all households. Spouses’ assets are added to those of the head and the combined
unit is counted only once. For the individual-based analysis the assets of a married
household are divided by two and then each adult in an age band is counted as a
distinct unit.
The differences in age profiles across different types of unit are not trivial. In
particular, stock wealth is differentially lower for the youngest tax units, so that the
resulting tax unit and individual age gradients are much steeper than household
unit gradients. For example, at a household unit in the UK, the ratio of median
stock wealth of the 60–69 age band is 6.4 times that of the 20–29 age band. The
comparable number at the tax unit or individual unit level is about 10.6. These
ratios are sufficiently different to raise questions about the sensitivity of age profiles
to the widespread use of household unit analysis, particularly across countries, and
suggests this as an interesting topic for future research .
In this chapter we have discussed what we can learn about the age profile of saving
and saving rates from UK household data. As in other countries, inc ome and
expenditure profiles display strong hump shapes, both in cross section and also after
adjusting for possible cohort effects. Taking the difference between these profiles
to calculate a residual measure of saving we find that saving rates rise throughout
working lives, as might be expected . Somewhat surprisingly, however, saving rates
do not decline during retiremen t, although there may be a role for differential
mortality in explaining some of this lack of decline amongst the very oldest ages.
Of course, such a residual definition of saving is not without problems. Not least,
it will fail to count as saving any (unspent) capital gains on existing funds, or any
contributions into savings products that are not measured as income (such as emplo-
yer pe nsion contributions). Such effects will explain why saving rates may appear
to be ‘‘low’’ during working life, but would not explain the positive saving and
saving rates of the elderly.
We have also shown, however, that there has been substantial variation, over age
cohorts and ultimately over time, in both the contributions to, and incomes from,
various types of pension income. Such changes have occurred as a result of both
economic factors (part icularly productivity growth and cyclical factors) and reform
to either the rates or structure of the UK pension system. The UK now has three
groups of households, or individuals, distinguished by their experiences of the
pensions and savings environment. The first group are the oldest pensioners, who are
typically poor and rely almost entirely on flat-rate state pension benefits, as they were
already on the verge of retirement when SERPS w as introduced. The second group
are those currently retiring, who have built up substantial SERPS or occupational
pension benefits but who w ould not have had much chance to contract out into
private DC schemes. These pensioners are younger than the first group and also tend
7.6 Conclusions 303

Get Life-Cycle Savings and Public Policy now with O’Reilly online learning.

O’Reilly members experience live online training, plus books, videos, and digital content from 200+ publishers.