Commentary: pensions and pensions policy.
Broadbent, Simon ; van de Ven, Justin ; Weale, Martin 等
The Pensions Commission is to make its recommendations on pension
policy reform later this year, in the light of comments on the
challenges identified in its first Report. Widespread agreement on the
need for such reform is partly a consequence of problems which have
already surfaced, such as the insolvency of some pension schemes, moves
to curtail benefits and mis-selling scandals. However the main impetus
comes from the fact, clearly demonstrated by the Pensions Commission
(2004), that for pensioners to enjoy the current relative standard of
living in fifty years time will require people either to (i) retire
later (ii) save more; or (iii) pay higher taxes. In the
Commission's view the key issues are to find a consensus on the
appropriate combination of these three, or of the alternative of lower
pensioner incomes, and to find ways of bringing about such an outcome.
We focus on aspects of such a reform and suggest one measure which could
be implemented without further ado.
For many years the British system of a relatively frugal contributory state pension combined with funded occupational and
personal pensions appeared relatively successful. Pensioners'
incomes grew faster than average earnings, although some people did
better than others--anyone relying upon the Basic State Pension (BSP)
saw a steady erosion of their position. People who had changed jobs a
lot, and many women, also did poorly. In retrospect this 'golden
age' seems to have been something of a fool's paradise.
Unusually and unsustainably high investment returns (and hence annuity
rates) allowed pension schemes to meet their obligations while running
up surpluses and even taking contribution holidays. Looking ahead, it
must be assumed that real investment returns will be closer to the
long-term rate of growth of the economy and that life expectancy will
rise and with it the number of pensioners. Any new arrangements will
have to embody some combination of the three changes mentioned above.
Pensions and the Government
It may seem self-evident why the Government should prescribe the
structure of the pension system in great detail. Pensions are, after
all, the largest single item of public expenditure, even before tax
incentives to encourage private provision are taken into account. At the
very least these resources need to be efficiently deployed. But this
requires a clear view of the most basic question: why have a pension
system at all? In today's society, people can expect to spend a
lengthy period of their lives in retirement. The fact that this is
expected, however, means that individuals can presumably plan for it.
One may take the view that the issue of when during their lives an
individual chooses to spend their income is not very different from the
issue of what they choose to spend their money on at any one time. The
state does not interfere to any great extent in the latter decision so
why should it need a policy over the former?
There are three obvious answers to this question. First, it is not
politically feasible for the state to wash its hands of people who reach
retirement with inadequate savings or pensions, however clearly they
have been urged to save for themselves. The conclusions of the
Commission's first report demonstrate the failure of policy that
leaves people to make their own arrangements. For more than twenty
years, people have known that the Basic State Pension (BSP) was linked
to prices rather than wages. The very elderly have seen their pensions
fall relative to wages as a result, but have not been in a position to
do much about it. People who are now reaching retirement age, however,
have had plenty of time to adjust their savings in light of price
indexation, and in many cases have not done so. It is possible that they
believed (correctly) that the pension regime would become more generous
by the time that they reached retirement, as it has done with the
introduction of the Pension Credit (PC).
A variant of this answer is that we need a pension system because
people do not understand or wish to think about the financial
consequences of retirement, and so need state assistance and perhaps a
degree of compulsion to make adequate provision for themselves.
A second need for pensions arises from the fact that some people
have low lifetime incomes and are therefore not in a position to save
for their retirement. It has become increasingly clear that policymakers
need to think in terms of individuals rather than couples in assessing
lifetime incomes, and that therefore many women, because they have
worked only part-time or not at all, fall into this category. Of course,
if people have low lifetime incomes, then it is not necessary to defer
helping them until they have reached retirement age. A possible
justification for delaying support until retirement is that
people's lifetime incomes are known more precisely after retirement
than at the start of their working lives. Although this argument may
seem persuasive, it is complicated by the potential for associated
behavioural responses during the working lifetime (e.g. that people face
reduced incentives to work while of working age if they know that the
Government will support them in retirement).
Thirdly, there is the question of whether any generation should
receive pensions worth more than their contributions (including
appropriate interest) at the expense of their successors. If this is the
case, then the pension scheme should be designed to transfer resources
to the related cohorts from future generations. This distributional
question is closely related to the second point referred to above, and
is consequently associated with many of the same issues. It is also
possible to argue that such a policy is justified if there is an
expectation of rising living standards, as was the case when the current
pension system was introduced. In such circumstances, it may seem
reasonable that children collectively should contribute to supporting
their parents.
In general then, a pension scheme can be designed to fulfil three
basic functions: it can transfer resources from people's working
lives to their retirement, it can transfer resources from people with
high lifetime incomes to people with low lifetime incomes (or vice
versa), and it can transfer resources between generations. Desirable
pension policy reform depends on views regarding these three
redistributive issues, and differences between the various proposals
that have been suggested often reflect different attitudes towards the
three motivated by self-interest.
The pension credit and means testing
Means testing is, in essence, an arrangement whereby a benefit,
such as the pension credit, is targeted at people in particular need,
usually because without the benefit they have low incomes. It is tapered so that it is withdrawn from people as their incomes rise. In essence
the taper is equivalent to a high rate of income taxation on people with
low incomes. If means testing is reduced (see Appendix), by replacing
some or all of the pension credit with a universal benefit, then the
gainers are the people who currently do not receive much of the pension
credit because of the means test. Within a fixed expenditure on pensions
the cost has to be met by reducing the amount of the benefit paid before
the means test, in which case poor people suffer.
Not surprisingly, the need for such a revision tends to be
justified by the beneficiaries not because it makes them better off but
on the grounds that means testing is a disincentive to building up
private wealth. This is not as obviously true as it might seem. The
means test has a substitution effect because it is, in effect, a tax on
the marginal consumption of the retired people who face it. As such it
is indeed a disincentive to save. But it also has an income effect;
people who face the means test receive lower retirement incomes than
they would in its absence. This income effect is an incentive to save.
It is by no means clear which effect will dominate, although work at the
National Institute based on a simulation model (Sefton and van de Ven,
2004) suggests that replacing the pension credit with a universal
benefit of equal cost will actually lead to lower average private wealth
holding in the long run (see chart 1). The effect on the wealth holding
of the top income quintile is particularly pronounced. A reversion to
the Minimum Income Guarantee with its more pronounced means testing has
the effect of raising pension saving relative to the pension credit.
While these distributional arguments suggest that there is a strong
case for maintaining means testing, there is nevertheless a powerful
argument against. This is that means-tested benefits are often not
claimed by the people who are entitled to them and who may be those most
in need of help. We see this as the most powerful argument for finding a
way to avoid means testing even if it results in greater deadweight
costs.
Private pensions
Properties of different pension schemes
In addition to the various changes which may take place to public
sector pensions, the Government cannot avoid the question of private
retirement saving. Many people with private pension schemes have been
disappointed recently. In some cases occupational pensions have failed
to deliver expected benefits, often because employers have become
bankrupt while their pension funds were in deficit. In other cases
people have saved for their own pensions; they have earned lower returns
than they had hoped and also had to face declining annuity rates.
Schemes in which pensions are linked to final salary or sometimes
peak salary are seen by many people as the gold standard of the pension
system. Nevertheless they embed major injustice. Final salary schemes
pay high pensions to people retiring on high salaries with long
contribution records. They thus give poor deals to people who change
jobs and often also to people who stay with one employer but whose
careers do not prosper, If one regards pensions as being earned by a
combination of employers' and employees' contributions, the
return to someone retiring on a high salary must be high compared with
the average return to the fund. Since this has to be paid for somehow it
follows that the return to someone retiring on a low salary is poor
compared with the return to the fund. Salaries are skewed, with a few
high earnings and many low earners and the dispersion increases as
people age. Thus the median earner receives a return below the market
return on his/her pension contributions; with the same contributions a
pension with a closer link to contributions might offer a better deal.
It should be noted that this particular abuse is much reduced if only
high-paid employees (e.g. executive directors) have access to the
defined-benefit scheme. Nevertheless, final salary schemes mean that
individuals' pension prospects are affected by career risk.
Of course occupational pension schemes may offer defined benefits
even if those benefits are linked to contributions rather than to final
salary, and in that sense there is no clear distinction between a
defined benefit and a defined contribution scheme. The latter tend to
reduce the career risk faced by scheme members. Reforms to the Civil
Service pension scheme proposed in March 2005, moving from a final
salary pension to a career average pension were, given unchanged
contributions, plainly fairer than the present system to the average
employee. They were opposed by the trade unions representing them. A
part of the support for final salary schemes may be simply that people
assume that if their employers want to change conditions of employment it must be a bad idea.
Defined benefit schemes, whether final salary or linked to average
salary, reduce the market risk that employees face and raise that faced
by employers when compared to defined contribution schemes. The latter
are subject to uncertainty about both market rates of return and, when
employees come to retire, annuity rates. In essence, with defined
benefit schemes employees receive not only their pension contributions
but also a form of insurance. This is not shown in their salaries but
is, on the face of it, a real cost to employers (1) and an additional
factor behind the recent closures of final salary schemes. A failure of
actuaries to price this insurance was one of the factors contributing to
the pensions fools' paradise from which the country is only now
emerging. Employers were not paying for the benefits employees believed
they would receive. That employers who close final salary schemes not
only aim to shed the burden of this unpriced risk but also want to
reduce the contributions that they actually make is a ready explanation
of why, in most cases, trade unions are rational in opposing such
changes. But, as we have noted with the Civil Service, trade unions can
also oppose changes when they are in the interests of the majority of
their members.
Defined contribution schemes exist largely because of the tax
reliefs offered to people who save in them. This has a number of
consequences. One has been that the charges levied by pension fund
managers, and particularly those at the retail level, have been high
compared with those imposed on other forms of investment such as unit
trusts and investment trusts. The industry justifies this with the
observation that the costs of running retail accounts are high.
Certainly the development of low-cost pension savings arrangements (such
as internet-based self-invested pension policies) has not put rapid
downward pressure on costs in other parts of the market. Salop and
Stiglitz (1977) offer some interesting insights into why normal
competition might not be felt in the financial services industry; the
continuing existence of suppliers with high fees and high charges may be
explained by the fact that many people have difficulty in understanding
the nature of the products that they are buying. It is not surprising
that pension funds argue for policy changes which will increase the
attractions of investment in their products (e.g. by increased tax
reliefs). Equally an important aspect of government policy should be to
address these market imperfections so as to reduce these deadweight
transactions costs.
Compulsion
An important question arises whether contributions to private
sector pensions should be compulsory. This is followed by a more sterile
debate about whether both employers and employees should be compelled to
contribute. The second debate is sterile because it is very difficult to
imagine that compulsory contributions are paid out of anything except
wages whether they are collected from employers or employees. To the
extent that this is also true with our voluntary system, employees have
less to lose from changes in employer contribution rates than they
perhaps believe. It is, moreover, erroneous to believe that compulsion
'would not work'. The fact is that part of the pensions system
is already compulsory. People who go out to work are obliged either to
contribute to the second state pension or to contract out by joining a
private scheme.
A major benefit of compulsion as compared to the present voluntary
arrangements is that the costs are much lower, at least for lower
earners. Stakeholder pensions have charges of 1.5 per cent per annum of
the value of the fund for the first ten years of its life, falling to 1
per cent per annum after that. The cost of running a compulsory fund
would be likely to be around 0.3 per cent per annum Since the charges
have to be paid out of a gross real return of perhaps 5 per cent per
annum this difference is very important. Whether investment is
compulsory or voluntary, people should have some choice about the way in
which their contributions are invested. Equity funds are likely to give
poor returns in some periods; the difficulties associated with this
would be avoided if investors also had the choice of a fund investing in
index-linked stock. There is a separate question about the propriety of
obliging people to contribute to a pension scheme when the consequence
of this is that they lose means-tested benefits. But actually such a
concern is simply the more general issue whether it is appropriate to
have high effective marginal tax rates on relatively poor people and the
issue is best addressed in the context of an overall review of the tax
and benefit system to establish overall principles about how to
structure it so as to balance the various goals of efficiency,
redistribution and financial soundness. One does not hear people arguing
that they should not contribute to the basic state pension because they
have some other resources and the basic pension is therefore, in effect,
means tested.
United States proposals and transactions costs
At this point is its worth noting the reforms proposed to the
United States Old Age Pension arrangements proposed by Mr Bush (Bush,
2005). He is attracted by the high average returns earned on stock
market investments and the belief that funding old age would be cheaper
if social security (i.e. national insurance) contributions were invested
in the stock market rather than supporting the current pay-as-you-go
scheme. Bush has rightly noted that the Government can operate a very
efficient collection system and does indeed collect taxes very cheaply.
Under his scheme people will be able to contribute up to 4
percentage points of their payroll taxes (with an upper limit fixed
initially at US$1000 per annum but rising steadily over time) into
individual accounts which can be invested in five low-cost broadly-based
funds. At the age of forty-seven, the default option is to move the fund
gradually from investments seen as high risk/high return to those seen
as low risk/low return; individuals and their spouses can, however,
jointly waive this. Participation is voluntary in that people can choose
to stay with the existing arrangements but it is plainly hoped that the
higher returns on offer will encourage them to contract out of the
existing scheme and into this new funded scheme.
The document is vague on detail but the presumption is that the
fund built up is intended to replace existing Social Security (as Old
Age Pensions are known in the United States). The risk that funds will
deliver lower benefits than existing Social Security is to be mitigated
by the portfolio restructuring which takes place on reaching the age of
forty-seven. It is nevertheless not clear from the document what happens
if the fund ends up unable to deliver benefits as large as those under
the existing Social Security scheme. In any case, it must be
questionable whether a contribution rate of only 4 percentage points of
payroll taxes will lead to a satisfactory level of retirement income,
let alone, as the document suggests, create something beyond this.
Nevertheless, there is a very positive aspect to the scheme. The
costs of running the scheme are estimated to be only 0.3 per cent of
fund value, (2) as compared to the 1.5 per cent with UK stakeholder
schemes. This low cost is made possible through the use of the tax
system to collect contributions and is based on a saving scheme already
available to US Government employees.
If public resources are to be deployed in encouraging saving, it is
almost certainly much better for them to be deployed in making available
the tax system as a means of facilitating contributions to private
schemes than in extending tax benefits available to private pensions,
say by reintroducing dividend tax credits. The first reduces social
transactions costs by making a cheap and efficient technology more
widely available. The second would raise expenditure on deadweight
transactions costs.
Rising and uncertain life expectancy
It is often argued that increasing life expectancy makes pensions
expensive and that uncertainty about future life expectancy compounds
the problem. These observations are both true but, without the benefit
of numbers which underpin them, it is impossible to form a view about
how important they are. A focus on life expectancy overstates the impact
of living longer on pension costs because the extra payments which fall
due arise quite a long way in the future and are discounted.
Assuming an index-linked rate of return of 1.5 per cent per annum,
a fair 100 [pounds sterling] annuity for a man aged sixty-five pays an
income of 7.43 [pounds sterling] per annum based on the death rates
shown in the Government Actuary's current life table. This reflects
current death rates rather than projections of the death rates of
current sixty-five year olds as they age. It suggests a life expectancy
for a man now aged sixty-five of 16.6 years, If all death rates up to
the age of 100 fall to half current levels, then the life expectancy
rises to 22.2 years and the fair annuity rate falls to falls to 5.69
[pounds sterling] per annum. These figures compare with quoted rates
(Financial Times, 2 July) ranging from 4.74 [pounds sterling] to 5.18
[pounds sterling]. Thus current annuity rates may already take account
of very substantial reductions in death rates, and/or build considerable
allowance for uncertainty into their pricing. However, they may also
reflect the fact that people who have saved for pensions tend to live
longer than average.
While private pensions face the question of life expectancy risk,
pay-as-you-go schemes additionally face the risks associated with
uncertain fertility rates. In some countries these are now as low as 1.2
children per woman. Such rates, if maintained, have a much greater
impact than does increasing life expectancy on the number of retired
people to be supported by people of working age and thus on the burden
imposed by a pay-as-you-go pension scheme. Sweden has addressed this
problem by linking the values of pay-as-you-go pensions to GDP at
retirement. A small population of working age implies a low GDP. Thus
the demographic risk is borne by the retiring cohort rather than those
currently of working age. Such an arrangement benefits those born in the
distant future relative to those already alive, when compared to
pensions linked to wages. But it does not impose much risk on people
within say twenty years of retirement because the size of the cohort of
working age up to twenty years ahead is known with reasonable accuracy.
The relevant people have already been born. It remains to be seen
whether such arrangements can actually be sustained in the more distant
future if Sweden experiences very low fertility for any length of time.
If one could be confident that low fertility rates were temporary,
then instead of leaving the implications of this to be carried by
pension recipients, as in the Swedish scheme, it would be more
satisfactory to spread the burden by means of public borrowing, If the
gap between receipts and payments arising from a small cohort of working
age is financed by borrowing, then the burden is in effect imposed on
the whole of the future instead of on just one cohort. But such a policy
helps current generations at the expense of future generations if low
fertility rates in fact turn out to be permanent. Thus debt finance
imposes extra risk on people who have no choice about whether they bear
it or not.
A role for National Savings
Returning to the question of whether pension arrangements are
designed mainly to ensure everyone has at least a basic standard of
living in retirement, or to reallocate incomes within people's
lifetimes, our own view is that the first issue should be the primary
purpose of pension arrangements. This means that tax relief or its
equivalent should be directed towards the first tranche of retirement
saving whether or not that is compulsory; at the same time the
Government should facilitate saving in safe media.
One means of delivering this would be through a National Savings
product which would combine a safe index-linked rate of return, similar
to that on index-linked debt, with an additional contribution from the
Government for people who invest in this medium. The additional
contribution would probably be fixed at a rate between those generated
by relief at the standard and higher rates of income tax. This, combined
with an upper limit to contributions, would ensure that the scheme was
directed at its primary purpose--provision of a basic standard of living
in retirement rather than, as with the current arrangements, tax reliefs
which are worth more to high tax payers than to low tax payers.
Contributions could be collected through the tax system in order to keep
costs low. The scheme would be complemented with the introduction of
Government annuities, again sold through National Savings.
The proposal has a number of attractions. It provides a simple
guaranteed product. Transactions and running costs will be considerably
lower than with many private sector schemes. And it allows tax relief to
be focused. From a macroeconomic perspective, this is just another form
of government borrowing and one tailored to meet the needs of savers. To
the extent that people save for their retirement in a National Savings
pensions scheme other forms of government borrowing will be reduced.
Conclusions
Ministers have talked of the need for a consensus on pensions. It
is not clear that they are in any hurry to act on the issue. For some
changes no doubt a consensus would be needed. Since many changes have
both winners and losers, establishing such a consensus is likely to be
difficult. But those changes which reduce transactions costs, by making
the technology of the PAYE tax system available to savers or which
increase choice, through introducing index-linked products sold by
National Savings, do not obviously require a consensus. Slow progress
with wholesale reform on issues such as means testing or changes to the
structure of tax relief need not be an obstacle to the introduction of
changes which reduce costs and increase choice.
Appendix: winners and losers from means testing
The way in which means testing generates both winners and losers
can be seen from the diagram below. On the horizontal axis we plot
income before taxes and benefits and on the vertical axis income after
taxes and benefits. The effects of two tax/benefit regimes are shown.
The line CC' shows the effects of a regime with a universal
benefit. The position of point C shows the value of the benefit. The
line CC' has a slope of less than 45 degrees because each household
keeps less than 1 [pounds sterling] of every [pounds sterling] that they
earn, since taxes are needed to pay for both the benefit and for other
government spending.
With means testing the line CC' is replaced by the kinked line
DD'. The value of the benefit is D, but the first segment of the
line is shallow, reflecting the fact that households do not keep much of
the first tranche of their incomes as a result of means testing. Once
their pre-tax/ benefit income rises to B the means tested component of
the benefit has been completely withdrawn. Beyond this point they keep a
higher proportion of their income than with the universal benefit
because the scheme is cheaper to operate and thus requires lower taxes.
Thus the second segment of the line is steeper than with a universal
benefit scheme.
It is plain from the positions of the two lines that people whose
pre-tax/benefit income is less than A are better off with means testing
than with universal benefit. So too are those with incomes higher than
B*. They keep more of their income because the means testing regime is
cheap.
Those with incomes in the range AB* are worse off with means
testing than without it. However only those with incomes less than B
face a disincentive to save as a result of means testing. People with
incomes higher than B face a stronger disincentive to save in the
absence of means testing because the tax rate is higher. Income effects
mean that people with incomes below A or higher than B* have less
incentive to save when means-tested retirement benefits are available
than when they are not, while the reverse is true for people in the
range AB*.
It might be thought that the effects of the higher tax needed to
pay universal benefits can be avoided if the retiring age is raised,
effectively leaving everyone better off with the universal benefit
scheme. Poor people will undoubtedly have to work longer with such an
arrangement. But if there is a universal benefit the total retirement
income of rich people will be higher than in the presence of means
testing; they receive the universal benefit while they did not receive
the means-tested benefit. Thus, if anything, they will be able to retire
earlier than in the presence of means testing.
NOTES
(1) Occupational pension schemes providing defined benefits have
historically been invested largely in equities because these give higher
returns. But, if the market is efficient, the risk adjusted return is
the same as that on indexed government debt (Cantor and Sefton, 2003).
Since the employer is bearing the risk, the difference between the two
rates of return can be seen as an additional cost borne by the employer.
(2) Although Diamond (1998) may imply a slightly higher cost.
REFERENCES
Bush, G. (2005), Strengthening Social Security for the 21st
Century, http://www.whitehouse.gov/infocus/social-security/200501/
socialsecurity.pdf
Cantor, A.C.M. and Sefton, J.A. (2002), 'Economic applications
to actuarial work: personal pensions and future rates of return',
British Actuarial Journal, 8, pp. 91-131.
Diamond, P. (1999), 'Administrative costs and equilibrium
charges with individual accounts', NBER Working Paper No. 7050.
Pensions Commission (2004), Pensions Challenges and Choices, http:/
/www.pensionscommission.org.uk/
Salop, S. and Stiglitz, J. (1977), 'Bargains and ripoffs. A
model of monopolistically competitive price dispersion', Review of
Economic Studies, 44, pp. 493-510.
Sefton, J.A. and van de Ven, J. (2004), 'Does means testing
exacerbate early retirement?', National Institute Discussion Paper
No. 244.
Wolf, M. (2005), 'A shameful pensions confidence trick',
Financial Times, 1 July.