinvestment. However, in addition, the long dura-
tion and large amplitude of the two TFP dips implies a substantial loss of
social welfare caused by failing even to consistently maintain previously
achieved productivity levels.
Comparing Productivity Estimates
To assess the reliability of our 20- year TFP estimate with the ten- year one,
we compare both estimates in Figure 5.5. This also includes two other offi-
cial UK government estimates of social security productivity for the last
decade, both using an Atkinson- derived methodology for cost- weighting
outputs. The first is from the Office for National Statistics (2008a) focused
Volume (1988 = 100)
Financial years
Total factor productivityIndex of outputIndex of input
Figure 5.4 Longer- term estimates of changes in total factor productivity
for UK ‘social protection’ services, from 1987 to 2008
Source: Authors’ calculations based on data from National Statistics and DWP/DSS
departmental reports.
M3049 - DUNLEAVY 9780857934987 PRINT.indd 149M3049 - DUNLEAVY 9780857934987 PRINT.indd 149 17/12/2012 09:1017/12/2012 09:10
Patrick Dunleavy and Leandro Carrera - 9780857934994
Downloaded from PubFactory at 07/28/2022 01:29:55PM
via free access
150 Growing the productivity of government services
on all social security benefits (not only those paid by DWP). The second
comes from DWP’s internal productivity unit (DWP, 2008). To make
comparison more straightforward, we have rebased all four productivity
series on the same base year, which is 1999–2000.
There is a striking level of broad agreement between the four data
series for their common period, 1997–2008. All four estimates show that
productivity levels fell appreciably, bottoming out in 2003–04. Taken
overall the ONS clearly offers the most benign interpretation of the last
ten years’ data, identifying the gentlest decline in productivity to 2003
and the strongest recovery since. The other three projections are broadly
convergent. Our long- term view agrees closely with the DWP series on the
2008 endpoint, while our short- term series is somewhat more pessimistic
on the revival of productivity achieved. Overall, our estimates suggest
that total factor productivity in 2008 was no higher than in 1988. In the
next section we turn to consider the factors that may help account for this
relative stasis.
Volume (2000 = 100)
Author’s 88–08 rebased 99/00
ONS rebased 99/00 DWP Paper rebased 99/00
Author’s 99–08 rebased 99/00
Figure 5.5 Comparing four estimates of total factor productivity in UK
social security, 1988 to 2008
M3049 - DUNLEAVY 9780857934987 PRINT.indd 150M3049 - DUNLEAVY 9780857934987 PRINT.indd 150 17/12/2012 09:1017/12/2012 09:10
Patrick Dunleavy and Leandro Carrera - 9780857934994
Downloaded from PubFactory at 07/28/2022 01:29:55PM
via free access
Social security 151
Three factors might help put the long- run lack of change in DWP’s pro-
ductivity into sharper focus:
1. A high rate of exogenous political, policy and organizational change
can be expected to have adverse implications for organizational pro-
ductivity in government. We have seen that DWP had a great deal of
ministerial alternation in office, including 12 top ministers (Secretaries
of State) in twenty two years, from 1988 to 2010, plus extensive
changes of junior ministers – none of which is helpful for the long-
range growth of productivity. Pursuing greater effectiveness, ministers
extensively remodelled benefits rules and introduced new benefits (like
Pension Credit), as well as requiring the DWP to undertake new tasks
(like the active labour policies demanded from Jobcentre Plus). In
addition, each new top minister and junior ministers tended to make
a series of smaller initiatives to put their mark on services, especially
under New Labour.
Large changes of benefits always create a rise in unexpected imple-
mentation difficulties. In machine bureaucracies backlogs of cases can
quickly build up. Redress cases (complaints, appeals and interactions
related to them) can spiral in a short period. Customer behaviours can
also be adversely affected, with many more people complaining more
readily when service levels pass tolerable limits. These issues then take
time to deal with and to renormalize, especially in an organization
without much ‘fat’ or slack in it. Major policy launches and organi-
zational restructuring tend also to distract senior managers from
improving day- to- day operations, perhaps delaying incremental sys-
temic improvements for several years, impeding training and causing
other adjustments useful for driving up productivity to be put on hold.
All of these factors clearly contributed to slower progress in DWP’s
productivity over time. There was a departmental and government
commitment to ‘benefits simplification’, but little evidence that it was
effective at this time (Work and Pensions Select Committee, 2006b).
More useful was improved DWP communications with its customers
(especially after 2005–06).
2. Government- wide pressure for greater efficiency was clearly important
in the Gershon Review period, from 2005 to 2008, when all the DWP
productivity series reviewed above bounced back (albeit with some
significant differences in timing). The chief impact here was through
staffing numbers being reduced by just under a fifth from their peak
M3049 - DUNLEAVY 9780857934987 PRINT.indd 151M3049 - DUNLEAVY 9780857934987 PRINT.indd 151 17/12/2012 09:1017/12/2012 09:10
Patrick Dunleavy and Leandro Carrera - 9780857934994
Downloaded from PubFactory at 07/28/2022 01:29:55PM
via free access
152 Growing the productivity of government services
levels (Work and Pensions Select Committee, 2006a). (Yet just outside
our study period, in late 2008 the upsurge in unemployment claims
led Jobcentre Plus to re- contact some 6000 staff who had just left the
agency, in the hope of re- employing many of them to cope with the
increase in demand.)
3. Investing in new business processes, increased capital spending and
organizational modernization should also have paid off for DWP in
terms of improving productivity. From 1999 onwards DWP spent a
lot of money on new IT systems, new buildings and capital invest-
ments and new working methods. Each of the many changes made
under this heading was necessarily supported by a strong Whitehall
business case, in which cost reductions played a key part. All the key
modernization changes were advocated partly on the grounds that
they would lead to some demonstrable increases in productivity.
From the literature reviewed in Chapter 1 we might expect that the use
of more and better ICT, improvements in customers contact methods,
the introduction of modern management practices, greater use of
outsourcing and bringing in consultants to speed up reorganizations
would all boost productivity growth. The mix of initiatives in DWP’s
case was in most respects quite comparable to that in other areas of
UK central government. But in fields like tax collection and customs
regulation we have seen that total factor and staff productivity trends
moved upwards much more strongly over these decades. (The same
mix of investments applied in prisons also, again with favourable
results; see Bastow, 2012, Ch. 4, and 2010.)
To explore this last issue in more detail, we have again assembled data
series for the 1998 to 2008 period, covering the same dimensions reviewed
in Chapter 4:
Expenditure on ICT strongly reflects the policy change factors
noted above, none of which can be implemented without altering
the underlying computer recording and payments systems. While
compiling NAO reports in early 1999 and again in late 2008, we
visited large DWP field offices in two northern regions and the
transitions made in the department’s ICT systems across this period
were impressive. In 1999 all back offices operated ‘dumb terminal’
systems, no staff could look at the internet, and the operations
were all conducted on mainframe systems from the Operational
Strategy period that were slow and problematic. A decade later
all staff were connected to high- powered departmental networks
with modern PCs. They could look at the internet, and in the call
M3049 - DUNLEAVY 9780857934987 PRINT.indd 152M3049 - DUNLEAVY 9780857934987 PRINT.indd 152 17/12/2012 09:1017/12/2012 09:10
Patrick Dunleavy and Leandro Carrera - 9780857934994
Downloaded from PubFactory at 07/28/2022 01:29:55PM
via free access
Social security 153
centres they were using web- based programs to run through a call
script using algorithms, and also calling up additional information.
However, we did note that the newer IT systems were time consum-
ing to operate, and that around four in ten of the staff preferred
to enter information directly into the ‘legacy’ IT systems wherever
they could do so. Experienced staff (the minority who were long-
term DWP folk) found it quicker to enter data directly into the
same ‘OpStrat’ computers of a decade earlier than to use the newer
screen- based systems. Managers frowned on this gambit, because
it reduced the possibilities of checking on benefits decisions. But
they were also dependent upon their most competent staff to meet
demanding workload targets, and so could not curtail the practice.
Outsourcing via Private Finance Initiative projects in DWP again
focused exclusively on major construction projects. Our two sets
of visits to large DWP offices showed huge changes in the depart-
ment’s built environment. In 1999 thousands of the department’s
Newcastle staff still worked in rundown offices, many appearing
to be in pre- fab buildings from the early post- war period. A decade
later they had all moved into purpose- built, suburban call centres
located on modern business parks. Some of the vast new buildings
almost resembled ‘Star Wars’ sets in their scale and modernity.
Again these substantial transitions were also strongly associated
with DWP radically redesigning its work teams and business proc-
esses. So we take PFI expenditure as a proxy for the extent of major
managerial change.
The use of consultants was again closely bound up with DWP’s
major reorganizations. Firms and agency staff were brought in to
create extra capacity to do much of the change management work-
load, while most in- house staff kept ongoing operations on track.
Like HMRC, the Department of Work and Pensions also brought
in consultants later in the noughties to help implement ‘lean’
approaches to redesigning service pathways.
Assembling relevant data for these factors was as difficult as for the
taxation departments, described in Chapters 3 and 4, and following
through the merger of DSS and the Employment Services Agency raised
some similar issues. Hence a general caveat still applies to data on these
three factors, that the information involved was assembled from a range of
sources. (And see the Appendix for more details of sources.)
Table 5.6 shows spending on ICT, PFI construction projects and the use
of consultants as shares of the total administrative expenditure for DWP
and its predecessor departments. The total amount of administrative
M3049 - DUNLEAVY 9780857934987 PRINT.indd 153M3049 - DUNLEAVY 9780857934987 PRINT.indd 153 17/12/2012 09:1017/12/2012 09:10
Patrick Dunleavy and Leandro Carrera - 9780857934994
Downloaded from PubFactory at 07/28/2022 01:29:55PM
via free access
154 Growing the productivity of government services
spending absorbed on these elements doubled from a tenth of total admin-
istrative spending in 2001–02 to a fifth in 2006–07, a major increase in
investment and capital intensification. However, in the intervening years
this share was less, and it also fell back again slightly in 2007–08. The
increases were greatest in ICT, but concentrated in the last two years of
our period. PFI spending on building grew less but was more consist-
ently up throughout 2004 to 2008. Consultancy spending showed a large
upwards jag in 2003–04, and was slightly higher also in the two adjoining
We followed the same procedures here as those discussed in section 4.4
in Chapter 4, lagging ICT, consulting and PFI (construction) expenditure
by one year against the cost- weighted output index level and fitting a basic
regression line. Again with so few observations we cannot read much
into the patterns obtained, but the tables obtained can still provide some
insights into how our three factors appear to be associated (or not) with
changing productivity levels.
Two of the charts below, Figures 5.6b and 5.6c, strongly suggest that
there is no connection between either PFI construction spending or con-
sultancy spending and the index of DWP outputs. Figure 5.6a at first sight
seems somewhat different – perhaps an eighth of the variations in outputs
might be attributable to IT spending as a proportion of administrative
costs. However, the regression line here is strongly shaped by data from
the last two years of our study period, when high levels of ICT investment
coincided with rising DWP output levels, which were in fact principally
Table 5.6 Expenditure on information technology, PFI construction
projects and consultancy as percentages of total administration
expenditure in UK social security, 1997 to 2008
Year 1997–
ICT 7.18 6.06 5.47 5.00 4.55 3.91 4.88 6.88 6.44 11.27 11.43
PFI (non-
na 5.11 4.73 4.49 4.50 4.18 4.85 6.64 7.15 6.94 6.88
Consulting na na na na 1.57 2.07 4.04 2.18 1.69 2.12 na
Total for
na 11.17 10.20 9.49 10.62 10.16 13.77 15.7 15.28 20.33 18.31
Note: Before the financial year 2001–02 data for ICT correspond to spending by the
Information Technology Services Agency. From this point onwards ITSA activities were
outsourced to EDS via a PFI contract. ICT values after 2001–02 are annual payments for
all ICT- related PFI contracts; na = data not available.
M3049 - DUNLEAVY 9780857934987 PRINT.indd 154M3049 - DUNLEAVY 9780857934987 PRINT.indd 154 17/12/2012 09:1017/12/2012 09:10
Patrick Dunleavy and Leandro Carrera - 9780857934994
Downloaded from PubFactory at 07/28/2022 01:29:55PM
via free access
Social security 155
due to the onset of economic recession. So overall, what is most interesting
about the DWP explanatory factors is how much weaker they apparently
were than was the case for HMRC and Customs. These factors critical
to the administrative reorganization of social security seem to be hardly
Output (97/98 = 100)
4.5 5.5
ICT expenditure (as percentage total administration costs)
y = 0.2974x + 90.652
R2 = 0.1251
6.5 7.5 8.5 10.59.5 11.5
Figure 5.6a Lagged ICT expenditure plotted against outputs for social
Source: Authors’ elaboration based on data from DWP departmental reports.
Output (97/98 = 100)
2.0 2.5 3.0 3.5 4.0
PFI non-ICT expenditure (as percentage of total administration costs)
Y = –0.177x + 92.577
R2 = 0.0118
Figure 5.6b Lagged expenditure on PFI construction projects plotted
against outputs for social security
Source: Authors’ elaboration based on data from DWP departmental reports.
M3049 - DUNLEAVY 9780857934987 PRINT.indd 155M3049 - DUNLEAVY 9780857934987 PRINT.indd 155 17/12/2012 09:1017/12/2012 09:10
Patrick Dunleavy and Leandro Carrera - 9780857934994
Downloaded from PubFactory at 07/28/2022 01:29:55PM
via free access
156 Growing the productivity of government services
correlated at all with improvements in outputs. Figure 5.6c plots lagged
expenditure on consultancy against outputs for social security.
This result can be best explained in terms of the overall conservatism of
the department’s efforts at restructuring. From the outset, the DSS/DWP
organizational culture was one that clung tightly to its past administra-
tive practices using paper forms and legacy IT processes. This mindset,
was hugely strengthened by the dominance of ‘mainframe guys’ (and they
were all men for many years) in its key ICT positions. It was accentuated
by outsourcing most IT responsibilities to large and conservative systems
integrator firms, especially EDS. Thus many different intellectual and
cultural barriers made senior DWP staff unable to grasp the magnitude
of the opportunities that they could have taken up at the end of the 1990s
to effect more far- ranging changes in how their new department would
go about its core mission, as in their conservatism about using the Web
(Dunleavy et al., 1999, paragraph 2.31). By 2002 the picture had not
shifted much. In particular, at the same time that the almost equally con-
servative Inland Revenue hierarchy were (slowly) getting to grips with
online self- assessment, and that Customs was trying and failing to move
VAT transactions online, first DSS and later DWP top officials rejected
any idea of developing online transactions. Instead, they persuaded min-
Output (97/98 = 100)
4.5 5.0
PFI non-ICT expenditure (as percentage total administration costs)
5.5 6.0 6.5 7.0 7.5
Y = 0.1454x + 91.36
R2 = 0.0316
Figure 5.6c Lagged expenditure on consultancy plotted against outputs
for social security
Source: Authors’ elaboration based on data from DWP departmental reports.
M3049 - DUNLEAVY 9780857934987 PRINT.indd 156M3049 - DUNLEAVY 9780857934987 PRINT.indd 156 17/12/2012 09:1017/12/2012 09:10
Patrick Dunleavy and Leandro Carrera - 9780857934994
Downloaded from PubFactory at 07/28/2022 01:29:55PM
via free access
Social security 157
isters to move the department lock, stock and barrel into a phone- based
model of customer contacts, and they put on one side any developments at
all of online transactions models.
The consequences are rather dramatically captured in Figure 5.7, which
shows the mix of DWP communications with its customers in 2005, at
the height of the mushrooming growth of partly abortive phone calls to
the department unleashed by the initial business process transition. By
2008 the reimposition of tight management of phone calls had succeeded
in reducing their numbers by 40 per cent. And, of course, by this time
face- to- face contact had been further reduced, except for Jobcentre Plus’s
surveillance of jobless working age people. But the net effect was chiefly
to increase once again the importance of paper forms and documentation
sent in by post. At the height of the internet boom, the department’s online
communications accounted for no more than half of 1 per cent of trans-
actions by 2005. By 2008 the online numbers had barely increased, and
before fieldwork began for the critical 2009 NAO report DWP still had no
benefits transactions being transacted in whole or in part online, nor any
publicly announced plans to do so.
The DWP’s first hurried efforts to create some online capability, partly
in response to the critical NAO report on its customer communications
(Dunleavy et al., 2009), also misfired. During 2008 the facility to pre-
register an initial JSA application online was no sooner announced than
Millions of customer contacts
340 000 out of
145 million contacts
Post Face to face Online
Figure 5.7 The changing pattern of the DWP’s customer contacts, 2005
to 2008
Source: Dunleavy et al. (2009, Figure 3, p. 13).
M3049 - DUNLEAVY 9780857934987 PRINT.indd 157M3049 - DUNLEAVY 9780857934987 PRINT.indd 157 17/12/2012 09:1017/12/2012 09:10
Patrick Dunleavy and Leandro Carrera - 9780857934994
Downloaded from PubFactory at 07/28/2022 01:29:55PM
via free access
158 Growing the productivity of government services
the take- up grew rapidly to 50 000 users a month. But all that happened to
apparently e- applications then was that the details given were recorded in
an Excel file, which was downloaded at the start of the next day’s business
and distributed to the 26 DWP call centres, who would then phone back
the would- be claimants involved, but only when they had spare capac-
ity. Because the call centres were often busy, it could be up to three days
before people were called back. Since customers expect speedy responses
to online submissions, by this stage many people had already begun to
worry that their application was lost and so had rung up anyway, thereby
starting duplicate applications running. When a more capable JSA online
process was started in mid- 2009, it had a patchy record. This first online
DWP transaction was expected to be used by 40 per cent of jobless claim-
ants by 2011, but its implementation proved problematic. ‘By April 2011,
20 months after the option became available, only 17% of new claims for
Jobseeker’s Allowance were made online’ (Public Accounts Committee,
2011b, p. 9). In early months many users found the JSA system crashed or
experienced other serious problems, with around a quarter of applications
started in this way failing to make it onto the DWP systems as they should.
Organizational conservatism also remained starkly evident in many
other aspects of DWP operations. The JSA process with its three stages
was estimated by some critics to be costing £450 million extra per year
because claimants had to interact twice with JCP, and then a separate
set of benefits decision- makers went over the paperwork for a third time
(Clory, 2009). The electronic kiosks in Jobcentre Plus shut out job seekers
from learning about electronic job search, instead of facilitating it. And as
the numbers of jobless increased in 2009, Labour ministers still refused to
allow JSA applicants to offer online proofs of looking for work (such as
e- mail job application letters). They insisted on low- risk as well as high-
risk claimants still showing up every two weeks at massively overcrowded
Job Centres, for contact interviews whose length perforce fell from the
previous ten minutes to two to three minutes each, arguably a waste of
time for all concerned.
However, it is important to note in closing that departmental officials at
first strongly contested this narrative and denied that opportunities to save
administrative costs have been foregone. They argued that retaining the
face- to- face interactions of its staff with working age people was essential
for DWP to properly develop ‘active labour market’ policies. In their view
the Jobcentre Plus model contributed to slowing the growth of unemploy-
ment in the 2008–10 recession considerably below previously forecast
levels, and thus it helped to contain the much larger rise in unemployment
payouts that could otherwise have occurred (Sharples, 2010). Moving to
more online transactions, in this view, would have perhaps shaved admin-
M3049 - DUNLEAVY 9780857934987 PRINT.indd 158M3049 - DUNLEAVY 9780857934987 PRINT.indd 158 17/12/2012 09:1017/12/2012 09:10
Patrick Dunleavy and Leandro Carrera - 9780857934994
Downloaded from PubFactory at 07/28/2022 01:29:55PM
via free access
Social security 159
istration costs, but at the risk of stimulating more jobless benefit claims
because of relaxed ‘disciplinary’ effects on those seeking work. The plau-
sibility of this counter- argument is hard to assess, but DWP’s long lags
in moving a whole range of transactions with customers (and not just job
seekers) online are beyond doubt.
In mid 2010 under new Conservative ministers DWP also completely
changed its stance on online applications for benefits, partly responding
to new ‘digital by default’ expectations from the Cabinet Office. As part of
the shift towards a single universal benefit, the department now announced
that it was aiming to move 80 per cent of its JSA customer transactions
online by September 2013. But MPs doubted that this represented realistic
business planning, as opposed to a target driven by austerity pressures. A
leading House of Commons committee noted that:
The Department could not explain the basis of the 80% target at the hearing.
Subsequent written evidence from the department stated that 86% of JSA
customers already use the internet and 67% have access in their homes, while
just over 40% are “ready, willing and able” to use online JSA services. (Public
Accounts Committee, 2011b, p. 9)
Despite extensive efforts at transforming DWP’s business processes, the
rather depressing conclusion we have reached is that its productivity
remained almost unchanged across 20 years. Even allowing for adverse
effects on productivity from frequent policy shifts and organizational
restructuring on DWP’s efficiency, there is a clear performance gap. DWP
was characterized by a conservative organizational culture, especially in
envisioning its major business processes and in all aspects of its IT opera-
tions. As a result, its large- scale organizational changes after 2001 were
exclusively directed towards an already outdated, phone- based adminis-
tration model. The department almost completely neglected to develop
the potential for ‘digital era’ changes to online transactions approaches.
The parallel here seems to be with those private sector industries in
earlier periods where ‘computers are everywhere except in the productivity
numbers’ (Solow, 1987). Private companies in the 1980s and early 1990s
invested millions of dollars in automation and new IT investments that
subsequently could not be traced through into increases in productivity
or corporate profitability. Similarly in our study period DWP seems to
have managed to modernize its business processes at huge expense but
without realizing sufficiently strong benefits to boost its productivity
levels. However, recent trends in productivity have been upwards, and the
external impetus from the Gershon Review cut staff numbers appreciably.
M3049 - DUNLEAVY 9780857934987 PRINT.indd 159M3049 - DUNLEAVY 9780857934987 PRINT.indd 159 17/12/2012 09:1017/12/2012 09:10
Patrick Dunleavy and Leandro Carrera - 9780857934994
Downloaded from PubFactory at 07/28/2022 01:29:55PM
via free access
160 Growing the productivity of government services
After a Conservative and Liberal Democrat government took office in
2010 (outside our main study period) a leading Tory politician (the former
party leader, Ian Duncan Smith) became the DWP Secretary of State. He
had specialized in social security matters for many years while in opposi-
tion, and was clearly strongly committed to reforming welfare systems and
operations. A key Conservative pledge was to introduce a single universal
credit (or benefit for those not working) that is also integrated with the
tax system. Implementing this pledge promises to inaugurate a new era
of radical change in UK benefits administration from 2015 onwards, a
shift in which radically new IT systems and business process capabili-
ties will again be absolutely central. DWP contracts for the IT aspects of
the universal benefit/universal credit transition were let in 2011, at initial
costs of around £1 billion. Meanwhile, DWP has also planned large staff
reductions, and is under intense austerity pressures to cut £2.7 billion
from its running costs within a few years, by early 2015 (Public Accounts
Committee, 2011b). The usual caveats apply about difficulties in shifting
conservative organizational cultures, reorganizations normally depressing
productivity, and the past poor record of major IT projects in UK gov-
ernment. Nonetheless, the conjunction of strong pressures suggests that
some future productivity gains may still be realized in UK social security
M3049 - DUNLEAVY 9780857934987 PRINT.indd 160M3049 - DUNLEAVY 9780857934987 PRINT.indd 160 17/12/2012 09:1017/12/2012 09:10
Patrick Dunleavy and Leandro Carrera - 9780857934994
Downloaded from PubFactory at 07/28/2022 01:29:55PM
via free access
ResearchGate has not been able to resolve any citations for this publication.
ResearchGate has not been able to resolve any references for this publication.