Abstract
This paper considers the evolution of the UK's pension programme in the light of various stated rationales for public intervention. It argues that the publicly-provided (tax-financed) pension programme has gone through four distinct stages since 1946. It examines some of the issues that have arisen in the context of private pension provision in the UK, both in the form of so-called ‘defined benefit’ and ‘defined contribution’ pension plans, as well as individual purchases of annuities.
Background: the semantics of pensions
There is now a vast literature on pensions and pension reform. ‘Pensions’, in United Kingdom (UK) parlance refer to both publicly and privately-provided pensions, and are contrasted with ‘social security’ more generally or ‘social welfare’ more specifically which may involve additional payments such as tax credits and housing support to older people. (This contrasts to US terminology, where ‘pensions’ typically refer to provision of pensions by employers and insurance companies, whereas social insurance programmes which make payments to households over a certain age are generally referred to as ‘social security’ and any additional income and wealth-tested components of old-age income support might be referred to as ‘welfare’).
These distinctions are pertinent because in countries such as the UK with complex pension programmes, most older households receive payments from both public and private sources, and from broadly contributory-based social insurance programmes as well as through add-ons and transfer payments such as free travel, subsidies to fuel bills and tax credits. A ‘multi-pillar’ approach of complex structures and diverse sources of pension income has its attractions insofar as there are different rationales for pension provision, some elements of which might best be handled by one sector or one form of provision and other elements by another. A standard framework for the analysis of pension provision, which goes back to at least Diamond (1977), identifies five rationales for state intervention: market failure, paternalism, redistribution, revenue-raising and administrative efficiency. So, for example, one might argue that poverty alleviation or the provision of an income ‘floor’ should be the responsibility of the state, whereas ‘income replacement in retirement’ might best be served by private markets (directly or indirectly through insurance provision); indeed this specific position was cogently argued by the World Bank (1994).
There is however an obvious risk that arises in pension regimes with complex provisions and multiple targets. By allocating pension instruments to targets, in some sort of analogue of a ‘Tinbergen rule’, by which achieving the desired values of a certain number of targets requires the policy maker to control an equal number of instruments (Tinbergen, 1952), there is the risk that different parts of the overall pension programme interact in such a way as to create distortions and inefficiencies. By analogy, the targets are not independent and hence individual instruments are not fully controlled. One obvious example is the distortion to household decisions concerning saving and labour force participation induced by the co-existence of public and private pension programmes. Another potential distortion arises in capital markets, where the rate of return on pension saving is potentially affected by unfunded public provision of transfer payments and by specific decisions concerning the regulation of the annuity market.
In addition to these possible distortions, it is rarely true, least of all in the context of the UK, that pension policy evolves as a smooth and regularised response to observed deficiencies in existing provision. UK pension policy has evolved in a series of frequent discrete steps that are often reversed by later policy decisions and that are at times based on short-term expediency. In one sense this somewhat myopic approach is disappointing insofar as the evolution of pension policy should be driven by long-run factors such as demographics, notably the increased longevity of the population, and by explicit consideration of intergenerational as well as intragenerational transfers. Yet arguably, UK pension policy has been more successful than many other OECD countries in being able, over the long term, to absorb demographic ageing without sharp rises in pension taxes and cumulative financial deficits.
With these issues in mind, how can we analyse the UK's pension landscape? In this paper I consider two broad issues: first, the evolution of the publicly-provided (tax-financed) pension programme in the UK and what I consider to be the four stages in the post-war evolution of public policy. Second, some of the issues arising in the context of private provision are considered, in the UK taking both the form of so-called ‘defined benefit’ and ‘defined contribution’ pension plans as well as individual purchases of annuities. Many of these topics will be covered in greater detail in this issue of the Review and hence this is a broad overview of the pension context in the UK.
Public pensions in the United Kingdom: the four stages of evolution
As mentioned in the preceding remarks, since the introduction of comprehensive pension provision in 1946, the UK public programme has had many changes, which can broadly be divided into four sequential phases.
Social insurance
Central to Beveridge's (1942) programme to eliminate poverty in old age was a contributory ‘social insurance’ system in which flat contributions paid for flat pension benefits beyond a specific ‘pensionable age’, differing between men and women, until death. This publicly-run insurance system was to be supplemented by an income-tested programme for those with inadequate contribution histories. These programmes were introduced in 1946 and 1948. The emphasis on poverty alleviation rather than earnings replacement in old age meant that the burden of achieving higher pensions than the minimum was to be borne by the private sector, largely through employer-provided pension schemes which partially covered the working population.
The financial viability of the social insurance component of the pension programme was arguably eroded from the very start, because old-age pensions were to be received almost immediately by early cohorts reaching pensionable age long before they had attained sufficiently long contributory histories. These expenditures exhausted the National Insurance Fund which was projected to be insufficient to finance later payments. From the early 1960s, contributions were therefore boosted by introducing an earnings-related element; even so many pensioners continued to depend on a combination of both partial social insurance benefits and means-tested benefits.
Earnings replacement
By the mid-1960s, the gulf in living standards in retirement between those with employer-provided pensions and those without, who often relied on means-tested benefits to supplement their National Insurance pension, provoked a political debate. Left-wing reform proposals wished to eliminate much of private sector provision by offering a generous earnings-related public pension programme to all, paralleling the ‘Bismarckian’ system of public provision that characterised much of Continental Europe. The right-wing of politics preferred a second tier of compulsory private provision (whether through employers or ‘individual accounts’) to supplement existing private provision. The result was a compromise, introduced in 1975: the State Earnings-Related Pension (SERPS).
SERPS was designed not just to provide a programme of public earnings replacement to supplement the flat pension, but also to provide a degree of redistribution to non-working wives and credits for specific periods of non-participation in paid work. It was designed to be both generous and redistributive, especially toward women. Rather than supersede private provision, however, the reform permitted approved defined benefit employer-provided pension plans to ‘opt out’ of the new second-tier SERPS in return for a lower contribution rate. However such plans had to provide benefits that were broadly at least on a par with SERPS provisions.
This consensus around SERPS, which initially seemed likely to become a permanent feature of the pension landscape, was itself rapidly eroded. Hemming and Kay (1981, 1982) pointed to several weaknesses of the new structure; in particular that the forecasts of future contribution rates had underestimated the potential cost of the programme once the ‘baby-boom’ generation began to reach pensionable age. The Conservative administration that had come to power in 1979 therefore set out to redress the balance between public and private pension provision in two ways: firstly by introducing tax incentives to encourage take-up of new forms of private pension (more properly dealt with in the next section) but second, by cutting back the generosity of SERPS. The latter was achieved by a number of changes in accrual structures (for further details see, for example, Disney and Emmerson, 2005 and Crawford, Keynes and Tetlow, 2014), an important consequence of which was that, in the long run, the second tier of public pension provision would essentially become an additional flat-rate supplement to the basic National Insurance pension: the so-called State Second Pension. This feature of reverting to a flat-rate structure would reappear in due course.
Tax credits
The onset of a Labour administration in 1997 saw yet another approach. In part, the new reforms recognised that rising numbers of pensioners would continue to put pressure on the public finances where benefits were universally increased in line with earnings; in part they reflected the then Chancellor of the Exchequer's interest in utilising tax credit programmes as the primary method of redistribution. Income-tested programmes such as a Negative Income Tax have long been held up as an alternative to contingency-related support programmes such as ‘social insurance’, but the 2000s were arguably the decade in which the largest experiment in such programmes took place in the UK, although a ‘Universal Credit’ was not in fact developed until the subsequent Coalition Administration between the Conservatives and the Liberal Democrats.
The 2003 reform saw the basic means-tested benefit (which had gone through various nomenclatures but which had been known as the ‘Minimum Income Guarantee’ or MIG since 1999) replaced by a Pension Credit Guarantee Credit, which was potentially withdrawn at a rate of 100 per cent as outside income increased. But this Credit was supplemented by a Pension Credit Saving Credit which was withdrawn at a rate of 40 per cent as outside income increased. Hence, this two-tier Pension Credit had some of the elements of a non-linear Negative Income Tax. Whilst the reduced taper in the Saving Credit was argued to have a weaker disincentive effect on pension saving than the 100 per cent withdrawal rate that had characterised the means-tested component up to that point, the net effect of this two-tier system would be to draw more people into the means-tested sector with an overall uncertain (but likely adverse) disincentive effect on pension saving.
A second change was to allow the MIG and, initially, the Pension Credit, to be indexed in line with earnings rather than prices, whereas the Basic State Pension – the social insurance pension – and most of the parameters of the second tier, continued to be indexed to prices. Apart from the complexity arising from differential indexation of components of the programme, one medium-term consequence of this policy would have been to increase the importance of the tax credit component at the expense of the social insurance component, since earnings typically grow faster than prices. In the longer run, the relative importance of the social insurance programme versus the tax credit programme would depend on the movement of the different indices relative to the evolution of the two-tier programme of flat benefits through the social insurance component. Further tinkering with the indexation arrangements of the different components of the Pension Credit component (see Bozio, Crawford and Tetlow, 2010) did little to enhance the transparency of the public programme in the late-2000s.
A citizen's pension?
The next round of public pension reform in one sense reflected a reversion to the original Beveridge conception that the primary state pension provision should be a flat contributory benefit, perhaps reflecting the fact that the principal government Minister involved was, like Beveridge, a Liberal (Democrat). The emphasis on tax credits was now to be downplayed, and the brief-lived regime of earnings-related state pensions and opting-out of ‘approved’ private pension plans formally abandoned. Under the Pensions Act 2014, a single-tier pension replaces the basic state pension and the additional second state pension with a flat-rate pension that is set above the basic level of means-tested support for people who have reached state pension age on or after 6 April 2016. Eligibility to the second tier of the pension benefit will now be extended to groups, such as the self-employed, who had previously been eligible only for the first tier i.e. the basic state pension. A so-called ‘triple lock’ guarantees that pensions rose in line with inflation, earnings or 2.5 per cent – whichever was the highest. Although its proponents claim that this reform is, in the short run, ‘cost neutral’, further financial savings will be obtained by raising the state pension age (now identical for men and women) to age 67 between 2026 and 2028.
This combined pension is considerably higher than the original basic state pension and, since rights to this pension can be accrued during specific periods of non-participation in paid work, and rights to the augmented pension had now been extended to groups such as the self-employed and ‘marginal’ workers, any residual link between the value of contributions paid and pension received is largely abandoned under the new public pension programme; hence this differs from the original Beveridge conception of ‘social insurance’. Moreover, with the abandonment of an earnings-related component and the downplaying of the ‘tax credit’ model, the current programme takes on much of the character of New Zealand's ‘Citizen's Pension’ where, broadly speaking, eligibility is subject to a duration of residency requirement rather than a contributory requirement. Hence the wheel of public pensions has not entirely turned full circle, as the post-2014 form of ‘social insurance’ looks very different from that enacted in the National Insurance Act of 1946.
Evaluation
In the period since 1946, the United Kingdom has arguably implemented each of the four main forms of public provision that might be envisaged: a contributory insurance ‘floor’, earnings-replacement, a modified negative income tax, and a universal ‘citizen's pension’. In some time intervals, several of these ‘models’ of pension provision has been operating simultaneously. Each round of reforms has been imbued with a promise of permanence which has proved illusory in practice.
In my introductory discussion, I suggested that pension provision might have several objectives, and that multiple ‘pension instruments’ might be needed to achieve more than one goal, reflecting the diversity of rationales for public intervention that were considered in the subsequent section. This desire to achieve multiple goals has to be weighed against the goal of administrative simplicity, and it is apparent from even this brief discussion that in certain periods, such as the mid-to-late 1980s and for most of the 2000s, complexity triumphed over simple principles. One motive that has clearly underlain all these periods of provision is the belief that a residual ‘contributory principle’ is a way of raising revenue, however divergent the actual accumulation of pension rights is from ‘insurance’ principles. A second, and worthwhile, goal has been to alleviate poverty although there has been much greater focus on intragenerational redistribution (some of it misplaced by failing to understand the implications of complex arrangements such as differential indexation procedures) relative to intergenerational redistribution. Arguably, with average real wages having been constant for a decade, the ‘triple lock’ is now carrying this principle of poverty alleviation among the elderly too far. And finally, the underlying principle that public provision should leave an important role for private provision of retirement saving has differentiated the United Kingdom model from the experience of many Continental European countries. And it is to private provision of pensions that I now turn.
Private pension provision in the United Kingdom
Employer-provided pensions
It is traditional to differentiate employer-provided pensions into two forms. In ‘defined contribution’ (DC) pension plans, the employer supplements individual contributions to a pension fund, and the pension that the individual ultimately receives will depend on years of service, pensionable age and the return on those contributions. The superiority of an employer-provided DC plan over an individually-purchased annuity arises from the employer's notional contribution, any gains from risk-pooling among the members of the plan, and any administrative savings from group purchase. Clearly earlier contributions into the pension fund accumulate a return over a longer time frame; hence a DC plan might be considered ‘front-loaded’. In contrast, in a ‘defined benefit’ (DB) plan, pension benefits are pre-set according to a formula depending on length of service, accrual rate and a measure of salary, often based on final years' of service. These typical provisions in DB plans (notably ‘final salary-based’ calculations of pension rights) bias pension rights in favour of long-serving members at the expense of short-tenured plan members – indeed this aim of incentivising long service may be the very rationale for such plans (Ippolito, 1997). The risk character of these plans also differs: in DC plans, individuals bear all investment risk, whereas tenure risk and earnings volatility are potential hazards to members of DB plans (Bodie, Marcus and Merton, 1988; Disney and Whitehouse, 1996). Most employer-provided pension plans are (notionally) funded in the private sector whereas many public sector pension plans are not.
Figure 1 depicts the evolution of employee participation in employer-provided pension plans in the UK over the period from 1953 to 2013. (These data are derived from surveys of occupational pension plans conducted by the Office of National Statistics and are subject to some definitional changes as particular enterprises were transferred in accounting terms between the public and private sectors). It will be noted that, while the number of participants in public sector plans ebbed and flowed, largely reflecting the changing size of the public sector itself, a large fraction of public sector workers are covered by pension arrangements. Indeed the fraction has tended to increase as the public sector has become more ‘white collar’. In contrast, private sector provision peaked in the late 1960s, and has shrunk steadily since, particularly in the mid-2000s.
Active members of employer-provided pension schemes, UK: 1953–2003 (millions)
There are several potential reasons for this shrinkage in private sector coverage. First, employer-provided pensions attracted various tax reliefs, which became less important as personal tax rates (especially among high earners) tended to shrink after the 1970s. Second, the composition of pension plans before 1975 was highly heterogeneous, with a mixture of DB and DC plans. The introduction of SERPS forced plans that wished to be ‘approved’ for contracting-out (see the previous section) to be of the DB form and of a certain generosity and risk-character; this deterred some employers from providing plans at all. Third, especially after 1997, cutbacks in tax reliefs allowed by pension funds coupled with changed accounting standards have probably discouraged companies from offering pensions. And finally, there may have been a change in the private sector labour market which discouraged employers from wishing to reward long-serving individuals with pension arrangements: the attraction of ‘pay deferral’ as a recruitment and retention strategy has been eroded.
Public sector pensions have proved more robust in terms of coverage. It may be that forms of incentive other than back-loading remuneration (such as basing pay on direct productivity measures) are less feasible in the public sector. Private sector pay tends to peak earlier in the life-cycle than in the public sector, and incremental structures differ across the sectors (and indeed within sectors). Hence, comparisons of ‘total reward’ in the public and private sectors have to take account of differential pay profiles as well as pension arrangements (Danzer and Dolton, 2012). Nevertheless, there have been attempts in recent years to cut back public pensions in the light of both the growth of unfunded public pension liabilities and also the perception that falling private pension coverage is putting the private sector at a disadvantage in terms of relative returns (Cribb and Emmerson, 2014).
Attention now seems to have shifted away from encouraging individual employers to offer pension plans at all. Cutbacks to the annual and lifetime allowance, which respectively limit tax reliefs on how much can be put into a pension fund in any one year and the total value of the pension fund itself, are likely to deter the use of pension contributions and hikes in pension rights as a tool in total remuneration strategies. However, the introduction of ‘Workplace Pensions’ after late 2012, by which all workers not covered by employer-provided pensions and earning more than £10,000 a year are ‘auto-enrolled’ into a generic DC plan with a matching employer's contribution, is intended to increase coverage of employees by some form of pension plan in order to supplement the new flat state pension. The current strategy appears to be to encourage a greater degree of participation in pensions, while providing no additional incentives for more generous employer-provided plans, accompanied by a steady cutback in public sector plan entitlements. Again, New Zealand seems to be the model for current provision, with Workplace Pensions having much the same character as the KiwiSaver plan introduced there in 2007 (Kritzer, 2007).
The annuity market and individually-purchased pensions
The core of the market for annuities arises through group-provided DC pension plans, as described in the previous section. The size of the voluntary annuity market, in which individuals purchase an annuity either through a retirement saving contract with an insurer, or with a cash sum, is harder to gauge in magnitude or operation. Finkelstein and Poterba (2002) cite figures suggesting that annual annuity payments to voluntary annuitants constituted around one sixth in value compared to annual payments to annuitants in the compulsory market and that voluntary annuities accounted for less than 6 per cent of the total paid to purchase new immediate annuities in 1998. This relatively thin voluntary market may result from problems of selection: however estimates of the potential impact of adverse selection on the pricing of annuities in the UK and on welfare differ (see, for example, Cannon and Tonks, 2004; Einav, Finkelstein and Schrimpf, 2010; and Finkelstein and Poterba, 2002, 2004). Moreover, some of these voluntary annuities will have supplemented purchases of annuities via employer-provided pension arrangements. Nevertheless, more recent panel data from samples of older people allow us to get a better understanding of holdings of annuity wealth (Banks, Crawford and Tetlow, 2015).
The major expansion in individually-purchased annuities arose with the development of Personal Pensions, which were introduced in 1988. These replaced so-called Retirement Annuity or s226 plans which were typically used to top-up existing pension arrangements and which had fairly restrictive conditions concerning for example date of annuitisation. In contrast, Personal Pensions had more flexibility in terms of date at which benefits could be first taken (from age 50) and an extremely generous incentive structure. Individuals could choose to ‘opt-out’ of either the second tier of the state pension programme (SERPS) or indeed an employer-provided pension plan, and receive a 5.8 per cent National Insurance ‘rebate’ which would be paid into their personal pension account, plus an extra 2 per cent contribution for a limited time period. With tax relief, this combined contribution was equivalent to a contribution rate of 8.46 per cent, largely financed by the state. Any additional contributions by the individual would receive standard pension tax reliefs with the limits on these reliefs (as a fraction of income) rising progressively with age. Compared to the ‘return’ on remaining in the second tier of the state pension programme, the implicit return on opting out of the public programme into a Personal Pension was extremely attractive, even allowing for falling real returns and transactions costs, so long as individuals continued to contribute to their Personal Pension (Disney and Whitehouse, 1992a; Chung et al, 2008). Large numbers of individuals opted to buy Personal Pensions, although relatively few chose to supplement their transferred rebates with substantial contributions of their own.
The impact of this experiment in individually-purchased annuities has proved controversial, and has been closely watched in several countries, not least the United States, that have flirted with introducing widespread individual retirement accounts as part of an overhaul of their pension programme. The UK government soon decided that the original structure of rebates had been too generous, and cut back on them (Disney and Whitehouse, 1992b). This reduced the return on Personal Pensions and undoubtedly caused a reduction in the number of participants in the programme. Since charges were typically levied upfront, this may have led to some participants not investing sufficient funds to recover their upfront costs. The desire to reduce upfront charges led the subsequent Labour administration to introduce a variant of Personal Pensions known as Stakeholder Pensions, which exhibited capped charges but a lower range of investment options. This stimulated some further take-up of these products, but this take-up seems to have been the result of an associated change in the structure of tax reliefs rather than the new pension instrument itself (Disney, Emmerson and Wakefield, 2010). The net effect on personal saving of these episodes seems to have been fairly small, although they did contribute to a shift away from second-tier public provision and towards individual private arrangements. It seems likely that the new Workplace Pension will prove a competitive substitute for saving through a personal pension, and it remains to be seen how important this expansion of the individual private pension market in the 1980s and 1990s will prove to be, in terms of overall contribution to pensioner incomes, once these cohorts start to retire.
Conclusion
This paper has summarised the various changes to public and private sector pension provision in the United Kingdom over the past fifty years. It has suggested a set of criteria or targets that might underpin an economic evaluation of the reform process. By attempting to achieve multiple objectives, the UK pension programme has sometimes fallen into the trap of excessive complexity but it has achieved the main objective of providing reasonable standards of living for the majority of pensioners in the face of steady demographic ageing of the population, without recourse to the skyrocketing costs that have been apparent in other countries. It has also, unintentionally, provided a test-bed for the economic evaluation of a variety of modes of pension delivery and reform processes. With better data now emerging on life-cycle saving and on the economic status of successive cohorts of older people, there is scope for continued analysis of these issues.