FAMILY FORTUNES

A GUIDE TO SAVING FOR RETIREMENT

A Discussion Meeting
[Held by the Institute of Actuaries, 22 April 2002]

Discussion Topics
1 Financial Institutions are increasingly faced with the problems of
persuading individuals to save saving for retirement is paramount (though
repayment of a mortgage is a high priority for some people). Government,
too, is concerned with liabilities on the Exchequer as the population ages.
Forecasts of a growing elderly population are top of the list on which the
Treasury seeks economic solutions.
2 For both the public and private sectors, the paper `Family Fortunes, a
Guide to Saving for Retirement', by D. R. Cooper, which is printed directly
after this list of discussion topics, is timely, in that it quantifies the amounts
which breadwinners must save in order to meet standard of living
expectations during their post-employment days. The model developed is a
yardstick for individual investment and for the levels of national savings
which the Government must encourage in order to maintain prosperity as the
population ages.
3 The model can be run on a wide variety of assumptions, and for
different sectors of the population. The conclusions from some scenarios
appear to contradict some long-held views on pensions and pension
provision. A number of interesting discussion points arise:
(1) Savings over the whole of a person's working lifetime are generally
considered prudent and desirable. However, for some households,
attempting to save while the wage earners are young will cause
unnecessary hardship and short-term debt. In these cases, savings for
pensions should start when the children have left home and the first
mortgage has been paid. Should occupational pensions schemes and
state schemes be adjusted to accommodate `cash rich' and `cash poor'
periods?
(2) Low wage earners have no financial incentives to save, because their
pension entitlements will not exceed state means tested benefits. A
rational decision for some people is not to save for retirement under the
present system. State encouragement to save, such as tax relief on
pension contributions and gross build up on investment earnings, are
not sufficient or appropriate for this group. Is some alternative form of
state encouragement to save required?
# Institute of Actuaries and Faculty of Actuaries

2
(3)

(4)

(5)

(6)

(7)

Family Fortunes A Guide to Saving for Retirement

Neither the state second pension nor stakeholder pensions at the lowest
level are sufficient to get many people above the minimum income
guarantee. Should the Government rethink it's pension policy, and
require greater contributions from individuals and employers in return
for higher benefits? Are greater and compulsory contributions required
from employers for stakeholder pensions?
Since the main period for saving for retirement is when people are
middle aged up to retirement, the financial services industry should sell
more protection insurance to young households to protect them from
health breakdowns before or during the critical savings period. The sale
of a pension policy to young people may not be `best advice'.
There has been a trend for males to retire early, when they are in their
middle fifties or early sixties, despite the fact that the population is
healthier and will live longer. Since there is less time to accumulate
savings and to build up pensions, how can people be encouraged to
work, earn and save longer? Will raising the state retirement age solve
the problem if the majority of people continue to retire well below the
age of 65 years?
The income of people in retirement could be enhanced for those who
own their own houses if they could borrow against the value of their
property. Unfortunately, there are few effective equity release schemes
in the market. Should financial providers be encouraged by the
Government to promote these schemes, and should the Government
give financial incentives for equity release?
Apart from defined benefit schemes, the majority of pension provision
is made through schemes that accumulate a lump sum, which is turned
into an annuity at the retirement date (or, in some cases, at a later date).
Should financial providers encourage the use of deferred annuity
products with a limited cash option? The deferred annuity approach
gives more certainty on the pension level. The uncertainty is limited to
the amount of the cash option, and not the pension level as at present.
Have providers lost sight of the fact that the priority for pension
schemes is the provision of pensions, and not lump sums which may/
may not provide an adequate pension?

Family Fortunes A Guide to Saving for Retirement

FAMILY FORTUNES
A GUIDE TO SAVING FOR RETIREMENT
By D. R. Cooper

abstract
The paper investigates the level and incidence of saving required in order to maintain the
standard of living that a household experiences whilst of working age into retirement. In order to
do this, a model has been constructed that follows the income and expenditure of a household,
allowing for tax and social security, as well as changing family circumstances. The model can be
used to explain how a household should save in order to achieve a given standard of living in
retirement.
The author concludes that the usual message, to save a fixed proportion of income
throughout a working lifetime, is at best not helpful and at worst could lead to a lower standard
of living over the household's lifetime. People can and should manage the timing of their saving
and borrowing in order to achieve optimum incomes.

keywords
Saving; Pensions; Financial Planning; Means Test

contact address
D. R. Cooper, William M. Mercer Ltd., 38-40 Trinity Square, London EC3N 4DJ, U.K. Tel:
+44(0)20-7977-8684; Fax: +44(0)20-7977-8892; E-mail: [email protected]

".

Introduction

1.1 The purpose of this paper is to investigate the level and incidence of
saving required to maintain the standard of living that a household
experiences whilst of working age into retirement. In order to do this, we
have constructed a model that follows the income and expenditure of a
household, allowing for tax and social security, as well as changes in family
circumstances. The model is based on one devised by Hamilton, that
considers the position for households in Canada (Hamilton, 2000). The
model can be used to explain how a household should save to achieve a given
standard of living in retirement.
1.2 The obvious way of saving for retirement is through a pension
scheme, which can provide benefits on either a defined benefit or a money
purchase basis (sometimes both). All defined benefit schemes and some
money purchase schemes are provided by employers, and so members benefit

Family Fortunes A Guide to Saving for Retirement

from a form of involuntary (and indirect) saving on their behalf. A

significant difference between the two types is that the former automatically
targets a retirement benefit related to pay, whereas the benefit provided by
the latter depends on the amount of contribution paid, as well as on
investment return and annuity rates. Thus, the former bears a direct
relation to the standard of living whilst in work, whereas the latter need
not. Regardless of its desirability, there appears to be a trend away from
defined benefit pension provision towards money purchase provision.
Considering the average contributions paid into money purchase schemes,
those currently relying on these schemes to provide retirement income
should expect a significant drop in their standard of living once they reach
retirement (for example, NAPF, 2000).
1.3 The model can incorporate both types of pension scheme, as well as
other, more widely available or more flexible, savings vehicles. The model
encompasses investment in housing, personal pension schemes, Individual
Savings Accounts (ISAs) and other stock market funds.
1.4 A household's financial circumstances and its ability to save vary
according to the number of earners and dependants. Saving will be easier in
some periods than in others. Whilst our model demonstrates that households
can maintain their standard of living in retirement whilst restricting saving
to the `cash rich' periods, this strategy involves certain risks, not least the
expectation that one can always enter the labour market. Indirect saving, for
example contributions by employers to occupational pension schemes or
governments to social security, which continues when personal saving is
difficult, reduces this risk. Households with access to these forms of financial
provision might be able to save less than others, and so experience a higher
standard of living.
1.5 Dependants affect, not only one's propensity to save, but also the
need. If a family with children saved the same amount as a family without
children, it might be able to afford a higher standard of living in retirement
than the one which it had whilst in work. It does not seem sensible for the
members of a family to experience hardship throughout their working
lifetime just so that they can be relatively affluent in retirement.
1.6 Financial advisors in the United Kingdom are governed by the
Financial Services Authority, which produces guidance as to procedures that
must be followed when talking to individuals about their saving needs. This
includes taking into account an individual's financial circumstances and his
or her attitude to risk. Whilst information on a prospective client's current
spending needs and lifestyle is gathered, once a savings package has been
`sold', there is no obligation for the advisor to revisit clients to determine
whether it still meets their needs. The model demonstrates that it is unlikely
that an optimal savings strategy could be expressed as a fixed percentage of
earnings throughout life, although this is a frequently used measure for
advice (for example, FSA, 2000). We will also see how sensitive a household's

Family Fortunes A Guide to Saving for Retirement

standard of living is to changes in family or work circumstances, as well as

in changing economic circumstances.
1.7 We begin, in the next section, considering patterns of expenditure by
age and income. The following section describes the model used. The
remaining sections investigate the results that the model produces.
.

Family Spending

2.1 Household expenditure varies by age, by income, and by the number

of people in the household. As we would expect, expenditure increases with
income, both in absolute terms and per capita. In addition, the average
number of people in a household tends to increase with income (or vice
versa), although there is an exception from the second to third income
quintiles, where the head of household was aged less than 30.
2.2 The following observations about spending stand out from the
tables in Family Spending (ONS, 1999).
2.2.1 At younger ages (less than 50), the largest difference in spending
between the lowest and highest quintile is in housing, where the highest
quintile spends about five times the amount of the lowest.
2.2.2 At all ages, the largest proportionate difference in spending
between income quintiles is on motoring and leisure services. The differences
for motoring are most marked for those aged under 30, where the highest
income quintile spends 12 times more than the lowest quintile, and, for the 75
and over age group, where the highest quintile spends 23 times as much as
the lowest. However, the average spend amongst this latter group is still only
half that of the age group 65 to 74. This provides strong evidence that,
whilst spending falls as people age, this is due to lower incomes rather than
due to less need or desire to spend.
2.2.3 Overall, household expenditure seems to increase about four fold
from the lowest to the highest income quintile, whereas median income
increases much faster. The increasing gap partly demonstrates the increasing
ability of the better off to save, but is largely explained by their greater tax
burden.
2.2.4 The average expenditure of those families in the bottom two
quintiles of the income distribution, where the head of household is aged less
than 65, is greater than the median income. At ages less than 30, the
median income in the third quintile is about 13,000 p.a., and it is about 20%
higher at the older ages. Thus, expecting groups earning less than this to
rely on personal, voluntary, saving for retirement is quite optimistic.
2.3 The immediate indication from Family Spending (Table 1.3, ONS,
1999) is that, on average, households are spending no more than 5% of their
income on life assurance and pensions, and this proportion does not vary
much across the earnings distribution. However, whilst this indicates a low

Family Fortunes A Guide to Saving for Retirement

level of average saving, and thus average income in retirement, it hides

more information than it gives. For example, the fact that the amount does
not vary by income distribution suggests that some of this `saving' is through
employer-sponsored schemes, to which there must be an employer
contribution.
2.4 It should be more useful to look at the proportion of income spent
on other saving and investment. This indicates that, excluding home
ownership, the top decile of the income distribution saves about 3% of
income, whilst all other groups save less than 1%. The greatest saving, as a
proportion of income, is on a home. This is about 4% of income, and does
not vary much across income groups. Taking into account all spending on
housing (including, for example, rent, mortgage interest, rates and repairs),
the pattern is very different. Those in the lowest income quintile spend over
40% of their income on housing (about 25%, net of rent rebates and other
allowances), whereas those in the highest quintile spend only 15%.
2.5 In summary, it seems that the top three deciles save about 11% of
their gross income, the middle four save about 8% and the lowest three save
about 4%. Whilst there is some rationale for lower income groups saving a
lower proportion of their income (Sinclair, 1998) apart from a reduced
ability to save the results from the model suggest that 4% is too low. The
results presented in Section 5 suggest that even the top deciles are likely to
experience a fall in their standard of living in retirement.
2.6 Average net household income varies by a factor of nine between
the lowest and highest income quintiles. However, average household
spending, excluding savings and income tax, which is illustrated in Figure 2.1,
varies by a factor of less than five (or four if we normalise for numbers in
each household). This spending includes, for example, food, fuel, clothing

Figure 2.1.

Household spending (ONS, 1999)

Family Fortunes A Guide to Saving for Retirement

Table 2.1.

Net weekly income of currently retired people who find life

financially easy or difficult

All retired
Men
Women
Socio-economic group
A/B
C1
C2
D
Source: NatWest, 1999

Easy (p.w.)

Difficult (p.w.)

172
211
137

115
128
102

247
189
145
114

155
127
115
106

and entertainment, and thus can be used to determine a `standard of living'

for each individual. The model, which is described in the next section, is
predicated on the assumption that people want to maintain a proportion of
the standard of living that they experienced in work throughout their
retirement. The proportion is selected by the model's user.
2.7 In 1999 a survey announced that the gross income required for a
`financially contented retirement' was 193 a week (NatWest, 1999), with
retirement being `financially difficult' on an income of 119 per week. The
NatWest survey estimated that 25% of working people were saving
enough to be comfortable in retirement, but 49% could expect to find
themselves with incomes of less than 119 per week. Drawing on Figure
2.1, we would expect these figures to vary across the income scale, as we
see in Table 2.1.
2.8 It is not possible to make a direct comparison between the socioeconomic groups used to produce Table 2.1 and the income quintiles used for
Figure 2.1. However, it does appear that people in retirement are able to
spend, on average, less than the amount that they believe would make
themselves `content', suggesting that their incomes are less than they would
wish.
2.9 This brief description of people's spending and saving habits will be
compared with the results that the model produces in Section 4.
.

The Model

3.1 A deterministic model has been used to project forward the income,
expenditure and wealth distributions of a household from a start age to an
`end age', which is the assumed age at death. The output, which is discussed
in the following sections, is determined by controlling the household's
standard of living, or consumption, throughout its lifetime.

Family Fortunes A Guide to Saving for Retirement

3.2 The household can range from one adult to a couple with up to four
children. The household's income can vary, as can the ways in which adults
contribute income to the household. The model permits one member of the
household to take a break from work, and could be extended to incorporate
periods in and out of work, or a changing share of income between adults.
3.3 The household starts with a basic salary and level of debt, controlled
by the user. Income will largely come from earnings, but households can also
draw down on savings and receive means tested benefits. The model permits
`general' spending (everything except house purchase and other saving), and
then allocates everything left over to paying off the initial debt and to the
various savings vehicles. The savings considered are home ownership,
pensions (including defined benefit and defined contribution occupational
schemes and personal pension schemes), ISAs and `unsheltered' vehicles. After
tax, National Insurance, general household spending and contributions to
occupational pension schemes (where appropriate), the priority order is:
(1) mortgage payments;
(2) payment of the initial debt;
(3) personal pension savings;
(4) ISA saving; and
(5) unsheltered saving.
It is assumed that the maximum amount possible will be put into the
relatively tax efficient savings' vehicles. Only any excess cash is invested in
`unsheltered' savings.
3.4 General spending can increase in line with earnings or inflation, if
preferred. Spending also increases by the number of people in the household.
Thus, each additional adult or child increases expenditure by 70% or 50%
respectively. These figures are suggested by the OECD (Hauser, 1997). After
retirement, it is possible to permit the level of spending to change relative to
the spending immediately prior to retirement.
3.5 The tax calculation takes into account most allowances, and
includes adjustments for MIRAS and capital gains, where appropriate.
Similarly, the income calculation takes into account child benefit.
3.6 The household's finances are projected forward, past retirement age,
to their assumed age at death. Any unsheltered savings are used to offset the
mortgage required when a home is purchased, and, in those years when the
net income after tax and general spending is less than required to pay the
mortgage, the household `draws down' on its ISAs. The household can also
borrow against the value of its house, or, if it is sufficiently wealthy, against
its unsheltered savings (this is in lieu of drawing down on those savings).
3.7 Finally, the adults can join either defined benefit or defined
contribution occupational pension schemes. The users can fix the age of
joining, but, thereafter, they are assumed to remain in the scheme for the rest
of their working lives.

Family Fortunes A Guide to Saving for Retirement

3.8 Clearly, such a model relies heavily on a broad range of assumptions,

and these are set out in detail in Appendix 1. Two important assumptions are
that the tax and pension regimes will remain as at present; that is that
allowances and other thresholds will continue to increase in line with price
inflation, as they have done for the past 20 years. In due course, even people
on less than average earnings will find themselves above the marginal tax
threshold, so this practice is probably unsustainable in the long term. It is
possible to alter the assumption to permit real growth in tax thresholds.
3.9 Thus, the main source of income before retirement age is salary, but
households might also receive some investment return, child benefit, means
tested social security and ISA drawdown. After retirement, income sources
could include a personal pension, Basic State Pension (BSP), State Second
Pension (S2P), protected rights pension (the user has the option to contract
the household in or out of the S2P), means tested benefits, ISAs, an
occupational pension and investment income. It is assumed that the personal
pension scheme and ISA funds are drawn down until an age fixed by the
user, at which stage the remaining capital is used to purchase an annuity.
3.10 The key to the model is a single figure, the multiple of a minimum
spend that can be maintained throughout life. The minimum spend has been
chosen as the level of spending recorded for the lowest income quintile in the
ONS survey Family Spending (ONS, 1999). The multiple is set assuming
that, at death, the household's net assets are equal to a multiple (which could
be zero) of the assumed market value of its home. Based on this level of
spend, the model produces information showing how the household's sources
and uses of income vary with age. It also shows how the level of spending
required to maintain the household's standard of living changes over time.
3.11 As an indication of the level of spending that those currently
retired expect in order to be `contented', we can translate Table 2.1 into
multiples of the minimum spend. This has been done assuming that all the
individuals' net incomes are available for general spending. As we shall see,
this might not be the case. The results are shown in Table 3.1.
Table 3.1.

Multiple of minimum spend for currently retired people who

find life financially easy or difficult

All retired
Men
Women
Socio-economic group
A/B
C1
C2
D

Easy

Difficult

2.26
2.78
1.80

1.51
1.69
1.34

3.26
2.49
1.91
1.50

2.05
1.67
1.51
1.40

10

Family Fortunes A Guide to Saving for Retirement

3.12
According to the NatWest (1999) survey, the `average' retired
person needs to achieve a multiple of 2.26 of the minimum spend to feel
financially secure in retirement. From the discussion in Section 2, we know
that there are significant numbers of the working population that cannot
maintain this level of spending, but these results give a broad indication of
what working people might aim towards.
3.13 Another interesting conclusion that we can draw from the
numbers in Table 3.1 is that those in lower socio-economic groups, who will
largely tend to be the less well off, have a very narrow margin between
`financially difficulty' and `financial contentment'. Those in higher economic
groups have a far higher margin, and so have a larger number of
outcomes where they can `make do'. This presumably provides greater
flexibility in saving.

`Standard of Living'

4.1 A common target benefit for occupational pension schemes is 2/3rds

of final salary. Most defined benefit pension scheme members could expect to
accrue this level of benefit over 40 years. Those with shorter service would
receive proportionately less benefit. The implication is that people require
less income in retirement than they do whilst in work. It is easy to see how
this might arise for those who own their own homes, since mortgage
payments would normally be completed before retirement. Similarly, one can
assume that most saving for retirement will be completed before retirement
occurs. However, other items of expenditure are less flexible. Whilst Family
Spending (ONS, 1999) does show that general household spending is less for
those in retirement than for those of working age, it is clear that this is not
always out of choice, but rather through necessity.
4.2 The results presented in this section, therefore, generally assume that
spending (per head and net of saving) will continue in retirement at the same
level as in work. (See the Appendix for a brief summary of different savings
models.) Results presented in Table 4.5.3 show how spending whilst in work
could increase if we allow spending in retirement to fall.
4.3 The financial assumptions underlying the results presented in
Section 4 are set out in the Appendix. Section 8 investigates how sensitive the
results are to some of the assumptions.
4.4 With so many parameters, it is not possible (and would not be
helpful) to present all the possible outcomes. Instead, a number of outcomes
are presented, so that the effect of the changes in assumptions can be
investigated. In all cases, adults enter the household aged 25 and households
are assumed to start out with an outstanding debt of 10,000. The
consequence of means tested benefits is discussed separately in Section 7.

Family Fortunes A Guide to Saving for Retirement

11

Table 4.5.1.

Couple with two children; 50/50 income split; draw down

stops at age 69; no career breaks

Joint income

Retirement age

30,000

65

At 30 only

60

At 30 and 40
At 30 only

65

At 30 only

60

At 30 and 40
At 30 only

50,000

House purchase

Age at death

Multiple

89
95
89
89
95

1.96
1.94
1.60
1.78
1.77

89
95
89
89
95

3.02
3.00
2.42
2.74
2.72

4.5 Couple with Two Children

4.5.1 Table 4.5.1 presents the multiple of the minimum spend that a
household consisting of a couple with two children would be able to
maintain, assuming that its aim is to have the same standard of living
throughout life. We have assumed that both adults receive the same income,
that they have no breaks from paid employment until they retire, and that
their first child is born when the couple reach age 30. Neither adult joins an
occupational pension scheme. At retirement they draw down from their
savings until age 69, when they purchase an annuity. At death their net assets
equal the value of their house.
4.5.2 From these results we can draw some broad conclusions:
The households are not exposed to much mortality risk, since annuity
risk is passed to an insurer at age 69.
Retiring at age 60 reduces the household's standard of living by about
10%.
Purchasing a new home at age 40 reduces the household's standard of
living by about 25%, but creates a larger inheritance.
The average gross household income of those in the top income quintile
is 55,000 (ONS, 1999). These households typically spend over four times
the `minimum spend' per individual member. The results of our
calculations suggest that families in this household and income category
are saving less than they need to maintain their standard of living in
retirement.
Comparing these results against the proposed levels of spending from
Table 3.1, the household with a joint income of 30,000 cannot maintain
the level of spending required for contentment, measured amongst all
socio-economic groups. However, it falls comfortably into the level for
the C1 group.

12

Family Fortunes A Guide to Saving for Retirement

The household with a joint income of 50,000 can meet the overall
target, but undershoots the target for socio-economic group A/B.
However, it does not undershoot by so much that it will find life
financially difficult.
4.5.3 We can investigate how sensitive the results are to some of the
household's general characteristics. The results for households with only one
member of the couple in paid employment, for households where spending
falls in retirement, and for households where one member of the couple takes
a career break are given in Tables 4.5.2 to 4.5.4. In each case the household
purchases one house at age 30, and reaches age 89.
4.5.4 Comparing the results in Tables 4.5.1 and 4.5.2, we can see, for
example, that the household with two incomes of 25,000 and a retirement
age of 65 has a standard of living 14% higher than the household where the
joint income of 50,000 is received by only one adult. This arises because the
former household gets the benefit of two single persons' tax allowances and
has a smaller proportion of income taxed at 40% (although a higher
proportion will incur National Insurance contributions). However, the
former household would presumably incur additional childcare costs relative
to the latter household, and these have not been built into the model.
4.5.5 The data presented in Family Spending suggest that a household
with an income of around 50,000 would be spending about four times
Table 4.5.2. Couple with two children; results where income is paid to one
member of couple; assuming one house purchase at age 30
Joint income

Split

Retirement age

Multiple

15,000

100%/0%

65
60

0.97
0.88

30,000

100%/0%

65
60

1.75
1.58

50,000

100%/0%

65
60

2.62
2.38

Table 4.5.3. Couple with two children; results where expenditure falls at
retirement; assuming 50/50 income split and one house purchase at age 30
Joint income

Retirement age

Fall in expenditure

Multiple

50,000

65

0%
10%
25%
50%

3.02
3.10
3.23
3.48

Family Fortunes A Guide to Saving for Retirement

13

Table 4.5.4. Couple with two children; results where one parent takes a
ten-year break from paid employment; assuming 50/50 income split, one
house purchase at age 30 and retirement at age 65
Joint income

Length of break

Multiple

30,000

10 years

1.62

50,000

10 years

2.51

the minimum spend, but also saving too little to maintain that lifestyle
in retirement. This appears to be borne out by the results of our
calculations.
4.5.6 Although it is becoming increasingly common for both partners to
work, even with young children, the calculations have been repeated to
consider the consequences of a working parent taking a break from
employment after the birth of the first child.
4.5.7 Comparing these figures with those in Table 4.5.1, we see that the
effect of a ten-year break is to reduce lifetime spending by about 15%.
Alternatively, considering the results in Table 4.5.2, we see that returning to
work can increase living standards considerably.
4.6 Occupational Pension Scheme
4.6.1 The majority of employees have access to an occupational pension
scheme at some point in their careers. This is a source of additional saving,
since employers have to contribute to occupational schemes, effectively by
deferring the employee's pay. The results in Table 4.5.1 can be reproduced
assuming membership of an occupational scheme, and some of these are
presented in Table 4.6.1.
Table 4.6.1. Couple with two children; results where one spouse joins an
occupational pension scheme; assuming 50/50 income split and one house
purchase at age 30; both spouses still contribute to personal pension
Joint income

Scheme

Retirement age

Multiple

30,000

DB

65
60
65
60

2.09
1.88
1.99
1.82

65
60
65
60

3.24
2.92
3.10
2.82

DC
50,000

DB
DC

14

Family Fortunes A Guide to Saving for Retirement

4.6.2 The calculations used to produce this table assume that the
individual joins at age 25 and remains a member until retirement age. This is
not entirely realistic, particularly for the defined benefit scheme, since it
ignores the effect of job changes.
4.6.3 Both the defined benefit (DB) and defined contribution (DC)
schemes are assumed to have a 5% member contribution. The defined benefit
scheme has 1/60ths accrual and the defined contribution scheme has a 5%
employer contribution. Neither scheme is assumed to contract out.
Comparing the results with those in Table 4.5.1, we see that the effect of the
defined benefit scheme is to increase general spending by between 6% and
7%. The defined contribution scheme does not increase the spending level of
the household quite so much (approximately 2%), but this will vary
according to the underlying assumptions. These answers seem reasonable,
since the pension schemes provide between 30% and 50% of income at
retirement.
4.6.4 Although the schemes do not appear to have much affect on
consumption, by this measure, they do have a significant affect on the
pattern of saving and replacement rates for the household. This is shown in
Section 6.
4.7 Household with No Children
4.7.1 Results similar to those in Section 4.5 can be presented for
households with no children.
Table 4.7.1.

Household with no children; assuming 50/50 income split and

one house purchase at age 30

Joint income

Household status

Retirement age

Multiple

30,000

Couple

65
60
65
60

2.29
2.09
3.57
3.24

65
60
65
60

3.59
3.26
5.35
4.84

65
60

1.91
1.74

Single
50,000

Couple
Single

15,000

Single

4.7.2 Comparing these results with those in the previous sections, we see
that, as most parents realise, having children reduces one's standard of
living.

Family Fortunes A Guide to Saving for Retirement

15

4.7.3 It is also interesting to compare the outcome for the couple with
an income of 30,000 with the single household with an income of 15,000.
The former household benefits from greater tax allowances whilst in work,
which enables a higher standard of living.
4.8 Different Order of Investments
4.8.1 Each household's standard of living will vary according to how
well its savings perform. Section 8 considers the results if investment returns
are different from those expected for this section. However, the return on
savings also depends on the investment media chosen. In this section we
consider the effect of different priority orders on savings on a household's
standard of living.
4.8.2 The results in Section 5 assume that personal pension scheme saving
takes precedence over ISA saving, and that no attempt is made to pay off the
mortgage early. We will consider what might happen when paying off the capital
outstanding under the mortgage is given priority over investing in an ISA.
4.8.3 Pay Off Mortgage
4.8.3.1 Savings for house purchase and in an ISA have some similar
characteristics: they are both made from net income; and they can both be
realised without incurring capital gains tax (assuming that the house is the
main home). The main differences are:
Flexibility. ISAs are liquid, whereas property is illiquid. However, in
this model we have assumed that the household can borrow against its
equity in the house. This provides liquidity, although, because of the cost
of borrowing, it is less cost efficient than drawing down on an ISA.
Return. ISAs are invested in mixed bond and equity portfolios that are
subject to an expense deduction. House purchase, therefore, gives a
higher rate of return, particularly since it reduces the cost of the
outstanding mortgage.
4.8.3.2 For most of the households that we are considering, altering the
investment order has very little effect on the level of consumption. Any
benefit depends, firstly, on the extent to which the household has free cash to
put towards the mortgage, and, secondly, on the cost of borrowing relative
to the return from investment. Most of the households have very little spare
cash, and rely on borrowing in retirement. Although, in isolation, paying off
the mortgage early seems optimal, as a consequence the households make
less investment. As a result, they need to borrow more in retirement, so the
cost of the mortgage is just deferred.
4.8.4 Give Priority to ISAs over Personal Pension Schemes
4.8.4.1 Personal pension schemes and ISA savings have quite different
characteristics (Cook & Johnson, 2000):

16

Family Fortunes A Guide to Saving for Retirement

Personal pension scheme saving is made out of gross pay, whereas ISA
saving is made from net pay.
Disinvestment from an ISA is flexible, and does not incur tax. On the
other hand, personal pension scheme investments only become available
when the policyholder retires. At retirement, 25% of the fund may be
taken as a tax free lump sum, but the remainder must be taken as income
that is taxed at the policyholder's marginal rate. So, for example,
households that pay tax at the higher rate whilst in work, but at the lower
rate whilst in retirement, are likely to get a better return from personal
pension scheme saving.
There are maximum contributions that can be made to both forms of
saving. For an ISA the amount is fixed (currently 7,000 p.a.), whereas, for a
personal pension scheme, the amount is a percentage of pay (up to a cap).
It is not possible to borrow against personal pension scheme savings.
4.8.4.2 The return available from each type of investment will depend
on the underlying assets. The model assumes that they are similarly invested,
so that the only difference in return arises because dividends received by an
ISA have a tax credit of 10% until 2004, so should provide a marginally
higher yield.
4.8.4.3 Because ISA and personal pension scheme savings are treated so
similarly by the model, altering the order does not have much effect on a
household's standard of living. The crucial difference appears to be the
flexibility afforded by ISA savings, which can be drawn down when income is
low relative to expenditure. If most of a household's savings have been
directed toward a personal pension scheme, then the household would have
to borrow instead, and this is relatively more expensive.
.

Distribution of Income and Expenditure

5.1 It is interesting to consider the way in which expenditure changes

with age. The model produces output that shows how the distribution of
expenditure between consumption and saving varies with household age and
family status. For example, a single person has relatively stable expenditure
over time, although it is slightly larger whilst in work, because of the
mortgage (we have assumed a 25-year term, capital repayment mortgage).
However, families with children have cash rich and cash poor periods, and so
need to target when they save more carefully. This is illustrated in Figures
5.1 and 5.2.
5.2 The drop in general spending at age 30 for the single person is due to
the assumed move from rented to owned accommodation. For the purpose of
this paper we have assumed that the rent is fixed, whereas the value of the
house purchase, and therefore the cost of the mortgage, depends on salary.

Family Fortunes A Guide to Saving for Retirement

Figure 5.1.

17

Household expenditure for a single person with a starting salary

of 15,000

Figure 5.2.

Household expenditure for a couple with two children and

starting income of 30,000

18

Family Fortunes A Guide to Saving for Retirement

5.3 Comparing the two figures, we see that planning for retirement is
more straightforward if your family status does not change. Figure 5.1 shows
how single people are able to save a stable proportion of their income
throughout their working lives. The savings vehicles shown are largely
housing and a personal pension scheme, because the individual has relatively
low earnings. Someone with higher earnings would have greater opportunity
to save, and so could take advantage of other types of saving.
5.4 Figure 5.2 illustrates how households with changing family
circumstances can have long periods when saving is not possible. This is a
deliberate consequence of the way in which the model is set up, since it
assumes that the household wants to maintain its level of consumption.
However, it illustrates that savings need not be constant throughout ones
working life in order to maintain ones standard of living in retirement.
5.5 We can also see that both of these households manage to continue
saving during the first part of their retirement. The single household first
pays off the debt that helped finance consumption whilst of working age, and
then saves to an ISA. As they get older, both households need to borrow
against the value of their homes and their ISA savings to maintain their
standards of living in real (with respect to earnings) terms.
5.6 The model also produces output that shows how sources of income
change on reaching retirement. Figures 5.3 and 5.4 show the income
distribution for couples with a starting household income of 50,000, shared
equally between the two partners. Between ages 50 and 65 household incomes
are not expected to grow faster than the rate of inflation. This assumption
can be varied, but the rate of increase in earnings does tend to decline after
about age 50. The income received after age 69 is also level, since at that age

Figure 5.3.

Household income distribution for a family with two children

and starting income of 30,000

Family Fortunes A Guide to Saving for Retirement

Figure 5.4.

19

Household income for a family with two children where one

parent takes a break for 10 years

the household stops drawing down on its ISA and personal pension savings,
and purchases annuities linked to the RPI. It is partly to balance RPI linked
income and wage linked expenditure that the households used for Charts 5.1
and 5.2 have to borrow in retirement.
5.7 Figure 5.3 would only change slightly if the household had no
children. There would be no child benefit and no need to draw down on the
ISA before retirement, and the level of income in retirement would be
greater. Figure 5.4 shows how the income distribution of the household used
for Figure 5.3 would change if one member took a 10-year break.
5.8 In both cases there is a single house purchase at age 30, and neither
member of the couple joins an occupational pension scheme:
Note how the ISA is used by the second couple to cover expenditure
whilst one partner is taking a break from employment. When the ISA is
exhausted, the model assumes that they are able to defer mortgage
payments until their level of income recovers.
Note also that, whilst the sources of income in retirement are similar in
both cases, the level of income is, of course, lower for the household
where one member has a career break. Because ISA savings were needed
during the career break, they are not available at retirement. This is a
useful illustration of the benefit of saving in ISAs rather than in personal
pension schemes, since the latter do not afford this flexibility.
5.9 To sustain its level of consumption, as well as relying on the ISA
during the career break, this household has to borrow and defer paying off
the mortgage. It only pays off its mortgage by age 75, and still has a small
amount of debt at age 89, although it can meet the debt out of its
`inheritance'. Figure 5.5 shows its expenditure distribution.

20

Family Fortunes A Guide to Saving for Retirement

Figure 5.5.

Household expenditure for a family with two children where

one parent takes a break for 10 years
.

Saving and Replacement Rates

6.1 Traditionally, when planning for retirement, people are advised to

maintain a certain level of saving throughout their working lifetime. The
income at retirement that the saving produces is often illustrated as a
`replacement ratio', or percentage of salary immediately prior to retirement.
Standard advice has been for individuals to aim for a `replacement ratio' of
2/3rds at retirement.
6.2 Figures 5.1 and 5.2 show the various forms of saving included in the
model, with other items of expenditure. However, it is useful to investigate
saving in isolation. We will look at the level and timing of the savings
necessary to maintain the model's imposed standard of living throughout
retirement, and the replacement ratio that this level of saving produces. We
should be able to see whether the standard rules of thumb are based on
reality.
6.3 Figures 6.1 to 6.4 show the savings' rates and replacement ratios
at each age for households with a starting income of 15,000p.a. and
50,000p.a., under various possible scenarios. In each case the household
comprises a couple with two children. The higher lines denote the
replacement ratio and the lower lines denote the level of saving.
6.4 The `replacement ratio' prior to retirement is the total income in the
year expressed as a percentage of salary. It demonstrates the extent to which
the household has had to draw down on its savings during its working

Family Fortunes A Guide to Saving for Retirement

Figure 6.1.

Savings and replacement ratios for retirement age 65

Figure 6.2.

Savings and replacement rates for retirement aged 60

21

22

Family Fortunes A Guide to Saving for Retirement

Figure 6.3.

Figure 6.4.

Savings and replacement rates for retirement aged 65 and

membership of a DB scheme

Savings and replacement rates for retirement aged 65 where a

career break is taken

Family Fortunes A Guide to Saving for Retirement

23

lifetime in order to maintain its standard of living. The replacement ratio

post retirement is calculated as the income in the year expressed as a
percentage of salary in the year immediately prior to retirement.
6.5 One, perhaps surprising, point is that household income appears to
have remarkably little influence on the pattern or level (as a percentage) of
saving. We will investigate this further in the next section, which includes
means tested benefits in the analysis. There is a material gap in the first five
years, which arises because each household is assumed to rent until they
purchase a house aged 30. The model fixes the amount of rent, regardless of
income, whereas the cost of the mortgage is proportionate to income.
6.6 As might have been expected, saving is most volatile in the
household where one person takes a career break (Figure 6.4). Apart from
this case, when sometimes no saving is possible, saving varies between about
15% and 35% of salary, depending on household status at the time.
6.7 Saving is least volatile when individuals have access to an
occupational pension scheme (Figure 6.3). This is partly because the scheme
imposes a minimum level of saving, but also because the household benefits
from the employer's contributions, and so does not have to compensate for
those years when savings has been low by increasing savings at other times to
the same extent.
6.8 All the figures, but particularly Figure 6.4, demonstrate that the
advice, `save 15% of earnings throughout your working lifetime', might not
be most useful. Clearly, guidance is needed as to the effects of reducing,
stopping or delaying saving.
6.9 Replacement ratios vary by level of household income; lower income
households have higher replacement ratios in all cases. This happens for two
reasons:
The BSP and S2P provide a greater proportion of post-retirement
income for those on low incomes (about 23% for household's with
combined income of 50,000 and 44% for household's with combined
income of 30,000).
Since the tax system is progressive, those on lower incomes pay a lower
percentage of their income in tax, and so, according to the model, have
proportionately more left for general spending and saving. It would
probably be more sensible if, once spending fell to a certain level, all
saving ceased. This is investigated in Section 8.
6.10 The replacement ratios at retirement provided by private saving
and state pensions are given in Table 6.1. This table also shows the average
savings rate up to retirement age, including the cost of housing.
6.11 The results in Table 6.1 do not take into account the particular
circumstances of each individual or the arrangements of the household. In
particular, the household where one member takes a career break has a
higher replacement ratio than the household that remains in work and

24

Family Fortunes A Guide to Saving for Retirement

Table 6.1.

Replacement ratios and savings rates

Household status

Joint income

Replacement ratio

Average saving

Retire aged 65

15,000
50,000
15,000
50,000
15,000
50,000
15,000
50,000

74.3%
66.3%
66.3% (49.8% at 60)
59.3% (48.1% at 60)
88.1%
83.5%
93.6%
75.8%

15.9%
19.2%
19.2%
23.6%
13.2%
16.8%
20.3%
24.0%

Retire aged 60
Career break
DB scheme

Table 6.2.

Replacement ratios partner 1 in DB scheme

Joint income

Member of household

Replacement ratio

50,000

Partner 1
Partner 2

111%
56%

appears to have a lower level of saving. This is because the household has
to continue saving in retirement to pay off the debt incurred during the break
from work, so that it does not result in a higher standard of living.
6.12 Similarly, it might be necessary for one member of the household
to achieve a far higher replacement rate than the other, if the household is to
achieve its desired replacement rate. The results presented in Table 6.2
illustrate this point.
6.13 Ignoring the contribution to the occupational pension scheme, both
partners have similar savings ratios. (Partner 2 saves marginally less, since it
is assumed that all elective tax allowances, apart from the single person's
allowance, go to partner 1). However, partner 1's employer's occupational
scheme imposes a certain level of forced saving, as well as providing an
additional source of saving. Effectively, the employer's saving replaces part
of partner 2's saving.
6.14 The model does not allow for family breakdown. If the trends
observed in the 1990s continue, it is likely that there will be an increased
number of single households, many arising from divorce (Murphy & Wang,
1999). In view of Pension Sharing and other divorce legislation, it might be
reasonable to assume that partner 2 has not adopted too much risk by
reducing saving (and, therefore, contributing more to general expenditure) in
the expectation of benefiting from partner 1's higher replacement ratio.
Generally, however, the financial outcomes from divorce are not equitable,
and one member (the one who has the main child caring responsibility,
usually the woman) ends up being significantly worse off.
6.15 We see from the results in this section that a replacement ratio of
2/3rds, including state benefits, is quite modest, even for higher income

Family Fortunes A Guide to Saving for Retirement

25

groups. It is probably better for such a target to be placed high, so that it

can be undershot without causing extreme difficulty.
6.16 The savings levels that are needed to reach the replacement ratios
recorded are higher than those observed in Family Spending (ONS, 1999). If
these are reliable measurements, households are likely to experience a drop in
their standard of living at retirement. This might be reasonable for some
households, although, as we observed in Section 2, it is likely that pensioner
households would prefer higher levels of consumption. However, unless
saving levels amongst the lower paid can be improved, they will experience
increased poverty in retirement.
.

Means Testing

7.1 The results in the previous sections concentrate on the effects of

private savings, including state benefits, only to the extent that they are
`earned' by virtue of National Insurance payments. However, certain groups
are eligible for means tested benefits. (The means tested benefits included in
the model are summarised in the Appendix). Some research has been done on
the way in which means testing affects people's work and savings' decisions,
but most of it is theoretical (for example, Feldstein, 1987; Friedberg, 1999) or
qualitative (for example, evidence given by Age Concern to the Select
Committee in Social Security, Fifth Report). By incorporating means testing
into the model that we have used for this paper, we can see how the
expectation of a certain level of means tested benefits will affect the need to
save and the `standard of living' of the household.
7.2 Table 7.1 presents these figures for our `standard' household of a
couple with two children, assuming that:
they have a total income of 15,000, split equally between the couple;
they retire aged 65 and live until age 89; and
the families that buy a house leave the value of the house as an inheritance.
Table 7.1.

Buy house at 30
Rent inheritance

Minimum spend and average saving

Means test?

Multiple

No
Yes
No
Yes
Yes

1.14
1.34
0.83
0.95
1.23

Rent no inheritance
Notes: # average saving to age 65
*
figure in brackets is saving averaged over lifetime

Average saving (%)

including house excluding house
11.5# (18.1)*
11.4 (18.1)
^
^
^

7.4 (11.4)
6.9 (11.2)
20.8 (29.1)
26.0 (33.9)
9.0 (6.6)

26

Family Fortunes A Guide to Saving for Retirement

7.3 Before commenting on the figures in Table 7.1, it is worth noting

that, throughout their working lifetimes, these households pay tax and
National Insurance. Their entitlement to income support depends largely on
their children, although they also receive some in the years immediately
preceding retirement.
7.4 Once they reach retirement, their entitlement to the minimum
income guarantee (MIG) depends on their income. The household that buys
a house receives an increasing proportion of its income (from under 10% to
approximately 25% at 85) from the MIG. The proportion of income provided
from means tested benefits increases throughout retirement, since:
state pensions in payment only increase in line with prices;
we have assumed that annuities bought from private savings will only
increase in line with prices; and
the MIG is increased in line with earnings.
7.5 The household that lives in rented accommodation and wants to
leave an inheritance receives virtually none of its retirement income through
means tested benefits. To match the inheritance of the family that bought a
house, it has saved a high proportion of its income whilst in work, and has to
continue to save at an even higher rate throughout retirement. This can be
seen by comparing the average rate of saving over the household's lifetime to
the average rate up until age 65. In this case the former is higher than the
latter, whereas normally saving would fall after retirement. The income
generated by its savings exceeds the income test for the MIG, although most
of it has to be reinvested.
7.6 The results suggest that receipt of means tested benefits enables
higher levels of consumption that can be maintained in retirement partly
through increased saving. This is contrary to the received wisdom that means
testing reduces the incentive to save (for example, Feldstein, 1987), but
consistent with the logic of the model. We have assumed that the household
targets a fixed inheritance, and consumption and saving are the only
variables that can be modified to achieve this.
7.7 We could, instead, assume that a household wants to maintain a
fixed level of consumption, and look at how savings and inheritance must
alter to accommodate this. For example, it seems reasonable to assume that
no household would permit its consumption to fall below the `poverty' spend.
However, in Table 7.1 we see that, if there are no means tested benefits, the
household that rents and wants to leave an inheritance would only be able to
sustain consumption at 83% of the poverty spend. If we impose a minimum
level of consumption equal to 100% of the poverty spend, then the household
will end with a debt equal to about 25% of the value of the house.
7.8 If means tested benefits were introduced, the household could use
the additional income to increase consumption, to increase saving, or both.
The outcomes from the former two alternatives are shown in Table 7.2.

Family Fortunes A Guide to Saving for Retirement

27

Table 7.2. Effect of means testing for a household that rents, if

consumption and inheritance are prioritised differently
No means test
Means test
Notes:

#
*

Priority

Multiple

Inheritance
Consumption
Inheritance
Consumption

0.83
1.00
1.00
1.23

Average saving
20.8%#
10.3%
23.0%
8.9%

(29.1%)*
(2.6%)
(26.4%)
(6.6%)

Inheritance
100%
26%
70%
0%

average saving to age 65

figure in brackets is saving averaged over lifetime

7.9 The actual level of consumption is likely to be between 1 and 1.23

times the poverty spend, and the inheritance and saving will vary
accordingly. However, if the household's priority is to maximise
consumption (which at low levels of consumption seems reasonable), means
tested benefits will not increase the level of saving.
7.10 The figures in Table 7.1 suggest that housing is a good
investment, certainly relative to renting and trying to save in addition. This
is partly a function of the assumptions made about the growth in the value
of property. However, eligibility to some means tested benefits depends on a
capital test. Frequently the value of a home will be excluded from the
capital test, whereas other savings will not. So, saving elsewhere can make a
household that rents ineligible for means tested benefits. On the other
hand, home ownership is a very illiquid investment, whereas ISA savings,
for example, could be far more flexible. Since we have assumed that
households can borrow against their property, this is not an issue that the
model addresses.
7.11 Now we will look at the position of different households.
Table 7.3.

10-year break
No children
Single parent (15 K)
Single parent (7.5 K)
Notes:

#
*

Minimum spend and average saving

Housing

Multiple

Purchase
Rent
Purchase
Purchase
Purchase

1.20 (0.95)y
0.85 (0.65)
1.33 (1.31)
1.72 (1.47)
1.47 (0.83)

figures in brackets ignore means tested benefits

average saving to age 65
figure in brackets is saving averaged over lifetime

Average saving (%)

including house excluding house
15.9
^
20.2
21.9
36.6

9.6
27.0# (33.7) *
13.4
15.0
29.9

28

Family Fortunes A Guide to Saving for Retirement

7.12 The standard of living of the household with two children, where
one parent takes a ten-year break from paid work after the birth of the first
child, is about 12% less than that of the household where both parents work
throughout. They also have different income and expenditure patterns. The
latter household pays off its mortgage by age 59, having had to accrue extra
debt when the second child is born. Once it has paid off this debt (and its
initial debt), it does not have to borrow (apart from the mortgage), apart
from a small amount each year after age 65. This could be viewed as equity
release on the house.
7.13 In contrast, the former household does not pay off its mortgage
until its death, having made no payments for the seven years between the
birth of the second child and the `caring' partner's return to work.
7.14 Both households pay tax and National Insurance throughout their
working lives, and continue paying tax throughout retirement.
7.15 Turning to income, the household where one parent takes a break
receives more means tested benefits during that break than the household
where no break is taken, even though the latter household is eligible for child
care support. The household that rents also receives housing benefit during
this period. Once in retirement, the former household also receives more
means tested benefit, although, once the `caring' partner returns to work, it is
able to save towards a personal pension.
7.16 The household without children has an easier time, financially. It
pays off its initial debt and mortgage within the allotted time, and is able to
save from age 30 towards a personal pension. It receives virtually no means
tested benefits whilst in work, and only becomes eligible for means tested
benefits aged 78. By age 85, about 10% of its income is due to the MIG.
7.17 The single parent with an income of 15,000 has the highest
standard of living of those in this section. Whilst in work, the household
receives means tested benefits similar to those of the other households with
two children, as means testing is based on household income. However, the
MIG is lower for a single person than for a couple, and this person's savings
are sufficient to lift the household above the eligibility threshold for the
single person's MIG.
7.18 The household with a single parent and an income of 7,500
receives a significant proportion of its income, at all ages prior to retirement,
from means tested benefits. It has to extend the period over which it pays
back its initial debt, and it takes an extra two years to pay off its mortgage.
At the same time, it is able to save enough so that, in retirement, it can
generate sufficient income to be ineligible for the MIG until reaching age 79.
However, the ability to save is a function of the model. When awarding a
mortgage or permitting borrowing, the model does not exercise credit ratings
nor deal with the effects of social exclusion. In its world, payment of the
mortgage can be extended and increased borrowing allowed, to the extent
that it is secured against the inheritance. In real life, people in lower income

Family Fortunes A Guide to Saving for Retirement

29

groups will frequently be refused credit, except, perhaps, at exorbitant rates,

and will find their homes repossessed if they cannot afford their mortgage
payments. Also, to keep comparisons consistent, the house price has been
fixed at two times income. In most parts of the U.K. people will struggle to
find a house, on which a building society would be prepared to lend, for
15,000.
7.19 The Government's policy has been to target means tested benefits
at low income families with children. The results of the model show that, for
the households which we have considered, this has been quite effective. The
standard of living of the household without children has been increased by
1.5% due to means tested benefits. The other households, with incomes of
15,000 p.a., experience increases of between 17% and 31%. The standard of
living of the household with an income of 7,500 is increased by 77%.
.

Sensitivity Analysis

8.1 Investment Returns

8.1.1 Except for the defined benefit scheme, all the investments are
`money purchase' in type. That is, their value depends on investment
markets, and can go up or down. (In theory, the pension from a defined
benefits scheme will not change in value solely due to market movements.) A
deterministic model does not capture this effect, and so we will investigate
the consequences of different investment return assumptions.
8.1.2 Clearly, if we assume, in this section, that investment returns are
lower than those which we assumed in the previous sections, the household's
standard of living will fall. What is interesting is how sensitive the standard
of living is to falls in investment returns. Some of the results of Section 4 are
repeated here, assuming that real investment returns are about 2% lower on
average (see Appendix). The result of this `poor' investment return is
compared against the `normal' return assumed in Section 4.
8.1.3 By comparing the two columns of multiples in Table 8.1.1, we can
see that the lower investment returns have reduced each household's standard
of living by about 6% for those retiring at age 65 and 10% for those retiring
at age 60. However, this assumes that the households have had the benefit of
foresight and anticipated the poor investment performance by increasing
their rates of saving. If they expected the higher rate of investment return
they would have saved less, and so their standard of living in retirement
would fall further. For example, suppose that the first household in the table
had maintained a standard of living of 1.96 times the minimum spend whilst
in work and that investment returns had been `poor' rather than our original
expectation. Then, on retirement, its standard of living would have had to
fall by approximately 20% (that is, to 1.57 of the minimum spend) if it kept
the same target inheritance.

30

Family Fortunes A Guide to Saving for Retirement

Table 8.1.1.

This is as Table 4.5.1, but with lower investment returns

Joint
income

Retirement
age

30,000

65

At 30 only

60

At 30 and 40
At 30 only

65

At 30 only

60

At 30 and 40
At 30 only

50,000

House
purchase

Age at
death

Multiple
(`poor')

Multiple
(`normal')

89
95
89
89
95

1.84
1.82
1.52
1.64
1.62

1.96
1.94
1.60
1.78
1.77

89
95
89
89
95

2.80
2.76
2.27
2.47
2.44

3.02
3.00
2.42
2.74
2.72

8.1.4 The groups that lose most in these circumstances are the
households with no children, since they have been able to save more than
those with children. They lose both because their savings grow less than
expected and because they produce a lower income. The results in Table 8.1.2
show standards of living falling by over 10% in some cases. Again, this
presupposes that they have saved enough to compensate them for lower
investment returns. Had they anticipated the higher rates of return, they
would experience a further fall in their standard of living in retirement.
8.1.5 Defined benefit occupational pension schemes are the only
private savings that households in the model have access to that are (at least
notionally) protected from poor investment returns, but `public' savings
have similar characteristics. State pensions are essentially defined benefit,
Table 8.1.2.

This is similar to Table 4.7.1, but with lower investment

returns

Joint
income

Household
status

Retirement
age

Multiple
(`poor')

Multiple
(`normal')

30,000

Couple

65
60
65
60

2.11
1.87
3.27
2.89

2.29
2.09
3.57
3.24

65
60
65
60

3.24
2.85
4.84
4.30

3.59
3.26
5.35
4.84

65
60

1.80
1.59

1.91
1.74

Single
50,000

Couple
Single

15,000

Single

Family Fortunes A Guide to Saving for Retirement

31

and so provide a floor below which income cannot fall, thus limiting an
investors' `downside' risk. Similarly, means testing cushions savers, to some
extent, from the effect of poor investment returns. For example, allowing for
means testing, the standard of living of the single person with an income of
15,000, retiring at age 60, would only fall to 92% rather than 91% of the
Table 4.7.1 level. Couples with children on this level of income would only
see a fall of 2%, if there were means tested benefits, rather than of 6%, if
there were no means tested benefits. However, we have seen, in Section 7,
that means tested benefits are only significant at quite low levels of income.
It is unlikely that those with earnings greater than average would want to
retire with incomes that leave them eligible for the MIG.
8.2 Inheritance
8.2.1 Assuming that households leave the full value of their homes as an
inheritance clearly limits their ability to spend whilst in retirement. In this
section we repeat some of Table 4.5.1, but allow for different levels of
inheritance.
Table 8.2.1.
Joint income
30,000

This is similar to Table 4.5.1, but with different inheritance,

retirement at age 65 and death at age 89
House purchase
At 30 only
At 30 and 40

50,000

At 30 only

Inheritance

Multiple

0%
50%
100%
0%
100%

2.21
2.08
1.96
2.12
1.60

0%
100%

3.44
3.02

8.2.2 We see that households which make only one house purchase can
increase their standard of living by about 13% if they are able to spend
against the value of their home. The household with two purchases can
increase its standard of living by about 32%. The difference represents the
higher value of the second house, as well as the relatively favourable rate of
property growth that the model assumes.
8.2.3 This demonstrates the potential attractiveness of equity release
schemes, where they can be devised in a way acceptable to both the consumer
and the provider (Le Grys et al., 2001).
8.3 Annuitisation
8.3.1 Given the current interest in income drawdown and forced

32

Family Fortunes A Guide to Saving for Retirement

annuitisation, it is interesting to see what happens to each household's

consumption if it decides not to purchase an annuity, but to continue
drawing down on its savings. Because this is a deterministic model, the risks
that the household becomes exposed to are limited to the effect of longevity
and the different costs involved. Formal income drawdown schemes are
relatively expensive, but this model ignores this cost. Effectively, we are just
comparing an investment in gilts, with a mortality guarantee, with an
investment in a mixed equity/gilt portfolio. We would expect the latter to
give a better return for those households that end at an earlier age, and the
former to give a better return for those households that are longer lived.
8.3.2 We illustrate the cost by repeating Table 4.5.1, assuming that
assets are freely drawn down until death.
Table 8.3.1.

This is as Table 4.5.1, but also includes the multiples when

draw down only stops at death

Joint
income

Retirement
age

30,000

65

At 30 only

60

At 30 and 40
At 30 only

65

At 30 only

60

At 30 and 40
At 30 only

50,000

House
purchase

Age
at death

Multiple
(d/d)

Multiple
(d/d stops)

89
95
89
89
95

1.97
1.90
1.61
1.80
1.74

1.96
1.94
1.60
1.78
1.77

89
95
89
89
95

3.05
2.93
2.43
2.77
2.66

3.02
3.00
2.42
2.74
2.72

8.3.3 The result is exactly as we expected. Households with an end age

of 89 are marginally better off if they do not purchase an annuity at 69,
whereas those that survive until 95 are worse off. The result reflects the
expectation of life inherent in the mortality tables used to calculate the
annuity.
.

Conclusion

9.1 It is apparent from the output of this model that saving is a

moveable feast (or famine), and that both the need and the ability to save
depend crucially on household circumstances. It is quite powerful in
demonstrating how a household's standard of living is vulnerable to family
circumstances and to savings decisions. Figure 5.2 shows how even a

Family Fortunes A Guide to Saving for Retirement

33

household receiving about twice the median income could experience a long
time when it will be unable to save (when it has children). When children
eventually leave home, we can also see that it is important that the income no
longer needed to support the child is saved, in order to maintain the
standard of living in retirement.
9.2 From Section 6, we see that, whilst saving rates can be low, or even
zero, at younger ages, once children have left home they should be increased
to around 30% of earnings. In many cases it could be necessary to continue
saving in retirement. This pattern is not sensitive to income, and savings'
rates at each age increase only slowly with household income. However, the
model assumes that the adults in the household can continue to find paid
employment, and that their earnings maintain their real value until
retirement. For many people this is not the case. Households that follow the
investment strategy that the model appears to recommend, and then find
themselves without paid employment in the years prior to retirement, will be
very vulnerable in retirement.
9.3 Although it is not optimal in terms of consumption, a more cautious
approach would be to vary one's standard of living and maintain a more stable
saving's rate. Although this might lead to some hardship whilst with young
children, for example, it could have the advantage of bringing forward saving,
thus reducing vulnerability to loss of employment prior to retirement.
9.4 Other sources of information (for example, NatWest, 1999) imply
that, whilst wealthier households need a higher weekly pension in order to be
`content' in retirement, there is a large difference in the level of income for
contentment or distress. Wealthier households, thus, have a wide margin of
error when setting the target for their savings. Households with lower
incomes have narrower margins between contentment and distress, so they
are more vulnerable to downside risk. Downside risk could arise due to
market falls, the increasing cost of annuitisation, or because households
underestimate the level of saving required. Section 8 gave some indication of
how households might be affected by these events. In the example used for
poor investment returns, consumption had to fall by 1/5th.
9.5 The usual target for income in retirement is 2/3rds salary, which the
model's output suggests is quite low (see Section 6). Occupational pension
schemes cannot provide a replacement rate higher than 2/3rds, so even those
with good expectations from their employer's scheme should probably be
saving elsewhere.
9.6 The most certain ways currently available to protect against adverse
experience are through insurance or through state support. The results in
Section 7 showed how means testing could protect the downside risk of lower
income households, but also demonstrated that it could have other
consequences that were not desirable, such as reducing the incentive to save.
`As of right' benefits, such as the BSP and S2P, also reduce downside risk,
but might not have the same effect on saving.

34

Family Fortunes A Guide to Saving for Retirement

9.7 The value of an insured income stream was demonstrated in

Section 8.3. Annuitisation protects against longevity as well as falls in
the stock market. However, annuity purchase locks the purchaser into
fixed-interest returns, and so limits the opportunity to benefit from
potentially better returns elsewhere in the market. Also, for those with
impaired life expectancy, a normal annuity would probably not offer
value for money.
9.8 There are some groups for whom private saving is not sensible, and
it would be wrong for providers or government to encourage these groups to
assume that the private sector is sufficient to meet their financial needs.
This model can indicate when saving is not possible, or when the rational
decision for a household would be to decide not to save for retirement.
However, there are other important reasons for saving. For those households
where retirement saving is not necessarily a priority, we have seen that it
might be better to save in more flexible arrangements, such as ISAs.
Although these do not have quite such beneficial tax status, they can add
value, because they have fewer constraints.
9.9 The model presented in this paper is not a blueprint for saving
success. It only illustrates a desirable savings pattern, given certain criteria,
with no assurance that the savings pattern will continue to be desirable if
circumstances change. People need to be made aware, continually, of their
current and future expected financial circumstances if they are to plan
appropriately for retirement. The annual benefit statement proposed by the
Government could be a step in this direction (PFAG, 2000). However,
information is not necessarily power. A simple benefit statement will not give
people the level of understanding needed to plan for the future, and those in
lower income households might not be able to adjust their savings in reaction
to the information provided.

References
Cook, M. & Johnson, P. (2000). Saving for retirement: how taxes and charges affect choice.
FSA Occasional Paper Series 8.
http://www.fsa.gov.uk/pubs/occpapers/op08.pdf
FSA (2000). FSA factsheet: stakeholder pensions and decision trees.
http://www.fsa.gov.uk/pubs/public/cred27ap.pdf
Feldstein, M. (1987). Should social security be means tested? Journal of Political Economy,
95, 3, 468-484.
Friedberg, L. (1999). The labor supply effects of the Social Security Earnings Test. NBER
Working Paper No. W7200.
Hamilton, M. (2000). Retirement saving strategies for Canadians. CIA National Symposium
on Ageing.
Hauser, R. (1997). Adequacy and poverty among the retired. AWP3.2, OECD.
Le Grys, D.J. et al. (2001). Report on equity release mechanisms.
http://www.actuaries.org.uk/equity___ release/rpt.pdf

Family Fortunes A Guide to Saving for Retirement

35

Murphy, M. & Wang, D. (1999). Forecasting British families into the twenty-first century. In
Changing Britain: families and households in the 1990s. Ed: S McRae, Oxford University
Press.
NAPF (2000). Annual survey 2000. National Association of Pension Funds.
NatWest (1999). NatWest pensions index. Vol. IV, National Westminster Life Assurance
Limited.
ONS (1999). Family spending, 1998-99. London: HMSO.
Pension Forecasting Advisory Group (PFAG) (2000). Planning your future.
http://www.dss.gov.uk/publications/dss/2000/penfor/penfor.pdf
Sinclair, P.J.N. (1998). Why do the poor save so little. Discussion Paper 20/98, Department
of Economics, University of Birmingham.

36

Family Fortunes A Guide to Saving for Retirement

APPENDIX
ASSUMPTIONS AND BACKGROUND

Results have been presented for the following combinations of

assumptions.
A.1

Household Statistics

Start age
Retirement age
Drawdown age
Age at death
Household type
Household income
Share
House purchase
Rent
Spending
Annuity
A.2

25
60 and 65
69
89 and 95
Single; couple; couple with two children
15,000; 30,000; 50,000
100%, 0% and 50%, 50%
Two times income at 30; two times income at 30,
and four times income at 40
3,600 p.a. (before purchase of first house),
increasing in line with inflation
`poverty' spend 3,945, no drop and drop at
retirement
Index linked

Economic Assumptions

Inflation
Real salary growth
Dividend yield
Real dividend growth
Real bond yield
Real property growth
Mortgage interest
Investment expenses
Cost of unsecured
borrowing

Standard basis

`Poor' basis

0%
1.5% (in excess of inflation)
2.0%
3.5%
2.0%
3.5%
1.0% in excess of bond yields
1.0%

0%
1.5% (in excess of inflation)
1.0%
2.0%
1.5%
3.0%
1.0% in excess of bond yields
1.0%

2.0% in excess of bond yields

2.0% in excess of bond yields

A.3 Tax and State Benefits

A.3.1 These are assumed to be as at the tax year 2000/2001, and
increase in line with inflation.
A.3.2 State pension allows for the introduction of the State Second
Pension from 2002.

Family Fortunes A Guide to Saving for Retirement

37

A.3.3 The means tested benefits included are:

the Working Families Tax Credit;
Housing Benefit; and
the Minimum Income Guarantee and Pension Credit.
A.4 Savings Models
A.4.1 There are several different reasons why individuals choose to
save. Commonly identified models are:
a lifecycle model, where the desire is to smooth consumption;
a bequest model, where the desire is to leave an inheritance; and
a precautionary model, where the desire is to protect against the risk of
unlikely events, such as early death.
A.4.2 The model that we use is a combination of the lifecycle model and
the bequest model, since the target is to maintain the household's standard of
living whilst permitting an inheritance on death. By combining these two,
the household, effectively, has savings put aside, should an unexpected event
occur.
A.4.3 The risks of following this approach are:
(1) that the households will not have sufficient precautionary savings; and
(2) that the household is `myopic', that is that it underestimates its
longevity.
A.4.4 Using insurance can reduce both these risks. For example, using a
portion of the `inheritance' savings to buy insurance can reduce the risk that,
on early death, the inheritance will be less than expected. Similarly, the
model effectively protects against `myopia', by assuming that savings are
used to purchase an annuity no later than at age 70. If income drawdown
were continued indefinitely, however, the household would be exposed to
mortality risk.

Printed by Bell & Bain Ltd., Glasgow