The Impact of Global Aging on Saving,
Investment, Asset Prices, and Returns
for the Q Group – The Institute for Quantitative Research in Finance
Gary Burtless
The Brookings Institution
Washington, DC USA
October 18, 2005
Carlsbad, California
Population ageing is occurring in
nearly all countries
„
„
In rich countries, aged population obtains
much of its income & consumption from
government budgets
Populations in both rich and poor
countries are growing older
‹ Falling
mortality rates / Longer life
expectancy
‹ Declining fertility
‹ Immigration is variable & uncertain
Impact of ageing on financial
markets and the economy
ƒ
Relative declines in rates of saving
and investment
ƒ
Will aged societies be net lenders or net
borrowers (current account balance)?
ƒ
Implications for asset prices and
returns
ƒ
Open-economy aspects
ƒ
ƒ
Current account balance (national saving
less domestic investment)
Interaction with demographic change
Life-cycle consumption theory: Age profile of saving
Annual earnings and consum ption
Earnings
25,000
20,000
15,000
Consumption
10,000
5,000
0
20
30
40
50
60
70
Age
The difference is annual saving
or dissaving.
80
Life-cycle consumption theory: Wealth accumulation
Wealth
200,000
150,000
100,000
50,000
0
-50,000
20
30
40
50
Age
60
70
80
Do changes in the age structure have the
impact on aggregate household saving implied
by the life-cycle hypothesis?
ƒ Microeconomic analysis
ƒ Weak life-cycle influences
ƒ Many people have no significant
retirement saving
ƒ Highly disparate patterns of wealth
accumulation complicates aggregation
ƒ Average dollar, not average person
ƒ Wealthy savers have out-size effect on aggregate
saving
ƒ Importance of cohort effects
Life-cycle consumption theory: Cross-Section Data
Factor income received by average 45-49 year-old = 100
120
100
100
103
94
Capital income
Labor income
94
88
77
80
81
62
85
88
87
72
74
71
60
59
49
38
48
40
33
32
31
26
32
21
20
20
7
6
10
0
15-19
25-29
35-39
45-49
Age group:
55-59
65-69
United States
6
75-79
3
2
85-89
3
Age pattern of U.S. saving rates:
Consumption Survey Results
Saving measured as after-tax income minus consumption expenditures (CEX)
18
Percent of after-tax income
16
15.8
14
12
11.5
10
8
10.6
9.6
10.8
8.6
6
4
2
0
25-34
35-44
45-54
55-64
65 and older
Total
s am ple
Age group
Source: Bosworth, Burtless & Sabelhaus (1991).
Age pattern of U.S. saving rates:
Wealth Survey Results
Saving measured as change in wealth minus estimated capital gains (SCF)
16
Percen t of after-tax in com e
14
12
13.6
10
10.1
10.3
10.6
9.5
8
6
4
2
2.5
0
25-34
35-44
45-54
55-64
65 and older
Total
sample
Age group
Source: Bosworth, Burtless & Sabelhaus (1991).
Macroeconomic analysis
ƒ Shows stronger evidence of demographic effects on
saving
ƒ Most pronounced for Asia
ƒ Results for industrial economies sensitive to countries
and time period.
ƒ But – the aggregate trend in personal saving often
contradicts theory’s prediction: U.S. saving has
declined since mid-1980s.
ƒ Multi-country panel data sets
ƒ Cross-national differences are correlated with other
determinants.
ƒ Within-country demographic changes are small
compared to changes in saving
Do changes in the age structure have the
impact on investment demand implied
by the life-cycle hypothesis?
ƒ Slower labor force growth should
reduce growth of required capital
stock.
ƒ Limited research on link between
labor force growth and technical
change
There are only a few studies of impact of aging
on investment demand.
ƒ Higgins (1998) concludes that the decline in
investment will exceed that of saving for highincome economies out to 2025.
ƒ Bosworth and Keys (2004) also obtained
strong demographic effects, but -ƒ Decline in saving projected to exceed that of
investment by 2050
ƒ High-income countries would have current account
deficits.
Do changes in the age structure have a
significant impact on asset prices?
ƒ Asset prices assumed to vary in response to
changes in the capital-output ratio.
ƒ Price of capital assumed to depend on marginal
product, or relative scarcity
ƒ Age-related demand for assets
ƒ Mankiw-Weil (1989) – Housing
ƒ Demographically determined asset prices
Problem with Mankiw-Weil prediction
Demographic measure of
"change in housing demand"
Prediction
Annual rate of change in hous ing price s
(5-ye ar lagge d moving ave rage )
Estimation period
5.0%
Change in price
(left axis)
4.0%
3.0%
2.0%
1.0%
0.0%
-1.0%
-2.0%
-3.0%
-4.0%
-5.0%
1945
1950
1955
1960
1965
1970
1975
Year
Source: Bosworth, Bryant & Burtless (2004).
1980
1985
2.6%
2.4%
2.2%
2.0%
1.8%
1.6%
1.4%
1.2%
1.0%
0.8%
0.6%
Change in demand
0.4%
(right axis)
0.2%
0.0%
1990 1995 2000 2005
ƒ Pension fund accumulation and decumulation
Schieber-Shoven (1997)
ƒ Age-related portfolio preferences
Poterba (2001) and Davis and Li (2003)
ƒ Proportion of portfolio devoted to risky assets
rises up to age 60, but does not decline
much thereafter
ƒ Effects of demographic structure on returns
varies across countries
Age pattern of household portfolios:
Risky asset as % of net financial assets (U.S.A.)
Common stocks as a percent of net financial assets
60
50
53
50
45
40
42
41
30
33
38
36
33
31
27
20
22
10
0
0
15-19
20-24
25-29
30-34
35-39
40-44
45-49
50-54
55-59
60-64
65-69
70-74
75+
Age group
Source: Poterba (2001).
Age pattern of household portfolios:
Risky asset as % of household net worth (U.S.A.)
Common stocks as a percent of net worth
20
18
18
16
14
15
12
16
16
15
16
17
12
10
8
9
10
6
6
4
2
0
3
0
15-19
20-24
25-29
30-34
35-39
40-44
45-49
50-54
55-59
60-64
65-69
70-74
75+
Age group
Source: Poterba (2001).
Age pattern of household net worth:
Tabulations of U.S. Survey of Consumer Finances (1983-1995)
Net worth (1995 dollars)
210,000
180,000
150,000
120,000
90,000
60,000
30,000
0
15-19
20-24
25-29
30-34
35-39
40-44
45-49
50-54
55-59
60-64
65-69
70-74
Age group
Source: Poterba (2001).
75+
Research Agenda
ƒ Saving
ƒ There is a large disconnect between the micro
and macroeconomic research on saving
ƒ Investment
ƒ Panel data for large sample of countries in
order to refine estimates of demographic
effects
ƒ Linkage between labor force growth and
technological change
ƒ Asset Prices
ƒ Improve the microeconomic data to
incorporate pension funds
Cross-Border
Impacts of Aging
Enhanced Cross-Border Economic
Integration
ƒ
2 sets of underlying causes:
¾ Technological innovations and social/cultural
changes reducing economic and psychic distances
between nations
¾ Lowering of barriers at borders
ƒ
2 secular trends in cross-border “substitutabilities”:
¾ Goods: household and firms more willing to
substitute home and foreign goods in response to
relative price changes
¾ Financial: savers and investors responding more
strongly across borders or currencies to changes in
relative expected returns
Implications for an Economy’s
Saving-Investment Balance
ƒ Macroeconomic variables more closely linked across
borders than before
ƒ Macroeconomic adjustments channeled relatively more
through external-sector transactions; cross-border and
cross-currency adjustments more important relative to
domestic adjustments
ƒ Hence larger swings in current-account balances due to
heightened goods and financial “substitutabilities”
ƒ Larger imbalances occur, not only ex ante but also ex
post, between national saving and domestic
investment
Main conclusion to keep in mind –
ƒ As economies become more open, there may be a
reduction in the correlation between domestic investment
and national saving.
ƒ Equivalently, there may be a tendency for current-account
imbalances to become larger and more variable.
ƒ Hence cross-border and cross-currency adjustments to
policy and non-policy shocks rise in importance relative to
purely domestic adjustments.
Demographic Shocks in Open Economies
ƒ
ƒ
Different pace and intensity of demographic
change in the world, among developed
economies, and especially between industrial
and developing countries
External-sector transactions, exchange rates and
variations in current account balance are
important parts of macroeconomic adjustments
to demographic shifts
Uneven pace of population
ageing across countries
„
Fast-ageing
areas
„
Europe
‹ Korea
‹ Japan
Indonesia
‹ Nigeria
‹ Philippines
‹
„
Slow pace of
ageing
‹
Intermediate
pace
USA / Canada
‹ Australia
‹ Latin America
‹
Unequal change in old-age dependency
Old-age dependency ratio
(Ratio of people 65 years old and older to persons age 20-64)
Japan
78.2
80
70
Russia
60
Thailand
50
USA
40
Japan
30
20
10
0
1950
38.2
1960
1970
1980
1990
2000
Year
Source: U.N. Population Projections (2004).
2010
2020
2030
2040
2050
Thailand
Demographic Shocks in Open Economies
ƒ A nation’s dependency ratios – youth
and elderly – may be correlated with
variations of its current account (S-I)
balance
ƒ International capital flows and externalsector goods transactions can “cushion”
some of the impacts of demographic
change on domestic macroeconomic
variables.
Conclusions -„
Micro & macro studies find age profile roughly
corresponds to life-cycle prediction
„
BUT – micro studies show smaller effect of aging …
possibly because pension saving is missed in micro
studies
„
Recent saving trends in many industrialized
countries don’t fit the theory: Aggregate saving has
dropped in spite of more middle-aged savers
„
Slower growth in working-age population should
depress investment demand
„
Whether declining saving or shrinking investment
demand will have bigger effect is open question
Conclusions -„
Different age groups have different preferences for risky &
non-risky assets
„
In theory, this could mean that aging will influence demand
for different asset classes and expected returns
„
Micro & macro studies have found age effects on asset
holdings and returns of different asset classes
‹
In micro studies, demand for risky assets among very
aged implies little if any decline in overall demand for
risky assets in foreseeable future
‹
Macro studies fail to uncover consistent pattern of
change in asset demand with aging
‹
My guess: The “predictable” impact of age structure
on asset demand is small relative to unpredictable
sources of change