Consumption Structure, welfare goods
and retirement income:
linking the ageing puzzles
Najat El Mekkaoui de Freitas
(Laboratoire d’économie de Dauphine, Université Paris-Dauphine) Joaquim Oliveira Martins
(OECD, Université Paris-Dauphine)
Some empirical facts on consumption, pension and
saving do not fit with theory:
¾ First puzzle: ageing/consumption puzzle : why
consumption decrease in old age ?
¾ Second puzzle: ageing/saving puzzle : Why
individuals continue to save at old age, in particular in countries with generous PAYG and health care systems (welfare goods)?
¾ Third puzzle: saving/capitalisation puzzle : why
countries with fully funded systems have the lowest private saving rates ?
Structure of the presentation
1. Brief discussion of the three puzzles
2. A theoretical life cycle model where we consider
the main determinants of savings, inc. welfare goods, social transfers and pension replacement rates
3. The three puzzles are interconnected. Part of the
explanation relates to the interactions among the change in the structure of consumption with age, provision of welfare goods and retirement income
4. An econometric analysis of the impact of the age
structure, welfare systems on saving patterns across OECD countries.
1st Puzzle: The Ageing/Consumption Puzzle
Why consumption is not smoothed throughout the life cycle ?
80.0 100.0 120.0 140.0 160.0 180.0 200.0 220.0 under 25 25-34 35-44 45-54 55-64 65-74 75 + C ons um pt ion of t h e un de r 2 5 a g e gr o up = 100
United States European countries
40.0 60.0 80.0 100.0 120.0 140.0 160.0 180.0 under 30 30-44 45-59 60 + C ons um pt ion of t h e unde r 3 0 a g e gr oup (E U avera g e ) = 100 Fin Swe Deu Gbr Ita Fra
There may be cohort and time effects biasing the relationship, but some studies based on individual data suggest that these bias are relatively small
2nd Puzzle: The Ageing/Saving Puzzle
3rd Puzzle: The Saving/Capitalisation puzzle Why fully-funded systems have a lower
household saving rate?
Average Household Saving rate
0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 18.0 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Low Funding High Funding PAYG systems Funded systems
Some interpretations of ageing/consumption puzzle:
Bernheim, Skinner and Weinberg (2001): Lack of
forward-looking optimising behaviour to determine retirement saving;
Hurd and Rohwedder (2006): poor health could
explain the decline in spending;
Miniaci, Monfardini and Weber (2003): both shocks
to retirement wealth and leisure substitution;
Sarah Smith (2007): retirement is assumed to be
involuntary, largely reflecting health problems and redundancy.
Some interpretations of the ageing/saving puzzle:
Bloom et al. (2003) and Sheshinski (2004) : The higher life
expectancy may increase the need for additional precautionary savings.
Bernheim et al. (2001) : household behavior is not governed
by rational, far-sighted optimization.
Moore and Mitchell (1998) : households are not preparing
adequately their retirement period.
household would need to save 20 per cent of annual
earnings to have a replacement rate=61%.
A single woman would need to save around 32% of her
income to have a replacement ratio of 54% at age of 62
the majority of older households will not be able to
maintain current levels of consumption into retirement without additional saving.
Some results on capitalization & savings
Feldstein (1974) highlighted a negative link between PAYG
pension systems and household savings.
Baillu and Reisen (1997): analyse the effect of funded
systems on saving (public and private) using a panel of countries 1982-93 period. They found a positive and
statistically significant impact of pension funds on saving
Lopez Murphy and R. Musalem (2004): considering 43
countries for the period 1960-2002, the authors found that mandatory contribution to funded pension systems increase national saving
Bosworth and Burtless (2004): analyse the same effect for 11
countries and during the period 1971-00 and find an impact on private saving not significant econometrically
2. What determines savings ? A simple model
In the first period, each agent splits her disposable wages
into consumption (Ci) and saving (Si):
Where
α
is the rate of social contributions. In the second period, only a welfare good consumption
(e.g. health) is considered Hi.
To finance this consumption, the agent receives a PAYG
pension with a replacement rate
β
, the accumulatedsaving accrued by the return on capital r and a given amount of social transfers T:
i i i i w r S T H =
β
. + (1 + ) ⋅ + i i i S w C + = (1−α). Maximising the utility of each agent under the budget constraint, we obtain :
(
)
[
]
r w r T w r H C t s H u p C u H C u E Max i i i i i i i i i + ⋅ + + + ⋅ − ≤ + + ⋅ + = 1 1 ) 1 ( 1 . . ) ( ) ( , β αWhere pi is the survival probability at the beginning of
and the saving rate:
Expected signs: (+) pi, r (-) T, β, α[
i i]
i i i i i i i w T w r p p H r w r T w p C . ) 1 ( ) 1 ( 1 1 . 1 ) 1 ( 1 1 β α β α + + ⋅ + ⋅ − ⋅ + = ⎥⎦ ⎤ ⎢⎣ ⎡ + + + + ⋅ − ⋅ + = ⎥ ⎦ ⎤ ⎢ ⎣ ⎡ + + ⋅ + − + ⋅ − = α β i i i i i i w T r p p p w S ) 1 ( 1 1 1 1 ) 1 (3. Proposed explanations in this paper
The three puzzles are linked together
The interaction between consumption structure,
provision of welfare goods and the level of retirement income can explain part of the puzzles
First fact: an increase of the welfare goods (health,
long-term care) at old age with a concomitant decrease in other goods & services
Second fact: In most countries welfare goods are
publicly provided and heavily subsidised
Third fact: Replacement ratios have increased and are
currently higher in PAYG than in the funded systems
NB: note that PAYG are unsustainable and
replacement rates in funded systems are expected to be increase relative to PAYG
Two groups of countries considered:
Group Funded systems: Funded pension
systems have been developed and pension replacement rate are relatively low:
US, UK, CAN, CHE, NLD, AUS, DNK
Group PAYG: countries dominated by public
PAYG pension systems and pension replacement rate is relatively high:
AUT, GER, FINL, FRA, JAP, COR, NOR,
90 110 130 150 170 190 210 230 Under 25 25-34 35-44 45-54 55-64 65-74 75 and + Co n s u m p ti on o f t h e un de r 2 5 g ro u p = 100 90 110 130 150 170 190 210 230 250 270 290 Under 25 25-34 35-44 45-54 55-64 65-74 75 and + Co n s u m p ti on o f t h e un de r 2 5 g ro u p = 100 50 70 90 110 130 150 170 190 210 230 250 Under 25 25-34 35-44 45-54 55-64 65-74 75 and + C ons um pt io n o f the un de r 2 5 g roup = 10 0 0 20 40 60 80 100 120 Under 25 25-34 35-44 45-54 55-64 65-74 75 and + Co n su m pt io n o f th e u n de r 2 5 g roup = 1 00 Housing Energy Entertainments Education 90 140 190 240 290 340 390 440 490 540 590 Under 25 25-34 35-44 45-54 55-64 65-74 75 and + Co n su m pt io n o f th e u n de r 2 5 g roup = 1 00 Health 40 60 80 100 120 140 160 180 200 Under 25 25-34 35-44 45-54 55-64 65-74 75 and + C o ns u m pt io n o f th e u nde r 2 5 g ro up = 10 0 Vehicles
USA Household Panel Survey
2
stfact: Health spending is heavily subsidised
Share of public health expenditures in total health expenditures (in %)
1970 19711972 1973 1974 1975 1976 1977 1978 1979 19801981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 199319941995 19961997 19981999 2000 20012002 50 55 60 65 70 de pp ub r 1970 1980 1990 2000 2010 styears Portugal 1970 19711972 1973 1974 1975 1976 1977 19781979 19801981 19821983 1984 1985 19861987 198819891990 1991 1992 1993 1994 199519961997 19981999200020012002 36 38 40 42 44 46 de pp ub r 1970 1980 1990 2000 2010 styears USA 1970 19711972 1973 197419751976 197719781979 1980 1981 198219831984 19851986 1987 19881989 1990 1991 199219931994 19951996 199719981999 20002001 2002 84 86 88 90 92 de pp ub r 1970 1980 1990 2000 2010 styears Sweden
3
stfact: Retirement incomes have generally
increased
average replacement rates (in %)
20 40 60 80 1980 1985 1990 1995 2000 26 RR annees Graphs by group(pays) 65 70 75 80 1980 1985 1990 1995 2000 11 RR annees Graphs by group(pays) 40 45 50 55 1980 1985 1990 1995 2000 30 RR annees Graphs by group(pays) 45 50 55 60 1980 1985 1990 1995 2000 28 RR annees Graphs by group(pays) Portugal France USA Sweden
How the links among these facts can explain the
puzzles?
The ageing-consumption and ageing-saving puzzles: if
old-age consumers shift their consumption structure towards goods that are heavily subsidised (higher T)
and receive increased retirement income (higher β), this could both induce a decline of total consumption and a surplus of saving at older ages.
But there is an explained fact: why fully-funded save
Capitalization seems to influence negatively
savings
Panel 25 OECD countries for the period 1993-2005 193 ) .40 4 -( _ _ 12 . 0 4 . 11 193 ) 72 . 4 -( 47 . 0 23 . 9 1 , 1 , = − + + ⋅ − = = − + ⋅ − = − − N t student effects Fixed Country Cap Sav N t student Cap Sav it t i it it t i it ε ε
Even in the group of fully-funded systems over
time (within) the relation is negative
aus aus ausau s au s aus aus aus aus aus aus can cancan can cancan cancan ca n can can
che che che che ch e che dnk dnk dnk dnk dn k dnk dnk dnk dnk dnkdn k gbr gbr gbrgbr gbr gbr gbr gbr gbr gbr gbr nld nld nl d nl d nld nl d nl d nl d nld nld nld usa usa usa
usa usausa usa usa usa usa usa -5 0 5 10 15 SA V 20 40 60 80 100 120 CAP
Countries High funding (in % GDP)
65 ) 55 . 4 ( _ _ 12 . 0 9 . 12 65 ) 45 . 5 ( 84 . 0 06 . 0 1 , 1 , = − − + + ⋅ − = = − + ⋅ + = − − N t student effects Fixed Country Cap Sav N t student Cap Sav it t i it it t i it ε ε
Illustration: in the USA and Canada household Saving rates have been declining over time in parallel with
increased share of pension funds in GDP
1993 1994 1995 1996 1997 1998 1999 2000 2001 20 02 20032004 2005 0 2 4 6 SA V 50 60 70 80 90 CAP 1993 19941995 1996 1997 19 98 1999 2000 2001 2002 2003 2004 2005 2 4 6 8 10 12 SA V 35 40 45 50 55 CAP USA Canada
This decline of savings within countries can be partly explained by the
increasing replacement rates, higher subsidisation of welfare goods and the increased share of old people in the population
Ricardian Equivalence also helps explaining
the capitalization/saving puzzle
PAYG Systems Funded systems
The Ricardian Equivalence is particularly strong in PAYG systems, where budget consolidation has been weaker than in fully-funded systems
aut aut aut
aut aut autautaut
aut aut bel be l bel bel bel bel bel bel deu deu de udeudeu deudeu deu deu de u esp esp esp esp esp esp esp esp fin fin fin fin fin finfin fin fin fin frafrafra frafrafrafrafrafra
ita ita itaita ita ita ita ita ita jpn jpn jpn jpn jpn jpn jpn jpn nor nor nor norno r nor nor nor
nor prt prt prt prt prt prt sweswe swe swe swe swe swe swe swe swe 0 5 10 15 20 25 SA V -15 -10 -5 0 5 10 nlgqa aus aus aus aus aus aus aus aus aus aus can can can can can can cancan can can chechecheche
cheche che che dnk dnk dnk dnk dnk dnk dnk gbr gbrgbr gbr gbr gbr gbr gbr nld nl d nld nld nld nld usa usausa
usa usa usa usa usa usa usa -5 0 5 10 15 SA V -15 -10 -5 0 5 nlgqa
4. An econometric test
A reduced form approach taking into account a number of determinants of household saving rates identified in the literature:
(i) Short-term and macroeconomic determinants:
- GDP per capita (wealth effects) - Long term real interest rate
- Public budget balance (in % GDP), in order to test the Barro-Ricardo equivalence hypothesis (Barro, 1974)
(ii) Social security and welfare systems determinants:
- Average replacement rate (in public and private schemes)
- Public share on total health expenditures (proxy for the provision of welfare goods)
(iii) Structure of the population:
- Shares of prime (25-59) and old-age (60+) population
Panel estimates of household saving rates
Total PAYG systems Fully-Funded
systems
Public budget -0.653*** -0.664*** -0.423** (-7.35) (-7.04) (-2.31)
Real interest rate 0.269*** 0.124 0.340* (2.96) (1.27) (1.91) GDP p.c. 0.00000700*** 0.00000769*** 0.0000826 (3.40) (3.55) (1.02) Share of pop 25-59 -0.963 0.449 -0.333 (-1.43) (0.60) (-0.18) Share of pop 60-99 -3.399*** -1.947** -3.103** (-4.82) (-2.23) (-2.23)
Share of public health exp. -0.312*** -0.394*** -0.0723 (-3.45) (-3.57) (-0.49) Life expectancy at 60 -1.089 -2.318* 2.485 (-1.01) (-1.69) (1.06) Replacement rate -1.446** -0.763 -2.357 (-2.28) (-1.12) (-1.41) Replacement rate*Pop 25-59 0.0150 0.0125 0.00688 (1.15) (0.95) (0.19) Replacement rate*Pop 60-99 0.0555*** 0.0140 0.143*** (4.65) (1.07) (4.06)
Life exp. at 60*Pop 60-99 -0.0239 0.00524 -0.0968***
(-1.57) (0.34) (-3.20) Time trend 0.165 0.0321 -0.535 (0.95) (0.15) (-1.26) Constant 155.5*** 85.70* 85.51 (4.63) (1.76) (1.14) Number of observations 245 134 111 Number of countries 18 11 7
Summary & further research
Our empirical results show that the three puzzles are linked together.
The changing structure of consumption with age, together with a large subsidy for welfare goods and increasing replacement rates provides an explanation for both the ageing-consumption and ageing-saving puzzles
In line with a Ricardian equivalence effect, the level of the public
budget balance has a negative impact on savings. This explains the observed saving-capitalisation puzzle.
In line with standard life-cycle effects, we also showed that an
increase in the share of the old-age population has a strong negative impact on the saving rate.
Finally, an additional longevity-saving puzzle emerged. This would