Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                

National Saving

Download as pdf or txt
Download as pdf or txt
You are on page 1of 28

re p070 24-11-95 13:48:07

Ricerche Economiche (1995) 49, 329–356

National saving and social security in Italy


NICOLA ROSSI
Department of Economic and Financial Studies, University of
Rome “Tor Vergata”, Via di Tor Vergata, 00133 Rome, Italy

IGNAZIO VISCO
Research Department, Bank of Italy, Rome, Italy

Summary

We have recently suggested that net government transfers have sub-


stantially contributed to the decline of the aggregate Italian saving
rate. Changes in social security laws and regulations which took place
in the late 1960s and early 1970s apparently weakened the link
between contributions and benefits permitting a time path of aggregate
consumption in excess of what would have occurred in the absence
of such changes. In this paper, these results are revised and extended
and, if anything, strengthened: slightly less than half the reduction in
the private “equilibrium” saving rate observed over the last 30 years
appears to be due to the increase of social security wealth. The paper
also assesses the likely impact of the 1992 social security reform and
suggests that in the long run the impact might be sizeable accounting
for up to a 3 points higher saving ratio.
 1995 Academic Press Limited
J.E.L. Classification: H55, E21.
Keywords: Saving behaviour, Social Security.

1. Introduction

According to official sources, aggregate Italian saving rates have


been recently experiencing a substantial decline. The timing and
the extent of such decline largely depend on the specific definition
of saving as well as on the researcher’s attitude toward the host
of measurement issues involved.† However, for gross as well as
† The standard national income accounts (NIA) definition is inadequate on
various respects. First of all, consumption is usually defined as total expenditures
on consumer goods and services. The former include expenditures on consumer
durables which should be more properly considered as a component of household
investment. A proper (economic) definition of consumption should instead only
include the services provided to households by the stock of durables. A second
issue arises when we consider that in the NIA definition of income interest
payments and receipts are included at their nominal value. As it has been
recognized for quite a time, however, income should be adjusted for the loss of
329
0035–5054/95/040329+28 $12.00/0  1995 Academic Press Limited
330 N. ROSSI AND I. VISCO

for net rates, for national and for private saving ratios, for both
“adjusted” and “unadjusted” magnitudes, major trends appear to
be beyond dispute and provide a somewhat unexpected description
of the behaviour of Italian consumers. As it turns out, from a peak
of about 24% at the beginning of the 1960s, the net national saving
rate has declined to about 9% in 1993.
In particular, in the early 1990s private saving rates (defined
as net of depreciation, adjusted for durables and inflation and
computed as ratios to net national disposable income) have been
fluctuating around 10%, nine percentage points down from the
average saving rate prevailing in the 1960s. In the same years,
government saving (adjusted for inflation) has plunged to an
unprecedented −1%, after more than 30 years of current account
surpluses averaging to 2%. As a result, today’s overall Italian
saving rate is about 15 percentage points below the one observed
during the so-called “Italian economic miracle”.
As can be easily gauged from Figure A where national, private
and government saving rates (corrected for durables and inflation)
are presented, the beginning of the decline in the Italian saving
rate can be placed just short of the middle of the 1970s, in
conjunction with the first oil shock, when the growth rate of the
Italian economy first started to fall. It is only in the 1980s,
however, that the reduction of the national saving rate became quite
substantial, fully reflecting the progressive decline of government
saving along with the negative trend of private saving.
Several alternative explanations for the decline of the saving
rate have been put forward by the most recent research.† In a
recent article (Rossi & Visco,1994), we have scrutinized the impact
of fiscal policy on private sector behaviour to suggest that a
thorough assessment of the interplay of private and public decisions
can shed some light on the evolution of the Italian saving rate in
the most recent decades. In particular, focusing on the 1951–1990
purchasing power that inflation induces on the stock of nominally denominated
debt. This is particularly relevant when national saving is split between its
private and government components. Ignoring the loss of principal due to inflation
implies an overestimation of private savings: this simply amounts to neglecting
a possibly important source of revenue for the government, the so-called “inflation
tax”. Finally, even if it is controversial, it is likely that saving should be considered
net of the depreciation of the capital stock. In this section, the movements of
national, private and government saving rates (all computed in relation to the
net national disposable income) are examined allowing for a proper treatment of
consumer durables and accounting for the capital stock depreciation and the
“inflation tax”. For detailed statistics on post war Italian saving trends see the
Statistical Appendix to Ando, Guiso and Visco (1994). The updated time series
used in this paper are contained in its working paper version, available on request
from the authors.
† Besides the papers by Fornero and Castellino (1990) and Jappelli (1991), the
interested reader is referred to Modigliani and Jappelli (1993) and to Ando, Guiso
and Visco (1994).
NATIONAL SAVING AND SOCIAL SECURITY IN ITALY 331
25

20

15
%

10

–5 51 53 55 57 59 61 63 65 67 69 71 73 75 77 79 81 83 85 87 89 91 93
52 54 56 58 60 62 64 66 68 70 72 74 76 78 80 82 84 86 88 90 92

Year

FIGURE A. Adjusted net saving rates. –Ε–: National; –Χ–:


government; –Μ–: private.
Source: Ando, Guiso and Visco (1994), Statistical Appendix, updated
and revised.

period, we submitted that the changes in social security laws and


regulations which took place in the late 1960s and early 1970s
severely weakened the link between contributions and benefits
permitting a time path of aggregate consumption in excess of what
would have occurred in the absence of such changes. It is worth
recalling that those changes entailed both a replacement effect
due to the introduction of an unfunded social security system
for specific segments of the population as well as, for a given
institutional structure, a lifetime wealth increment. The evidence
presented in our paper apparently provides a clear-cut outcome,
indicating that the increase in social security benefits has sub-
stantially contributed to the decline in private saving that has
been observed in the course of the 1970s and the 1980s. We
estimated that one-third (or about three percentage points) of the
decline of private saving (in terms of net private disposable income)
observed since the 1960s may be attributed to the rise in the net
social security wealth to income ratio.
That changes in laws and regulations determining retirement
benefits might be at the root of important economic phenomena of
the last 20 to 30 years comes as no surprise. Italy is the OECD
country with the highest pension expenditures to GDP ratio and,
in fact, Italy’s pension system, since the early 1980s, has been the
subject of lively discussions aiming, on the one hand, at controlling
the future pattern of expenditures and at suggesting, on the
332 N. ROSSI AND I. VISCO

other hand, appropriate legislative changes. Many attempts at


structurally reforming the system took place in the 1980s in an
effort to halt the rapid growth in the pension expenditures. At
last, in 1992, a social security reform was enacted: while certainly
contributing to reverse the present negative trends, it was however
far from sufficient. An overall social security reform has been
recently approved by Parliament: its effects still need to be thor-
oughly assessed.
The time is ripe, therefore, to revise and update Rossi and Visco’s
(1994) exercise. A reassessment of the likely impact of social
security on the aggregate saving rate is the (admittedly limited)
purpose of this paper. In Section 2 a brief review of the main
patterns of the Italian social security system is presented, providing
quantitative evidence on its more recent evolution. Section 3 con-
siders the time-series evidence on the substitutability of social
security and fungible wealth. Section 4 provides some micro-
economic evidence on the same topic. Section 5 makes use of the
time-series and cross-section evidence to provide a preliminary
assessment of the impact of social security reforms on the future
evolution of the saving rate and concludes the paper.

2. The Italian social security system

Social security transfers have been rising in Italy dramatically


over the last 40 years from a level of 7 to one around 22% of net
national income. The entire rise (about 15 points) is accounted for
by the movements of pensions, public health services having grown
in line with income over the period.
An analysis of the determinants of the rapid growth in pension
expenditures highlights the role of demographic trends and of
substantial increases in the number of pensions (throughout the
period) as well as in average pension benefits (from 1976 onward).†
The growth of pension expenditures which took place from the
turn of the 1960s onwards reflects “the impact of the introduction
of special plans for self-employed workers and increased use of
disability pensions for welfare purposes. This rapid growth carried
over into the 1970s in connection with the switch in 1968 from a
contributions-based system to an earnings-based system and the
far-reaching reform of INPS pensions in 1969, in particular the
introduction, and subsequent improvement, of methods of indexing
pension benefits to prices and real earnings. By contrast, in the
1980s the growth of pension expenditure slowed down. This re-
flected the decrease in the number of disability pensions . . . , the

† Franco (1992) reviews the post-war evolution of government transfers in


Italy and provides a detailed account of their determinants.
NATIONAL SAVING AND SOCIAL SECURITY IN ITALY 333

gradual winding down of the hyperindexation of pensions to prices


. . . , the slowdown in growth due to the adjustment of pensions to
real earnings, the attenuation of the effects of self-employed
schemes when they reached steady state, and the introduction of
means testing for welfare benefits.” (Franco & Morcaldo, 1990: p.
107).
From the point of view of government finances, it is crucial to
underline that the spectacular increase of social transfers (and
particularly of pensions) has not been matched by that of social
contributions from whatever source (from employers as well as
from employees). Between 1951 and 1993 total transfers increased
in relation to net national income about three times (with three
“jumps” in 1965, 1975 and 1981, when major institutional changes
took place, with further consequences for the following years). By
no means the corresponding increase of social contributions was a
modest one, having amounted to two and a half times that of
national income. This was not sufficient, however, to avoid a rise
of net transfers (from about 1% of income until the middle of the
1970s to about 5% in most of the 1980s) entirely due to the
generosity of the national social security system. As a result, in
1992, the equilibrium contribution rate (that is, in a pay-as-you-
go system such as the Italian one, the ratio of pension transfers
to wages) was projected to reach by the year 2010 levels well above
any future foreseeable level of the contribution rate. For example,
the 2010 equilibrium rate for the (private) dependent workers
scheme was projected to reach 54%.†
As Franco et al. (1994) have shown, there was a serious imbalance
in Italian generational policy: as of 1990, future generations of
Italians would have to face lifetime net tax burdens much higher
(almost double) than those faced by those just born. A deep reform
of the system was therefore unavoidable and took place at the end of
1992. As a result, theoretical equilibrium rates were substantially
lowered and, for the dependent workers’ scheme, were projected
to reach between 40 and 47% in the year 2010 depending on the
indexation scheme (INPS, 1993). Equilibrium in the long run was
however far from achieved, projections for some of the major
pension funds still implying a long-run equilibrium rate which
largely exceeded any economically meaningful level of the con-
tribution rate. Moreover, the existence of separate schemes still
tended to generate undesirable distributional effects.
The post-war evolution of the Italian social security system is
clearly reflected in the time path of social security wealth (Figure
B). The net pension wealth to national disposable income ratio

† See INPS (1993: p. 132).


334 N. ROSSI AND I. VISCO
7

0 51 53 55 57 59 61 63 65 67 69 71 73 75 77 79 81 83 85 87 89 91 93
52 54 56 58 60 62 64 66 68 70 72 74 76 78 80 82 84 86 88 90 92

FIGURE B. Social security wealth to net disposable income ratios.


–Ε–: Social security wealth (retirees); –Χ–: social security wealth
(gross); –Μ–: social security wealth (contributions); –Ε–: social
security wealth (net).
Source: Beltrametti (1995).

rises dramatically from 0·6 in 1951 to about 4·5 in 1992 to drop to


2·9 after the 1992 reform.† The increase is due to demographic
trends as well as to the deepening and widening of the system
which took place since the late 1950s. At each point in time, it
concerns both the workforce and (to a lesser extent) the retired
population.
The information content of social security wealth should not be
over emphasized, though. Net social security wealth should meas-
ure the actuarial value of the pension benefits which individuals
expect to receive net of the actuarial value of the social security
taxes which they expect to pay. As such, this quantity should
reflect changes over time in such factors as the population age
structure, social security coverage and benefit and tax rules. It is,
therefore, a rather elusive quantity whose measurement involves
formidable assumptions at the individual as well as at the ag-
gregate level. It should be stressed, though, that, whatever the
measurement issues involved, social security wealth is positive by
definition in institutional settings such as the Italian one: it simply
measures the cost of switching from a public pay-as-you-go system
to a private fully-funded system. It tells almost nothing about the

† Social security wealth computations for 1993 assume changes in workers’


retirement choices corresponding to the changes in the legal retirement age.
See Beltrametti (1995), where the major assumptions behind pension wealth
computations are spelled out in detail.
NATIONAL SAVING AND SOCIAL SECURITY IN ITALY 335

long-run sustainability of the social security system which is,


instead, correctly measured by the equilibrium contribution rate.
Consequently, international comparison of pension liabilities (such
as in van den Noord & Herd, 1993) can to a large extent be
questioned (see also Franco, 1995).
However, social security wealth might have a lot to say about
household saving behaviour and labour supply choices. An increase
in pension benefits in the form of, say, changes in the rules
determining social security payments (such as the changes re-
peatedly observed in the 1970s and 1980s) could affect the saving
behaviour of all individuals through their impact on expectations,
since, in a life-cycle framework and for a given retirement age,
such changes would, ceteris paribus, create an incentive for the
current generation to increase its consumption.† Therefore, in
such circumstances, an increase in transfer payments signalling
a permanent increase in pension benefits‡ (the bill being eventually
footed by later generations) could be expected to have an ap-
preciable negative impact on private savings and hence, for a given
level of current income, on the private saving rate. The shrinking
private saving rate would add up to a larger government deficit,
thereby leading to an even larger cut in the national saving rate.

3. Private saving and social security wealth: time-series


evidence

Needless to say, the impact of social security on saving is far from


being a novel issue, having been debated widely at both the
theoretical and empirical levels.§ The controversy over the validity
of the life-cycle model extended à la Feldstein (1976) is still very
much an open issue and, indeed, Auerbach and Kotlikoff (1983),
thoroughly examining the available cross-section and time-series
tests of the role of social security, have argued that, given current
data, neither type of information has much potential for settling
the controversy.
Rossi and Visco’s (1994) results suggest, however, that the Italian

† It should be recalled that, if the retirement age is not fixed, such effect may
be partly offset by induced phenomena of early retirement.
‡ It is implicitly assumed here that the increase in transfer payments is
perceived as sustainable by the agents. This warning forcefully applies to the
Italian case where the inherent risk of unsustainability of the social security
system had been recognized since the late 1970s.
§ See the pioneering works of Feldstein (1974) and Barro (1974). Apart from
Rossi and Visco (1994), aggregate (but inconclusive) time-series evidence for Italy
is reported in Modigliani and Sterling (1983).
336 N. ROSSI AND I. VISCO

experience may represent an ideal case study providing for sub-


stantial variability in social security regulations and hence in the
social security wealth variable. As Figure B shows, the net social
security wealth to income ratio accurately reflects the discretional
adjustment of pensions to prices which took place in 1952, 1958,
1962 and 1965 as well as the automatic indexation first introduced
in 1969, extended in 1976 and subsequently amended in 1983. It
also incorporates the effects of the introduction of special schemes
for self-employed farmers in 1957, for artisans in 1959 and for
shopkeepers in 1966. It points out the major switch from a con-
tributions-based system to an earning-based system (in two steps,
the first one in 1969 and the second in 1976). It finally highlights
the far-reaching consequences of the 1992 reform (as possibly
perceived by the agents). As in Rossi and Visco (1994), we shall
exploit the information contained in the behaviour of the economy
around the periods of change, hopefully avoiding what Auerbach
and Kotlikoff correctly regarded as a major source of mis-
specification.
Following Modigliani (1986), the aggregate behaviour of Italian
consumers may be described by the simple tale of rational, utility
maximizing, agents allocating optimally resources to consumption
over their entire life, as depicted by the stripped down version of the
life-cycle hypothesis of saving. As Feldstein (1974) first suggested,
though, the social security system should influence family’s con-
sumption. If labour-leisure choices are unaffected, this influence
should only pass through the effect of social security benefits and
contributions on its lifetime budget constraint. If there are no con-
straints on net worth prior to the time of death, the social security
impact is summarized by the family’s lifetime social security wealth
which is simply the difference between the discounted value of life-
time net social security benefits and contributions.
In such a case, consumer decisions would be conditional on
human wealth (which may be assumed roughly proportional to
current labour income) and on a variable given by the sum of net
non-human (real and financial) wealth and net pension wealth.
The latter could be weighted by a constant parameter, h, allowing
for a difference between private assets and net social security
wealth due, among other things, (i) to the presence of a bequest
motive, which would imply h>1 (Williamson & Jones, 1983), (ii) to
income uncertainty, which would entail h<1 (Samwick, 1994), and
(iii) to the perception of a risk of unsustainability of the social
security system that would also imply h<1.
In the specification of an aggregate saving function within such
an extended life-cycle framework, the issue of double counting
between disposable income and social security wealth should how-
ever be addressed. In particular, according to the standard NIA
NATIONAL SAVING AND SOCIAL SECURITY IN ITALY 337

definition, disposable income (yd) is inclusive of social security


benefits, essentially to current retirees (ys), and is net of social
security contributions by workers and firms (yc).† In a pay-as-you-
go system such as the Italian one, whereby the largest part of
contributions are directly paid by firms, the latter choice is justified.
This would amount to define, as usual, yd as the net real disposable
income of the private sector, and to revert to a gross definition of
social security wealth (wsg=ws+wsc, where ws is net pension wealth
and wsc is the discounted stream of current and future con-
tributions).‡
Furthermore, if current social security transfers to households
are already accounted for in the pension wealth variable, they
should either be subtracted from disposable income to avoid
double counting or should enter the aggregate saving function
in a restricted form. Under a set of somewhat restrictive
assumptions, the aggregate equilibrium behaviour of consumers
would then translate into a simple saving rate function such as

s
yd
=u 0+u 1 g+u2 r+u3
wr wsg
yd
+h
A B
yd
+u 4
ys
yd
, (1)

where the u’s are (approximately) constant aggregate coefficients,


and the restriction u4=1−u0 should be satisfied.
Equation (1)—that is an equilibrium linear relationship
between the saving rate (s/yd), the rate of growth of population
and productivity (g), the real interest rate (r), the net non-
human (real and financial) wealth to income ratio (wr/yd), the
gross social security wealth to income ratio (wsg/yd) and the
social transfers to income ratio (ys/yd)—may be arrived at
expanding around constant values of r, g, (wr/yd) and (wsg/yd)
the extended aggregate life-cycle consumption function,§ with
yd standing then for the sum of aggregate net labour income

† In what follows, all consumption, income and wealth variables are in real
terms.
‡ The alternative specification, with disposable income defined as gross of
current social security contributions and pension wealth being net of their present
value counterpart, could also be tested. In the empirical analysis summarized in
Appendix 1, this has been done with such a specification always turning out to
be dominated by the one discussed in the text.
§ See Appendix A in the working paper version of this paper.
338 N. ROSSI AND I. VISCO

(yl), net income from capital (rwr) and net social transfers to
households (ys).†
A thorough statistical analysis, summarized in Appendix 1,
led to the error-correction representation of the data generating
process reported in Table 1,‡ where coefficient estimates, associated
standard errors and diagnostic statistics, when available, for both
least squares and instrumental variables estimates are also re-
ported. Comparison of LS and IV estimates suggests, however,
simultaneity bias to be negligible. In Table 1,§ ln(c/yd) is the
logarithm of the consumption to income ratio,¶ where c stands for
“economic” consumption and yd is net private disposable income,
adjusted for inflation;†† z stands for real government consumption;
and r is defined as ln[(1+R)/(1+p)], where R is the nominal
interest rate on long-term Government bonds and p is the expected
annual rate of change of consumer prices.‡‡
The error-correction estimates have a plausible economic in-
terpretation, and appear to possess acceptable statistical prop-
erties. Needless to say, to assess the relevance of Auerbach and
Kotlikoff’s (1983) assertion about the likely instability of aggregate

† An equilibrium condition equivalent to equation (1) (with u4=1−u0) might


also be written excluding the ys/yd term from the right-hand side of the equation
and replacing yd by y=yd−ys. Alternatively, one might still focus on the standard
definition of the private saving rate, leaving the definition of disposable income
unchanged, replacing wsg with wsw (the discounted stream of pension benefits for
current workers) and leaving u4 unconstrained. In this case ys would also catch
the effect on private saving due to wsr (equal to wsg−wsw, i.e. the discounted
stream of pensions for current retirees). The relative merits of the three alternative
(and, in principle, entirely equivalent) formulations are scrutinized in Appendix
1, where the reasons supporting the use of equation (1) are indicated.
‡ Apart from adequately representing the data process, the error-correction
representation may implicitly account for other phenomena likely to affect the
Italian saving rate, such as financial market imperfections. Liquidity constraints
may be caught by the short-term components of the error-correction equation
while borrowing constraints may affect the target wealth to income ratio and
thereby the long-run relationship (see the “Introduction” to Ando, Guiso & Visco,
1994).
§ All variables are reported, with sources and definitions, in a data appendix
to the working paper version of this article. As far as the social security wealth
variables are concerned, it should be noticed that for the year 1992 they were
constructed as weighted averages of pre-reform and post-reform figures, with
weights equal to 0·75 and 0·25, respectively, to account for the fact that, although
enacted by the end of the year, the 1992 reform of social security was publicly
announced by the end of the summer of 1992.
¶ Clearly, −ln(c/yd)≈s/yd. Notice that while the saving ratio is a bounded
variable (at least from above), ln(c/yd) is not.
†† The empirical exercise undertaken in this section, therefore, assumes that
while households pierce the corporate veil, they do not pierce the government
veil. Evidence on this assumption is presented in Rossi and Visco (1994: Appendix).
‡‡ Unweighted average of semi-annual survey estimates of expected inflation.
TABLE 1 Non-linear error-correction representation (sample: 1954–1993, least squares and instrumental variables
estimates)

D ln ct= 0·459D ln ytd+0·098D ln(ys/yd)t−1−0·476D ln zt−0·340D ln zt−1+0·275D ln ct−1


(0·042) (0·043) (0·130) (0·168) (0·081)

− 0·545[ln(c/yd)t−1+(0·288−0·033((wr/yd)t−1+0·658(wsg/yd)t−1)+0·321rt−1+(1–0·288)(ys/yd)t−1)]
(0·091) (0·041) (0·003) (0·091) (0·106) (0·041)

Note: LS estimates; in parentheses, White’s standard errors.


R̄2=0·866, dw=2·010, r̂=0·0084, serial correlation: F(1, 29)=0·095, functional form: F(1, 29)=1·237, normality:
v2(2)=0·952, heteroskedasticity: F(1, 38)=0·537, arch: F(1, 29)=0·607.

D ln ct= 0·538D ln ytd+0·080D ln(ys/yd)t−1−0·533D ln zt−0·426D ln zt−1+0·315D ln ct−1


(0·085) (0·047) (0·150) (0·187) (0·096)

− 0·567[ln(c/yd)t−1+(0·298−0·033((wr/yd)t−1+0·650(wsg/yd)t−1)+0·336rt−1+(1–0·298)(ys/yd)t−1)]
(0·104) (0·041) (0·004) (0·090) (0·102) (0·041)
NATIONAL SAVING AND SOCIAL SECURITY IN ITALY

Note: IV estimates; in parentheses, White’s standard errors.


2
R̄ =0·850, dw=2·066, r̂=0·0089, validity of instruments: v2(5)=5·081, serial correlation: v2(1)=0·159, functional
form: v2(1)=2·169, normality: v2(1)=0·522, heteroskedasticity: v2(1)=1·683.
d
Instruments: besides predetermined variables and D ln zt, the list of instruments includes ln yt−1, ln(c/yd)t−2,
r d sg d
(w /y )t−2, (w /y )t−2 and rt−2.
339
340 N. ROSSI AND I. VISCO

time-series relationships, we have carefully investigated the sta-


bility of the estimated model and inspected the sequence of one-
step ahead residuals. The equation standard errors corresponding
to the latter show little variation, in spite of the substantial changes
in the series under observation. Inspection of similar evidence for
the estimated coefficients based on the recursive estimator, has
provided further evidence of parameter constancy.
All long-run coefficients appear to be rather precisely determined
and the restriction u4=1−u0 turns out not to be rejected by the
data. The estimates indicate h to be about two-thirds, thereby
implying in equilibrium a non-negligible replacement of private
savings by social security.†
The present estimate of h is about twice as large as the cor-
responding IV estimate in Rossi and Visco’s (1994). The large and
significant difference originates entirely from the proper treatment
of the double counting issue and not from the information contained
in the 1991–1993 observations. The estimate of h presented in
Table 1 is, in fact, very robust to changes in specification (in both
the short-run and the long-run components of the non-linear error-
correction specification) and in the methodology of estimation (if
anything, when compared with the other estimates presented in
Appendix 1, those of Table 1 seem to be on the lower side).
Turning to the interpretation of recent trends in the Italian
private saving rates on the basis of the equilibrium parameters
derived from the IV estimates presented in Table 1 (the upper
panel of Table 1 makes it clear that looking at LS estimates would
not change the results appreciably), Table 2 presents in the first
row changes in the net private saving rate (that is s/yd) over the
1954–1992 period, leaving 1993 for later discussion. The second
row reports the fitted equilibrium values of the latter and the
following rows present its decomposition.‡
It should be underlined that the fitted equilibrium values do not
incorporate the short-run dynamic effects of the error-correction
specification. Therefore they are not directly comparable to the
actual changes of the net private saving rate. Moreover, it is crucial
to understand that, in the light of equation (1), the saving rate
can be affected by the pension system in two conceptually different
ways. On the one hand, pension income adds up to other income
components to determine private disposable income and hence the

† It should be noticed that h measures the extent by which private non-human


wealth is replaced by social security wealth conditional on given interest rate,
productivity growth and net saving rate. In the light of the “error-correcting”
nature of equation (1) (see the “Introduction” to Ando, Guiso & Visco, 1994), it
can be shown that in a steady state equilibrium the replacement rate between
pension and non-human wealth may be much lower than h.
‡ The stability of the short- and long-run terms of the error-correction rep-
resentation suggests that such in-sample simulation exercise is a legitimate one.
NATIONAL SAVING AND SOCIAL SECURITY IN ITALY 341

TABLE 2 Changes in the aggregate private saving rate and their


determinants (changes between period averages: 1954–1992)

1962–72 1973–79 1980–92 1980–92


1954–61 1962–72 1973–79 1962–72
Changes in actual saving rate:
D(s/yd) 0·049 −0·040 −0·051 −0·094
Changes in equilibrium saving
rate: D[(s/ŷd)] 0·031 −0·083 −0·048 −0·131
Contributions of:
Dg −0·015 −0·047 −0·015 −0·062
Dr −0·008 −0·010 0·020 0·010
D(wr/yd) 0·066 −0·028 −0·034 −0·062
D(wsg/yd) −0·034 −0·022 −0·035 −0·057
D(ys/yd) 0·022 0·024 0·016 0·040

Note: Changes in the aggregate private saving rate are computed from period
averages of ratios of actual saving figures to the net disposable income of the
private sector adjusted for inflation. The contribution of g is arrived at setting the
short-run components of the error-correction representation at their equilibrium
values and solving for the growth equilibrium rate.

observed saving rate. On the other hand, present and future


developments of the pension system influence present consumption
behaviour thereby affecting the saving rate.
As it can be seen from Table 2, the large increase of the private
saving rate observed between the 1950s and the 1960s turns out
to be mostly attributable to the declining non-human wealth to
income ratio which reflects the unprecedented growth experienced
by the postwar Italian economy until 1962. At the same time, the
negative contribution of social security wealth arising from the
appearance of new pay-as-you-go pension schemes is far from
negligible and averages to about three percentage points. In the
following time interval (between the 1960s and the years following
the first oil shock) rising housing prices coupled with the pro-
ductivity slowdown to induce an upward trend in the non-human
wealth to income ratio. The result, strengthened by a declining
real interest rate, is a three percentage points fall in the equilibrium
saving rate. In the same period, though, further structural changes
in the social security system began to feed into consumption
decisions thereby adding to the fall in the saving rate. In the
following period (ending in 1992), notwithstanding the rapid
growth of private financial assets (mainly public debt), the joint
contribution of the non-human wealth to income ratio and of
productivity growth somewhat tapers off and is partly coun-
terbalanced by the positive effect of the rising real interest rates,
with the “pension effect” accounting for much of the further fall
342 N. ROSSI AND I. VISCO

in the private saving rate. Finally, comparing the 1980s with the
1960s, slightly less than half of the entire fall of the private
equilibrium saving rate appears to be due to the increased gross
social security wealth to income ratio. The negative pension wealth
effect on saving in the period (some six percentage points) is
partially offset by the direct effect of pension expenditures on
disposable income (with a positive impact on the saving rate of
about four percentage points). To put it differently, it might be
said that, had the Italian households adhered to the standard life-
cycle framework (with h=0), the net private saving rate currently
observed would not be much different from the one prevailing in
the mid-1970s.

4. Household saving and social security wealth: cross-section


evidence

The foregoing time-series analysis has proposed that the rise in


the pension wealth to income ratio could have been responsible
for a substantial part of the decrease in the private saving rate
observed over the last quarter of a century. Recognizing that
aggregation and other sources of bias may affect the standard
aggregate consumption functions, in this section we supplement the
time-series evidence with the insights provided by the estimation of
a saving function on the microeconomic information derived from
the Bank of Italy 1991 Survey on Household Income and Wealth
(SHIW).†
The measurement of the net present value of social security
benefits and taxes is discussed in Appendix 2, where summary
information on the households included in the sample is also
provided. Measures of expected social security benefits were com-
puted under two sets of alternative assumptions. First, an attempt
was made to assign each worker to a specific social security fund
and to define consequently, on the basis of the relevant legislation,
age eligibility requirements, reference wages for computing pen-
sionable earnings, accrual rates and contribution requirements.
Second, reference was made to the subjective assessment of ex-
pected pension benefits and expected retirement age of active
population. Differently than Jappelli (1995), the latter alternative
was not confined to the subsample of respondents. To avoid sample
selection bias and account for endogeneity of households’ ex-
pectations about future benefits, separate reduced form equations
were estimated for the expected replacement rate and retirement
age and the results used to impute expectations to non-respondents.

† See Brandolini and Cannari (1994) for a full length description of the Bank
of Italy SHIW.
NATIONAL SAVING AND SOCIAL SECURITY IN ITALY 343

The empirical specification of the saving function replicates, at


the household level, equation (1):†

si/yin=w0+w1(wri/yin)+w2(wisg/yin), (2)

where real interest and productivity growth rates are assumed to


be constant across households, si/yin≈−ln(ci/yin), yin stands for nor-
mal age-adjusted net earnings of household i defined as the average
present discounted value of the stream of net future earnings, and
coefficients w0, w1 and w2 depend on a number of socio-demographic
variables including age of the household’s head, number of children,
education, location, and the like. In addition, regressors include
(the log of) normal earnings and dummies indicating the purchase
of life insurance or pension funds.‡ As in the case of the time-
series estimates, a gross definition of social security wealth is used.
Appendix 2 reports and comments on the cross-section data and
the parameter estimates based, alternatively, on the two measures
of social security wealth. Since both w1 and w2 are allowed to
vary with household-specific socio-demographic characteristics, the
cross-section results permit to derive household-specific estimates
of the degree of substitutability between fungible and pension
wealth, as measured by (h=w2/w1). The latter parameter, however,
only captures the (static) displacement effect of social security
wealth under the assumption of constant saving rates. The dis-
tributions of the h’s over the sample of about 5 600 households are
summarized in Table 3 in terms of their medians, means and
standard deviations.
Both measures of social security wealth convey the same mes-
sage: as of 1991, the static substitution effect of social security on
private accumulation tended to be large, and, on average, above
unity. However, both distributions tend to be skewed to the left,
with the overall median as low as 0·9, not far from the time-series
estimates.§

† It is worth noting that, contrary to the previous microeconometric literature,


we estimate a saving function and not a fungible assets equation.
‡ Normal earnings should not affect the saving ratio if preferences are homo-
thetic. It should be kept in mind, though, that other variables included in the
regression (such as the region of residence or dummies signalling contributions
to life insurance and pension funds) may also proxy for lifetime earnings.
§ The extent to which social security wealth substitutes for private wealth
might seem much larger than that obtained by Brugiavini (1987) and Jappelli
(1995), where estimation of a fungible assets equation yields an estimated offset
between one tenth and one fifth. It is worth stressing, however, that the difference
between those estimates and the present ones is entirely due to the reference
theoretical framework, in that estimation of a fungible asset equation (as opposed
to the estimation of a saving function) would yield results comparable to those
reached by the previous authors. This is not surprising since, as observed, the
two estimation strategies refer to two different notions of equilibrium.
344 N. ROSSI AND I. VISCO

TABLE 3 Distribution of h: summary information (1991 SHIW)

Social security wealth based on

legislation expectations

median mean S.D. median mean S.D.

Overall 0·933 1·052 0·521 1·172 1·275 0·547


By age of head:
less than 34 0·737 0·777 0·237 0·918 0·946 0·258
between 34 and 44 0·720 0·757 0·295 1·150 1·142 0·396
between 44 and 54 1·169 1·220 0·488 1·379 1·401 0·536
between 54 and 64 1·249 1·289 0·524 1·417 1·404 0·555
By region:
North and Central Italy 1·133 1·245 0·540 1·381 1·438 0·588
Southern Italy 0·688 0·734 0·276 0·989 1·008 0·327
By residence:
Rural areas and towns 1·009 1·131 0·521 1·236 1·357 0·545
Cities 0·617 0·643 0·263 0·828 0·858 0·324
By household size:
3 or less 1·004 1·109 0·497 1·174 1·290 0·569
more than 3 0·885 0·998 0·537 1·172 1·262 0·525
By number of income recipients:
1 0·849 0·927 0·437 1·215 1·330 0·504
more than one 1·009 1·134 0·554 1·113 1·240 0·571
By number of children:
2 or less 0·935 1·062 0·520 1·172 1·281 0·556
more than 2 0·913 0·990 0·522 1·212 1·242 0·491
No. of observations 5587 5587

Notes: (i) estimates of h are based on the cross-section estimates reported in


Table 7, (ii) the estimation sample excludes households with pension income only;
moreover, in computing the present table, households showing negative values
of h (for a total of 41 observations) have been excluded.

As Table 3 indicates, as of 1991, the effect of pension wealth on


saving tended to exceed that of non-human wealth in the case of
older households, while for younger households the opposite result
appears to hold. This outcome possibly indicates the perception
among the young of the unsustainability of the Italian social
security system. Interestingly, both the old and the young house-
holds may have proved to be right in that the 1992 reform left the
former group basically unaffected, severely hitting the latter. The
age effect is apparently the predominant one, the differences
between estimated h’s in different regions of the country being
largely determined by their different age compositions.
NATIONAL SAVING AND SOCIAL SECURITY IN ITALY 345

5. 1993 and beyond

The central tenet of this paper is that the changes in social security
laws and regulations which took place in the late 1960s and
early 1970s severely weakened the link between contributions and
benefits permitting a time path of aggregate consumption in excess
of what would have occurred in the absence of such changes. It is
worth recalling that those changes entailed both a replacement
effect due to the introduction of an unfunded social security system
for specific segments of the population as well as, for a given-
institutional structure, a lifetime wealth increment.
The evidence emerging from aggregate time series as well as
from cross-section data provides a rather clear-cut outcome, in-
dicating that a large part of the decline of private saving (in terms
of net private disposable income) observed since the 1960s may be
attributed to the rise in the social security wealth to income ratio.
Given the specification adopted in this work, a major role is played
by the productivity slowdown.
Past trends have been, however, substantially reversed in the
last couple of years. A major reform was enacted in 1992 and a
further attempt to halt social security expenditure has taken place
in the spring of 1995. As Figure B shows, the 1992 reform may
have determined a substantial fall (as large as 20%) in the gross
social security wealth (a gigantic “capital levy” of about 2000
trillions lire), leading the gross pension wealth to net national
disposable income ratio to fall to 4·6 in 1993 from 5·9 in 1992. This
outcome appears to be confirmed by the information available at the
individual level: preliminary computation of expected household
social security wealth for 1991 and 1993 suggests an average fall
of about 10% (the difference being given by postponed retirement).
To the extent that the overall reform of the Italian pension
system just approved by Parliament will confirm the implications
of the 1992 measures, on the basis of the parameter estimates of
Table 1 it might be inferred that such dramatic changes in the
social security arrangements could bring about a rise of the private
saving ratio of up to 3 percentage points in the steady state, with
figures lower than 3% accounting for the fact that the saving
augmenting replacement effect may be counterbalanced by a saving
reducing retirement effect.

Acknowledgements

We would like to thank Albert Ando, Orazio Attanasio, Andrea


Brandolini, Agar Brugiavini, Carlo Favero, Luigi Guiso, Tullio
Jappelli, Alberto Locarno, Roberto Tedeschi and one anonymous
346 N. ROSSI AND I. VISCO

referee for their most helpful comments and criticisms, Luca Bel-
trametti for providing his social security wealth data, Massimo
Rostagno for helpful discussions on the same topic and Alessandro
Venturini for excellent research assistance. The usual disclaimer
applies. The views expressed herein are those of the authors and
do not necessarily reflect those of the Bank of Italy. Financial
support from the Italian National Research Council is gratefully
acknowledged.

References
Ando, A., Guiso, L. & Terlizzese, D. (1994). Dissaving by the elderly, transfer
motives and liquidity constraints. In A. Ando, L. Guiso & I. Visco, Eds.
Saving and the Accumulation of Wealth, pp. 188–213. Cambridge: Cambridge
University Press.
Ando, A., Guiso, L. & Visco, I. Eds. (1994). Saving and the Accumulation of
Wealth. Cambridge: Cambridge University Press.
Ando, A. & Modigliani, F. (1963). The “life cycle” hypothesis of saving: aggregate
implications and tests. American Economic Review, 53, 55–84.
Auerbach, A.J. & Kotlikoff, L.J. (1983). An examination of empirical tests of
social security and saving. In E. Helpman, A. Razin & E. Sadka, Eds. Social
Policy Evaluation: An Economic Perspective, pp. 161–179. New York: Academic
Press.
Barro, R.J. (1974). Are government bonds net wealth? Journal of Political Eco-
nomy, 82, 1095–1117.
Beltrametti, L. (1995). On pension liabilities in Italy. Ricerche Economiche, 49,
405–428.
Brandolini, A. & Cannari, L. (1994). Methodological appendix: the Bank of Italy’s
survey of household income and wealth. In A. Ando, L. Guiso & I. Visco, Eds.
Saving and the Accumulation of Wealth, pp. 369–386. Cambridge: Cambridge
University Press.
Brugiavini, A. (1991). Effetti delle pensioni sul risparmio nella teoria del ciclo
vitale. In T. Jappelli, Ed. Bilancio pubblico e risparmio privato, pp. 43–68.
Milano: F. Angeli.
Cannari, L., D’Alessio, G., Raimondi, G. & Rinaldi, A.I. (1990). Le attività
finanziarie delle famiglie italiane. Temi di Discussione No. 172. Roma: Banca
d’Italia.
Doornick, J.A. & Hendry, D.F. (1994). PcFiml 8.0. London: International Thomson
Publishing.
Engle, R.F. & Granger, C.W. (1987). Co-integration and error correction rep-
resentation, estimation and testing. Econometrica, 55, 255–276.
Feldstein, M. (1974). Social security, induced retirement, and aggregate ac-
cumulation. Journal of Political Economy, 82, 905–926.
Feldstein, M. (1976). Social security and saving: the extended life cycle theory.
American Economic Review, 66, 77–87.
Fornero, E. & Castellino, O. Eds. (1990). Formazione e impiego della ricchezza
delle famiglie. Torino: Fondazione Giorgio Rota.
Franco, D. (1992). Alcune note sull’espansione della spesa pubblica in Italia
(1960–1990). In Ente Einaudi, Il Disavanzo Pubblico in Italia: Natura Strut-
turale e Politiche di Rientro, pp. 211–355. Bologna: Il Mulino.
Franco, D. (1995). Pension liabilities. Their use and misuse in the assessment of
fiscal policies. Economic Papers, No. 110. Brussels: European Commission.
Franco, D. & Morcaldo, G. (1990). The Italian pension system: development and
NATIONAL SAVING AND SOCIAL SECURITY IN ITALY 347

effects on income distribution and poverty. Labour, 4, 105–124 and 133–159.


Franco, D., Gokhale, J., Guiso, L., Kotlikoff, L.J. & Sartor, N. (1994). Generational
accounting. The case of Italy. In A. Ando, L. Guiso & I. Visco, Eds. Saving and
the Accumulation of Wealth, pp. 128–160. Cambridge: Cambridge University
Press.
Hansen, H. & Johansen, S. (1993). Recursive estimation of cointegrated VAR-
models. Preprint No. 1. Copenhagen: Institute of Mathematics and Statistics.
INPS (1993). Le pensioni domani. Bologna: Il Mulino.
Jappelli, T., Ed. (1991). Bilancio pubblico e risparmio privato. Milano, F. Angeli.
Jappelli, T. (1995). Does social security reduce wealth accumulation? Evidence
from Italian survey data. Ricerche Economiche 49, 1–31.
Johansen, S. (1988). Statistical analysis of cointegrating vectors. Journal of
Economic Dynamics and Control, 12, 231–254.
Modigliani, F. (1986). Life cycle, individual thrift and the wealth of nations.
American Economic Review, 76, 297–313.
Modigliani, F. & Jappelli, T. (1993). Why has the Italian national saving rate
declined? In S. Biasco, A. Roncaglia & M. Salvati, Eds. Market and Institutions
in Economic Development, pp. 117–137. London: MacMillan.
Modigliani, F. & Sterling, A. (1983). Determinants of private saving with special
reference to the role of social security. Cross-country tests. In F. Modigliani &
R. Hemming, Eds. The Determinants of National Saving and Wealth, pp. 24–55.
London: MacMillan.
Pesaran, M.H. & Pesaran, B. (1991). MICROFIT 3.0. An Interactive Econometric
Software Package. Oxford: Oxford University Press.
Rossi, N. & Visco, I. (1994). Private saving and government deficits. In A. Ando,
L. Guiso & I. Visco, Eds. Saving and the Accumulation of Wealth, pp. 70–105.
Cambridge: Cambridge University Press.
Samwick, A. (1994). The limited offset between pension wealth and other private
wealth: implications of buffer stock saving (unpublished).
van den Noord, P. & Herd, R. (1993). Pension liabilities in the seven major
economies. Working Paper No. 142. Paris: OECD.
Williamson, S.H. & Jones, W.J. (1983). Computing the impact of social security
using the life cycle consumption function. American Economic Review, 73,
1036–1052.

Appendix 1: Time-series estimates

As in Rossi and Visco (1994) an interesting starting point for a


time-series analysis of the determinants of the private saving rate
might be the equilibrium specification

−ln(c/yd)≈s/yd=u0+u1g+u2r+u3
A B
wr ws
yd
+h d
y
(1.1)

where the variables are defined as in Section 3 of the text. In Rossi


& Visco (1994) yd and ws, defined as usual, included (the former
as the current flow, the latter as the current and future discounted
flows) the pensions of current retirees (ys) and (with a negative
sign) the social security contributions (yc) paid in front of the
348 N. ROSSI AND I. VISCO

pension benefits of individuals currently at work.† In what follows,


we benefit from the availability of the different components of ws
(from Beltrametti, 1995) and introduce an estimate of the net of
tax social transfers component of disposable income, so that ws=
wsw+wsr−wsc and yd=y+ys (disregarding net government transfers
to firms which are assumed to have been already netted out from
all income variables) where wsw is the discounted stream of pension
benefits for current workers, wsr is the discounted stream of pen-
sions for current retirees, wsc is the discounted stream of social
security contributions by current workers and we recall that ys is
the (net of tax) pension income flow. In equation (1.1), then, one
should replace yd with y and ws with wsg=wsr+wsw=ws+wsc. Al-
ternatively, disposable income might be defined as y+yc and pen-
sion wealth as ws. The former case, on which the following estimates
are based, seems more appropriate for a pay-as-you-go system like
the Italian one.‡
It should be noticed, however, that wsr (as wsw) is constructed
under a set of specific assumptions which purport to mimic the
formation of expectations by current retirees on their future net
pensions. One could then include in the equilibrium relation only
the component of social security wealth that relates to current
workers (for which no substitute to wsw is currently available), and
use ys as a proxy for the expected stream of current pensions and
future benefits to the current retirees.§ We might then assume
two alternative equilibrium specifications, that is

−ln(c/y)=u0+u1g+u2r+u3
A B
wr wsg
y
+h
y
(1.2)

and

−ln(c/yd)=u0+u1g+u2r+u3
A B AB
wr wsw
yd
+h
yd
+u4
ys
yd
(1.3)

† In this paper, as in Rossi and Visco (1994), the income flows are net of taxes
while the present values of current and future discounted flows are not.
‡ For a formal test of the alternative specification, see footnote † on p. 23.
§ Being aware of the double counting issue and to evaluate the robustness of
our results, in our 1994 paper we also considered a specification such as (1.1)
with ws replaced by wsw (but without allowing ys to enter the regression with a
different coefficient than y), to obtain results very close to those eventually
reported in that paper, with, if anything, a slightly stronger effect of social
security wealth on private consumption. Moreover, in conducting a simulation
experiment on microeconomic data we fixed h at 0·5 (as opposed to an estimate
of 0·3 in the time series model) to account, again, for double counting.
NATIONAL SAVING AND SOCIAL SECURITY IN ITALY 349

where u4 is obviously a function of the wsr/ys ratio. In principle,


proper representations of such long-run equilibrium relations
might validly be estimated. However, it should be noticed that the
ratio of wsr to ys increases from about 3 in the early 1950s to over
11 in the early 1990s. It then follows that relying on ys as a proxy
of wsr, if may allow to consider a different expectations hypothesis,
might induce substantial bias in the regression estimates to ac-
commodate that rising trend. On these grounds we shall therefore
disregard equation (1.3) in what follows.
However, it is easy to show that equation (1.2) may be rewritten
as

−ln(c/yd)=u0+u1g+u2r+u3
A B AB
wr wsg
yd +h d +u4
y
ys
yd
(1.4)

with u4=1−u0.
If the above constraint is satisfied, equations (1.2) and (1.4) may
be considered equivalent representations of consumer behaviour.
In what follows we shall first focus on the somewhat simpler
equation (1.2) and proceed, subsequently, to test the restriction
u4=1−u0.
Recent developments in the theory of cointegration have allowed
to explicitly link economic equilibrium relationships between a set
of variables such as those depicted by equation (1.2) with statistical
models of those variables (Engle & Granger, 1987). In testing for
the existence of cointegration, it is necessary to establish first that
the individual series are I(1) and then that there exists some non
trivial function of them which is I(0). The most common approach
is to start testing for a unit root in the univariate representations
of the individual series and then in the least squares linear
combination. Dickey–Fuller (DF) and augmented Dickey–Fuller
(ADF) statistics have been computed for the following variables:†
(i) the log of the consumption to income ratio, ln(c/y), (ii) the
private wealth to income ratio, wr/y, (iii) the gross social security
wealth to income ratio, wsg/y, (iv) the real interest rate, r.
In the 1951–1993 sample, the hypothesis of a unit root at the
5% level could never be rejected for the variables listed above.‡
In a multivariate context, though, there may be a number of
cointegrating regressions linking the saving rate, the real interest
rate and the wealth to income ratios. We therefore revert to the

† All estimates reported in this Appendix have been obtained using Microfit
3.0 (Version 386, see Pesaran and Pesaran, 1991) or Pcfiml (Version 8.0, see
Doornick and Hendry, 1994), where specific references are reported.
‡ The hypothesis of a unit root could, instead, be rejected for productivity and
population growth.
350 N. ROSSI AND I. VISCO

TABLE 4 Maximum likelihood cointegration analysis (equation 1.2,


1954–1993)

Lag length in VAR (in years): 2; trended variables, no trend in DGP;


exogenous variable: D ln(zt)

Non-zero eigenvalues of the stochastic matrix:


0·655, 0·485, 0·298, 0·169, 0·102

W test (H0: vΖ4; H1: v=5) 4·289


W test (H0: vΖ3; H1: v=4) 7·399
W test (H0: vΖ2; H1: v=3) 14·61
W test (H0: v=1; H1: v=2) 26·519
W test (H0: v=0; H1: v=1) 42·585∗

Cointegration vector associated with the largest eigenvalue:

ln c ln y r wr/y wsg/y
−1·000 0·966∗ −0·155 0·022∗ 0·023∗

Homogeneity restriction: v2(1)=1·631

∗ significant at the 95% level.

recent work of Johansen (1988), testing for the maximum number


of cointegration vectors (v) and providing a maximum likelihood
estimate of such vector(s) (Table 4). In designing the vector auto-
regression underlying the cointegration analysis it should be care-
fully noticed that focusing on the ln(c/y) variable entails unduly
restrictive constraints on the short-run response of consumption
to income. It seems wiser to let ln(c) and ln(y) enter the VAR
separately and then test for long-run homogeneity of consumption
to income. As in Rossi and Visco (1994), government expenditure
is also allowed to play a role and hence D ln(zt) is allowed to enter
the VAR as an exogenous variable.† As it turns out, the hypothesis
of a single cointegration vector cannot be rejected at customary
confidence levels.‡ Moreover, the hypothesis of long-run homo-
geneity of consumption and income is not rejected by the data.
† Integration tests for this additional variable show it to be I(0).
‡ Statistically, it would be possible to further restrict the VAR to lag one. We
prefer though a slightly overparameterized system and we stress that the (single)
cointegrating vector implied by a VAR(1) turns out to be almost identical to those
above. It may be interesting to notice that in our previous paper we opted for a
VAR(4), imposing however short-run homogeneity between consumption and
income. In that case the overparameterized VAR apparently accounted for the
misspecification induced by the short-run homogeneity restrictions, delivering a
single cointegration vector rather similar to the one shown in Table 4. Overlooking
this point and simultaneously restricting the VAR’s length and the short-run
consumption to income response ends up misspecifying the cointegrating vector.
NATIONAL SAVING AND SOCIAL SECURITY IN ITALY 351

Finally, the stability of the cointegration space was explored,


carefully inspecting the constancy of the cointegration rank (see
Hansen & Johansen, 1993). In the two cases of a variable or a
constant short-term dynamics, the cointegration rank was found
to be remarkably stable.
In a relatively small sample, like the present one, an attractive
alternative to the multivariate time-series methods used above is
the estimation of a conditional error-correction model and the
formal test of the significance of the adjustment coefficient. As-
suming, therefore, on the basis of the results reported in Table 4,
that the variables entering the long-run equilibrium relations are
cointegrated, Table 5 presents the estimated linear error-correction
models which have the general form of the theoretical model but
contain additional lags and variables designed to shed light on the
short-run impact of income and government expenditure as well as
to take into account the possibility of some agents being constrained
from their optimal consumption plan in any period thereby showing
excess sensitivity of consumption to current income. In particular,
our specification search allowed for a rich short-run dynamics
which included, as in Rossi & Visco (1994), the lagged value of the
dependent variable, D ln c, and the first differences of ln z and of
ln yd. To allow for the possibility of different short-run responses
with respect to y and ys, besides the (current and lagged) first
differences of ln(yd), those of ln(ys/yd) were introduced with separate
coefficients. The possible presence of current and lagged first
differences of the real interest rate and of the rates of change of
the wealth variables (in real terms) was also investigated. Beside
the error-correction representation of equation (1.2), Table 5 also
reports the estimates obtained with the saving rate specification
(in terms of the standard definition of disposable income) of equa-
tion (1.4). Both the unrestricted and restricted (u4=1−u0) es-
timates are presented.
In Table 5† besides the short-term effects coming from
changes in y, ys and z, the saving rate adjusts to its equilibrium
level at a rate of about 0·5 per year; in the long-run equilibrium
the saving rate is significantly influenced by the real rate of
interest, the substitution effect (−u2) being about 0·3 (that is the
negative of the coefficient of r, as estimated in Table 5, divided
by the adjustment coefficient). In the long-run equilibrium the

† In Table 5 White’s standard errors are reported under the coefficient es-
timates. The mis-specification tests statistics are distributed as follows: serial
correlation F(1, n−k−1), functional form F(1, n−k−1), normality v2(2), het-
eroskedasticity F(1, n−2), predictive failure F(n1, n−k), where n1 is the number
of observations in the shorter time period (1991–1993), additional regressors F(8,
n−k−8).
352 N. ROSSI AND I. VISCO

TABLE 5 Linear error-correction representation (dependent variable: D ln ct ;


sample: 1954–1993)

Equation (1.4) Equation (1.4)


Equation (1.2) (unrestricted) (restricted)

Short-run terms
D ln(yd)t 0·456 (0·044) 0·457 (0·045) 0·459 (0·042)
D ln(ys/yd)t−1 0·093 (0·041) 0·093 (0·044) 0·098 (0·043)
D ln zt −0·482 (0·126) −0·507 (0·142) −0·476 (0·130)
D ln zt−1 −0·332 (0·168) −0·358 (0·174) −0·340 (0·168)
D ln ct−1 0·271 (0·078) 0·253 (0·083) 0·275 (0·081)
Error-correction term
ln(c/y)t−1 −0·538 (0·088)
ln(c/yd)t−1 −0·516 (0·102) −0·545 (0·091)
Long-run terms
intercept −0·126 (0·031) −0·128 (0·057) −0·157 (0·039)
rt−1 −0·198 (0·066) −0·184 (0·064) −0·175 (0·063)
(wr/y)t−1 0·015 (0·003)
(wr/yd)t−1 0·016 (0·005) 0·018 (0·004)
(wsg/y)t−1 0·010 (0·002)
(wsg/yd)t−1 0·014 (0·004) 0·012 (0·002)
(ys/yd)t−1 −0·546 (0·209) −0·388 (0·063)
Misspecification tests
R̄2 0·866 0·864 0·866
dw 2·030 2·079 2·010
r̂100 0·836 0·845 0·838
Serial correlation 0·147 0·273 0·095
Functional form 1·387 1·135 1·237
Normality 1·017 1·117 0·952
Heteroskedasticity 0·843 1·449 0·537
Predictive failure 0·911 0·729 0·894
Additional regressors 0·485 0·489 0·337
No. of observations (n) 40 40 40
No. of regressors (k) 10 11 10

Note: Additional regressors: D ln(yd)t−1, D ln(ys/yd)t, Drt, Drt−1, D ln wtr, D ln wt−1


r
,
D ln wsg
t , D ln wt−1.
sg

propensity to consume out of non-human wealth (u3) is about 0·03


and the coefficient of wsg is between 65 and 90% that of wr. Finally,
the estimates appear to be remarkably stable, even when the very
turbulent 1991–1993 period is included in the sample.†
† It should be observed that if we test for the equality of the coefficients of wsw
and wsr by adding (wsr/y)t−1 to the regressors of equation (1.2) we cannot reject
the null hypothesis (with a t-statistic equal to 0·938). Similarly, when we add
(wsc/y)t−1 and (yc/y)t−1 to the regressors, to check for a possible bias in considering
disposable income net of current contributions (rather than social security wealth
net of the stream of current and future contributions) we cannot reject the
assumption of no bias (with an F(2, 28) statistic of 1·343).
NATIONAL SAVING AND SOCIAL SECURITY IN ITALY 353

TABLE 6 Estimates of h

Equation (1.4) Equation (1.4)


Equation (1.2) (unrestricted) (restricted)

(1) 0·705 (0·095) (2) 0·871 (0·337) (3) 0·658 (0·087)


(4) 0·599 (0·099) (5) 1·149 (0·538) (6) 0·576 (0·104)
(7) 0·922 (0·125) (8) 0·952 (0·345) (9) 0·867 (0·114)

Note: in parentheses, White’s standard errors. Equations (1), (2) and (3) as in
Table 5. Equations (4), (5) and (6) as in Table 5 with additional regressors as
indicated. Equations (7), (8) and (9) as (1), (2) and (3), respectively, with wsg
t−1
s
replaced by wt−1 .

The evidence from the ECM estimates then confirms that stem-
ming from the analysis of cointegration. Table 6 presents a sum-
mary of the estimates of h in equations (1.2) and (1.4) as can
be obtained by the estimated coefficients presented in Table 5.
Estimates of the respective standard errors are also provided. It
is clear that this parameter appears to be sizeable and, as already
observed, not far from one. To further check the robustness of this
result a few attempts have been made, as shown in Table 6, to
check whether restricting to zero the coefficient of wsc would
dramatically change the picture. Again we obtain very similar
results, with, if anything, estimates of h even closer to one.†
Since model (1.4) can be derived from model (1.2) if the restriction
u4=1−u0 is satisfied and since such restriction turns out not to
be rejected by the data (F(1, 29)=0·555) it seemed safe to use in
the text the specification of column (3) of Table 5, which is written
in terms of the standard saving to disposable income ratio. The
corresponding (non-linear) restricted least-squares and in-
strumental variables estimates are presented in Table 1 of the
text.

† These estimates should be compared with those obtained in Rossi and


Visco (1994), where the long-run equilibrium relation (1.1) was considered, and
estimates of h between one fifth and one third were obtained. New estimates of
the same specification (with the revised time series and the extended sample
considered in this paper) come out very close to the original ones, with h turning
out to be about one fourth. Most strikingly, however, adding the pension to income
ratio, the estimate of h rises dramatically, again being in the neighbourhood of
one. The pension to income ratio provides then a correction to the double counting
(and subsequent constraint) implicit in the definitions of disposable income and
social security wealth considered in Rossi and Visco (1994).
354 N. ROSSI AND I. VISCO

Appendix 2: Cross-section estimates

The 1991 and 1993 SHIW provide data on consumption, income,


wealth and socio-demographic characteristics from two samples of
households representative of the Italian population. The 1991
and 1993 SHIW also report specific information on retirement
behaviour, providing data on ex post and ex ante retirement age of
current workers and (for 1991 only) on ex post and ex ante re-
placement rates. In order to evaluate the empirical exercise re-
ported in Section 4, some data definitions and manipulations are
worth recalling.
Both 1991 and 1993 SHIW do not provide information on the
imputed flow of the consumption of durables. Consumption data
therefore refer to consumption of non-durables and services. The
normal earnings variable is supposed to purge current earnings
from transitory earnings and age effects. In estimating adjusted
normal earnings, defined as the annualized value of the recipient’s
human wealth, the procedure suggested by Ando, Guiso and Ter-
lizzese (1994:208–209) has been followed. The household net worth
is given by the sum of its net financial and real assets. The former
are adjusted for non- and under-reporting as in Cannari et al.
(1990).
Gross expected social security wealth is defined as in Jappelli
(1995), as the stream of social security benefits of a person of a
given age, expecting to retire at a given point in time and to die,
with a given probability, anytime in the future. However, contrary
to Jappelli we account for survivors’ benefits. Since the future
stream of social security benefits may be conventionally defined
as a percentage of the agent’s expected earnings at retirement,
expected pension wealth crucially depends on the expected re-
placement rate and on the expected retirement rate.
Information on both quantities are provided by the 1991 SHIW
as answers to the following questions: (i) “Consider the moment
when you will retire. Setting your final monthly income before
retirement equal to 100, what do you expect your first monthly
pension to be?”, and (ii) “When do you expect to retire?”. Answers
to the above questions are available only for 4400 and 8500 labour
income recipients, respectively. Contrary to Jappelli (1995), to avoid
sample selection biases arising from a non random distribution
of non-responses in the population, reduced form equations for
expected replacement rate and expected retirement age were es-
timated on the respondents subsample and fitted values imputed
to the overall sample of labour income recipients, thereby implicitly
accounting for potential endogeneity of the social security wealth
variable arising from endogeneity of expected retirement age and
replacement rate.
In particular, expected replacement rates at the individual level
NATIONAL SAVING AND SOCIAL SECURITY IN ITALY 355

turned out to be affected significantly by variables defining the


social security rules applying to the specific individual such as sex,
number of years worked (proxying for the number of years of
contributions), occupation and sector of occupation, as well as their
interaction with other variables at the individual level such as
education. Expected retirement age, instead, proved to be strongly
dependent, along with the individual history of the agent, on his
social environment (location of the household, number of children
and income recipients in the household, and the like). Interestingly,
the expected retirement age turned out to depend, rather clearly,
on the expected replacement rate with a five percentage point
decrease of the pension to earnings ratio leading to a one-year
shift forward of the expected retirement age.
Alternatively, to check the previous computations and cross-
check the time-series aggregate estimates discussed in the main
text, gross social security wealth was computed assigning each
worker to a specific social security fund and defining consequently
age eligibility requirements, reference wage and period for com-
puting pensionable earnings, accrual rates and contribution re-
quirements. The task was made rather difficult by the complexity
of the Italian social security system, by the presence of a large
number of social security funds for various categories of employees
and self-employed and by the uncontrolled evolution of the le-
gislation in the last decades. It proved necessary to revert to a
number of simplifying assumptions of which the most important
ones related to the timing of retirement, assumed equal to the
earliest possible one.
For the 1991 sample household social security wealth averaged
at about 310 and 340 million lire (for the measures based on
expectations and legislation, respectively), a level largely com-
parable with Beltrametti’s (1995) aggregate time-series figures
used in Section 3. The difference between the two measures of
social security wealth was mostly concentrated in those occupations
and sectors where early retirement is allowed but is apparently
not accounted for in agents’ expectations. In particular, the simple
correlation coefficient between the two quantities is about 0·95 in
the 1991 sample.
Table 7 presents a summary of the estimates based on equation
(2) in the text. Since both intercept and slopes of equation (2) shift
with socio-demographic characteristics, estimated coefficients refer
to a specific household as described in the notes to Table 7.
356 N. ROSSI AND I. VISCO

TABLE 7 Cross-section estimates (1991 SHIW, least squares estimates,


dependent variable: negative of log of non-durables consumption to normal
net earnings ratio)

Social security wealth based on

Legislation Expectations

wr/yn −0·0190 (0·0030) −0·0174 (0·0030)


wsg/yn −0·0163 (0·0011) −0·0174 (0·0012)
ln(yn) 0·3366 (0·0140) 0·3532 (0·0158)
Constant −2·9847 (0·1474) −3·1602 (0·1717)
Purchase of life insurance
(×100) −0·0024 (0·0005) −0·0023 (0·0005)
Contributions to pension
funds (×100) −0·0030 (0·0006) −0·0029 (0·0006)
Plus socio-demographic
dummies and interaction
terms
No. of observations 5628 5628
R2 0·5360 0·5440

Notes: (i) in parentheses, White’s standard errors, (ii) the sample excludes
households with pension income only, (iii) the reported estimates and standard
errors refer to a household living in a small town in Northern Italy, formed by
two adults, both income recipients, and two children, and with the head of the
household in his late twenties or early thirties.

You might also like