Life-Cycle Savings and Public Policy

saving. In this sense, the two surveys are mutually consistent. It is well possible that

the SHIW understates ‘‘true’’ consumption and therefore overstates savings at

some ages. This bias is likely to be more severe for the upper tail of the saving distribu-

tion. Whether the difference between the two survey measures of saving depends

on sample composition, survey design, or other effects is the topic for future

research.

4.8 CONCLUSIONS

We use Italian micro-level data to construct several measures of saving: the relevant

measures are based on a ‘‘residual’’ definition (income minus consumption), whereas

in some cases we look at saving as the difference in wealth levels at two different

points in time. We have available a large cross-sectional sample of Italian households

for a number of years; this sample also contains a small panel component, which we

exploit to look at financial saving and real saving. We also propose a non-standard

definition of saving based on the notion of ‘‘earned income’’ and we refer to this as

‘‘mandatory saving’’ to accompany the traditional definition of saving based on

disposable income (called discretionary saving). In our study the former is a

definition of saving that takes account of both social security contributions and

contributions to a severance pay fund and it stresses the importance of ‘‘saving for

FIGURE 4.14 Comparing ISTAT and SHIW median expenditures, income, and saving (thousands

of 1998 euro). The figures plot consumption, income, and discretionary saving in the 1991 ISTAT

survey and in the 1991 SHIW survey. The fitted lines are obtained by a regression on a third-order age

polynomial. All values are expressed in thousands of euro.

4.8 Conclusions 143

retirement’’ through mandated arrangements. In line with the work of Jappelli and

Modigliani (1998), the novelty of this approach is that when using earned income

we also account for the benefits paid out to retirees from these retirement provisions

and treat them as a form of ‘‘mandatory dissaving’’. The reason for resorting to this

non-standard measure of saving, which we present along with the traditional

measure of private savings, is simply to stress the importance of the institutional

setting for old-age insur ance in measuring saving. We argue that in Italy a large

fraction of household resources is mandated to public pensions. Thus, whatever the

potential substitutability between the two forms of wealth, the desire to accumulate

resources for consumption in old age must be reduced for Italian consumers. For the

same reason also, the life-cycle pattern of mandatory saving is crucial to the

understanding of the dynamic of discretionary saving.

By making assumptions on the relationship between age effects, cohort effects,

and time effects, w hich typically affect saving choices, we are able to estimate

cohort-adjusted age profiles for the two definitions of saving. The cohort

adjustment proved very important in our data, particularly for older generations,

hence suggesting that differences in cohorts cannot be neglected when explaining

saving behavior.

We show that discretionary saving is positive at all ages. Similar patterns arise

when we consider a number of household characteristics, such as education and

employment of the head of the household, family composition, and homeowner-

ship. For all these groups, discretionary saving does not fall below zero, except for

non-homeowners at very old ages. This means that families hold a positive amount

of private wealth until the end of the life cycle and there might be a combination of

different motives for private wealth holdings. For example, private wealth is be-

queathable and the fact that non-homeowners decumulate more than homeo wners

suggests that bequest motives may play a role. Other potentia l explanations are

related to risk exposure or transaction costs. Whereas pension arrangements insure

against longevity risk and the national health system (essentially based on universal

coverage) partially insures against health risks, there might be other risks late in life or

transaction costs may be more severe. For example, elderly people may be

particularly concerned with permanent illness and disability and long-term care.

Hence, for Italian households, there seems to emerge some substitutability

between private wealth and pension wealth, but also different degrees of fungibility:

whereas pension wealth finances consumption during retirement, private wealth

may be used for intergenerational transfers and to cover other contingencies at very

old ages.

ACKNOWLEDGMENT

This paper is part of a research project on ‘‘Structural Analysis of Household Savings and Wealth

Positions over the Life Cycle.’’ The Training and Mobility of Researchers Network Program (TMR)

144 Chapter 4 Household Saving Behavior and Pension Policies in Italy