Demographic Changes, Saving, and Current Account - 麻省理工学院人工智能研究专业

Demographic Changes, Saving, and Current Account
Demographic Changes, Saving, and Current Account

Demographic Changes, Saving, 和
Current Account in East Asia*

Soyoung Kim
经济系
Korea University
Anam-Dong, Sungbuk-Gu,
Seoul, 136-701
Republic of Korea
soyoungkim@korea.ac.kr

Jong-Wha Lee
经济系
Korea University and the
Australian National University
jongwha@korea.ac.kr

抽象的
This paper analyzes the empirical relationships among demo-
graphic changes, saving, and current account balances in East Asia.
The panel Vector-Auto Regressive (VAR) model shows that an
increase in the dependency rate, especially the elderly depend-
ency rate, significantly lowers saving rates and subsequently wors-
ens current account balances. The result implies that the future
aging of the population in East Asia would have a significant
impact on global capital flows and current account imbalances.

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

6
2
2
2
1
6
8
2
3
4
2
A
s
e
p
2
0
0
7
6
2
2
2
p
d

乙
y
G
你
e
s
t

哦
n
0
7
S
e
p
e
米
乙
e
r
2
0
2
3

1. 介绍

Over the last three decades, the major industrialized coun-
tries have experienced large increases in elderly popula-
tion because of the drop in fertility rates and of the baby
boom generation becoming older. The aging of the popula-
tion is expected to accelerate over the next 3 几十年. 为了
advanced countries as a whole, the old dependency rate,
or the share of elderly population aged 65 and above in to-
tal population, is forecasted to increase from 15 百分比在
2005 到 26 百分比在 2050.

East Asian economies have also experienced rapid demo-
graphic changes, with Japan as the well-known case. 自从
1995, Japan has become an “aged society” with the old
dependency rate of 14 percent or higher, and is forecasted
to become a “super-aged society” by 2025 with the old
dependency rate of over 20 百分. The aging process is
also fast in emerging Asian economies such as Hong

* We are grateful to the participants at the September 2006 Asian
Economic Panel meeting, Keio University, 东京, for helpful
comments.

Asian Economic Papers 6:2

Institute of Technology

Demographic Changes, Saving, and Current Account

孔, 韩国, 新加坡, and Taiwan, which are expected to turn into “super-
aged societies” by 2025.

The purpose of this paper is to investigate empirically the implications of demo-
graphic changes in East Asia, 尤其, focusing on saving rates and current ac-
count balances. 理论上, the fast-aging population and the shrinking labor force
will have a signiªcant impact on national saving and investment. The current ac-
count is, by deªnition, identical to the imbalance between national saving and do-
mestic investment. 所以, a saving–investment imbalance in an individual coun-
try determines its current account balance and net capital ºow. It is an important
question if the demographic change has a signiªcant impact on current account
balances or net capital ºows in East Asia. 最近几年, enlarging global current
account imbalances have been one of the major concerns in the international com-
社区. The global imbalances are characterized by continued accumulation of cur-
rent account surpluses in East Asia mirrored by sustained current account deªcits in
美国. The East Asian current account surplus reºects their saving–
investment imbalances due to a savings glut or investment drought. Eichengreen
(2005) suggests that because of population aging, East Asia will have lower saving
rates and thereby reduce the imbalances. Despite this conjecture, not many studies
attempt to assess empirically the effects of demographic changes on saving rates
and current account balances in East Asia.

Compared to past studies, our empirical works have two distinct features in meth-
odology. 第一的, most of past empirical studies that analyze the macroeconomic effects
of demographic changes have adopted either cross-country regressions or simula-
tion approaches. The cross-country regression approach is mostly static in its nature,
although it is a data-based method. 另一方面, the simulation approach,
although showing dynamics, is less data-oriented. 相反, 这篇论文
adopts the panel Vector-Auto Regressive (VAR) 框架, which can combine two
nice features of previous studies: the empirical evidence provided here is data-
oriented and dynamic in its nature. 第二, by examining the effects of demo-
graphic changes on many related macroeconomic variables, we further explore the
detailed transmission and consequence of demographic changes. 例如, 我们
examine the effects on private saving, public saving, 投资, interest rate, 真实的
exchange rate, and GDP in order to infer the detailed transmission of demographic
changes on national saving.

This paper is organized as follows. 部分 2 describes the trend of demographic
changes and the current accounts in the world. 部分 3 explains economic theories
on the effects of demographic changes on saving and current account. 部分 4 关于-
views previous empirical studies on the links between demographic change, saving,

Asian Economic Papers

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

6
2
2
2
1
6
8
2
3
4
2
A
s
e
p
2
0
0
7
6
2
2
2
p
d

乙
y
G
你
e
s
t

哦
n
0
7
S
e
p
e
米
乙
e
r
2
0
2
3

Demographic Changes, Saving, and Current Account

and current account. 部分 5 introduces our data set consisting of 10 East Asian
economies and empirical methodology of a panel VAR model. In section 6, 我们
present and discuss the estimation results of demographic effects on saving,
current account, and other macroeconomic variables. Concluding remarks follow in
部分 7.

2. Demographic changes and current account in the world and East Asia

2.1 Demographic transition
One of the most important demographic changes is the aging of the world’s popula-
的. Most industrialized countries experienced a dramatic rise in births immedi-
ately following World War II. 随后, fertility rates dropped (数字 1). 作为
baby boom generation begins to retire, these countries face a signiªcant population
老化. The share of the population aged 65 和更老的 (elderly dependency rate) 有
increased steadily in advanced countries (数字 2). The aging process has been
most spectacular in Japan. The share of Japanese elderly population aged 65 和
above increased from 7.1 百分比在 1970 到 19.7 百分比在 2005.

数字 3 shows the trend of the share of young dependent population aged 14
and under (youth dependency rate). The share has declined steadily since 1960 在
advanced countries as a whole, changing from 28 百分比在 1960 到 17 百分比在
2005. As the rise in the elderly dependency rate begins to surpass the decline in the
young dependency rate in recent years, the total dependency rate declines in many
advanced countries. 换句话说, the share of working-age population (aged be-
补间 15 和 64) begins to shrink. The trend of increasing aged population and
shrinking working-age population is expected to continue in advanced countries
over the next several decades. Japan will experience the most rapid aging process
among advanced countries, with an elderly dependency rate of over 35 百分比在
2050. The downward trend of the working-age population share will also be most
spectacular in Japan: the working-age population share is forecasted to decrease
从大约 66 百分比在 2005 到 51 百分比在 2050.

Most developing countries are only at the beginning stages of the demographic
transition characterized by an increase in life expectancy followed by declining fer-
tility rates. In most developing countries, the share of the elderly population has not
increased for the last several decades (数字 2). Reºecting their higher fertility rates
and slow aging process, the share of the working-age population has continued to
increase since 1970 in developing countries as a whole (数字 3).

Over the next several decades, most developing countries are expected to experi-
ence an increase in old-age population, but as the decline in young-dependency rate

Asian Economic Papers

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

6
2
2
2
1
6
8
2
3
4
2
A
s
e
p
2
0
0
7
6
2
2
2
p
d

乙
y
G
你
e
s
t

哦
n
0
7
S
e
p
e
米
乙
e
r
2
0
2
3

Demographic Changes, Saving, and Current Account

数字 1. Trend of fertility rates (children per woman), 1950–2050

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

6
2
2
2
1
6
8
2
3
4
2
A
s
e
p
2
0
0
7
6
2
2
2
p
d

乙
y
G
你
e
s
t

哦
n
0
7
S
e
p
e
米
乙
e
r
2
0
2
3

来源: Population Division of the Department of Economic and Social Affairs of the United Nations Secretariat, World Population

Prospects: 这 2004 修订. 台湾, Statistical Yearbook of the Republic of China, 2004.

Asian Economic Papers

Demographic Changes, Saving, and Current Account

数字 2. Trend of elderly dependency (percent of the population aged 65 以上),
1950–2050

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

6
2
2
2
1
6
8
2
3
4
2
A
s
e
p
2
0
0
7
6
2
2
2
p
d

乙
y
G
你
e
s
t

哦
n
0
7
S
e
p
e
米
乙
e
r
2
0
2
3

来源: Population Division of the Department of Economic and Social Affairs of the United Nations Secretariat, World Population

Prospects: 这 2004 修订. 台湾, Statistical Yearbook of The Republic of China, 2004.

Asian Economic Papers

Demographic Changes, Saving, and Current Account

is faster than a slower increase in old-age population, they continue to experience an
increase in the working-age population share (数字 4). 然而, a number of de-
veloping economies, particularly in East Asia, are predicted to begin the fast-aging
process soon. 在东亚, fertility rates continued to decline over the last decades.
The fertility rates are currently below 2.0 in most East Asian economies except Indo-
内西亚, Malaysia, and The Philippines. The share of young-age dependent population
has continuously declined since 1970 (数字 3). Population aging has already begun
in many East Asian economies. 在 2005, the share of elderly population is over
9 percent in Korea and Taiwan, 8.5 percent in Singapore, 和 7.6 中国百分比.
The fast-aging processes are expected to accelerate. 在 2050, the share of the elderly
population will increase to 35 percent in Korea and Taiwan, 31 percent in Singapore,
和 24 中国百分比 (数字 2).

The population aging, combined with a relatively stable trend of youth dependency
rate in the future, will lead to a decrease in working-age population share. 数字 4
shows that the share of the working-age population is forecasted to decrease from
2010 in most East Asian economies. It is expected to decline from 73 百分比在 2010
到 54 百分比在 2050 in Korea and from 72 百分比到 61 percent in China over the
same period.

2.2 Current account balances
最近几年, there have been increasing concerns regarding the growing global
current account imbalances. The industrial countries as a whole have continued to
show current account deªcits or net capital inºows, mainly due to the U.S. 他们的-
tained current account deªcits (数字 5). 在美国, current account
deªcits widened signiªcantly since the 1990s. 在 2004, 美国. deªcit stood at
$666 十亿, 从 $136 十亿 1997. It amounted to 5.7 percent of GDP, 增加-
ing from 1.6 百分 7 years earlier. 相比之下, Japan has persistently accumulated a
large amount of current account surpluses. 在 2004, Japan had a current account sur-
plus of $172 十亿, amounting to 3.7 percent of GDP. Emerging market economies in East Asia also had big surpluses, especially since the 1997–1998 Asian ªnancial crisis (数字 6).The four East Asian emerging economies as a group had a surplus of $90 十亿, 或者 7.1 percent of their GDP. The ASEAN-4 economies as a group had a
surplus of $28 十亿, 或者 4.4 percent of their GDP. China also had a surplus of about $70 十亿 2004 或者 4.2 percent of GDP.

The global imbalances in the current account can be interpreted as showing the dis-
parities in the saving and investment in each region and country. 尤其, 这
massive East Asian current account surpluses reºect high saving rates relative to
investment in East Asia. Except for China, East Asian economies have experienced a
secular downward trend in both saving and investment rates since the mid 1990s

Asian Economic Papers

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

6
2
2
2
1
6
8
2
3
4
2
A
s
e
p
2
0
0
7
6
2
2
2
p
d

乙
y
G
你
e
s
t

哦
n
0
7
S
e
p
e
米
乙
e
r
2
0
2
3

Demographic Changes, Saving, and Current Account

数字 3. Trend of youth dependency (percent of the population aged 14 and under),
1950–2050

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

6
2
2
2
1
6
8
2
3
4
2
A
s
e
p
2
0
0
7
6
2
2
2
p
d

乙
y
G
你
e
s
t

哦
n
0
7
S
e
p
e
米
乙
e
r
2
0
2
3

来源: Population Division of the Department of Economic and Social Affairs of the United Nations Secretariat, World Population

Prospects: 这 2004 修订. 台湾, Statistical Yearbook of The Republic of China, 2004.

Asian Economic Papers

Demographic Changes, Saving, and Current Account

数字 4. Trend of working-age population (percent of total population), 1950–2050

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

6
2
2
2
1
6
8
2
3
4
2
A
s
e
p
2
0
0
7
6
2
2
2
p
d

乙
y
G
你
e
s
t

哦
n
0
7
S
e
p
e
米
乙
e
r
2
0
2
3

来源: Population Division of the Department of Economic and Social Affairs of the United Nations Secretariat, World Population

Prospects: 这 2004 修订. 台湾, Statistical Yearbook of the Republic of China, 2004.

Asian Economic Papers

Demographic Changes, Saving, and Current Account

数字 5. Changes in global current account balances, 1980–2005年

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

6
2
2
2
1
6
8
2
3
4
2
A
s
e
p
2
0
0
7
6
2
2
2
p
d

乙
y
G
你
e
s
t

哦
n
0
7
S
e
p
e
米
乙
e
r
2
0
2
3

来源: IMF, World Economic Outlook database, 九月 2005.

笔记: Emerging Asia-East Asian NIES plus ASEAN-4.

(人物 7 和 8). 然而, investment rates fell dramatically after the ªnancial
crisis, with their declines surpassing the declines in savings. In a group of four East
Asian Newly Industrialized Economies (NIES), investment rates dropped from
31.6 百分比在 1997 到 23.0 percent of GDP in 2003, whereas the national saving rate
decreased from 32.3 百分比到 29.4 percent over the same period.

3. Theoretical review

理论上, a change of population age structure has signiªcant effects on various
components of the current account. 更确切地说, an increase in youth and elderly
dependency rates affects saving and investment, and thereby the current account.

Asian Economic Papers

Demographic Changes, Saving, and Current Account

数字 6. Current account in East Asia (as a percentage of GDP), 1980–2003

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

6
2
2
2
1
6
8
2
3
4
2
A
s
e
p
2
0
0
7
6
2
2
2
p
d

乙
y
G
你
e
s
t

哦
n
0
7
S
e
p
e
米
乙
e
r
2
0
2
3

Asian Economic Papers

Demographic Changes, Saving, and Current Account

数字 7. Saving in East Asia (as a percentage of GDP), 1980–2003

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

6
2
2
2
1
6
8
2
3
4
2
A
s
e
p
2
0
0
7
6
2
2
2
p
d

乙
y
G
你
e
s
t

哦
n
0
7
S
e
p
e
米
乙
e
r
2
0
2
3

Asian Economic Papers

Demographic Changes, Saving, and Current Account

数字 8. Investment in East Asia (as a percentage of GDP), 1980–2003

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

6
2
2
2
1
6
8
2
3
4
2
A
s
e
p
2
0
0
7
6
2
2
2
p
d

乙
y
G
你
e
s
t

哦
n
0
7
S
e
p
e
米
乙
e
r
2
0
2
3

Asian Economic Papers

Demographic Changes, Saving, and Current Account

One of the most important consequences of an increase in dependency rate, fre-
quently discussed in past theoretical studies, is the decrease in private saving. 在
core of saving theory is the life-cycle hypothesis, which presumes that individuals
pass through three broad phases of savings and consumption during their lifetimes.
If the household’s labor income is hump-shaped—low in young age, rising, 和
then falling in old age—then consumption smoothing implies that saving rates are
high in their productive middle ages and low in young and old age (Modigliani
1970). 因此, in an economy as a whole, the theory predicts that saving rate tends to
be lower when the share of the youth and old-age dependent in total population be-
comes higher.

Other theories expand the life cycle theory to include the inºuence of other factors.
卡罗尔 (1997) highlights the role of uncertainty for precautionary saving. The uncer-
tainty in the timing of retirement and death can exert major inºuence on the pace of
dissaving among the old and retirees. Leaving bequests to children will also tend to
dampen the decline in saving during retirement.

Public saving is also likely to drop with increased dependency rates. A smaller work
force can result in lower tax revenues. If the government does not change its ªscal
arrangements—a social security plan, for example—an increase in dependents, par-
ticularly elderly people, will increase public spending. 所以, public saving is
likely to fall. 实际上, the increase in the elderly population has been raising pressures
on pension, 健康, and other areas of public spending, deteriorating ªscal positions
in almost all advanced countries.

An increase in dependency rates is also expected to affect negatively the demand for
domestic investment. Slower labor force growth and lower expected output growth
will decrease the rates of return to investment. 除非 (labor-augmenting) techno-
logical progress accelerates, domestic investment demand must decline. 然而,
in the short run, it can happen that investment increases because of the need to sub-
stitute capital for a falling supply of labor input.

The current account is equal to the difference between saving and investment, 和
to net capital ºows, though a non-negligible size of statistical discrepancy is often
found in the actual data. The effect of the dependency variables on the current ac-
count is simply equal to the net of their negative effects on saving and domestic in-
vestment. 因此, the relative force of changes in national saving and domestic in-
vestment determines the impact of the increased dependency on current account
balance. An increase in dependency rates, if it lowered national saving rates faster
than domestic investment, would affect the current account adversely.

Asian Economic Papers

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

6
2
2
2
1
6
8
2
3
4
2
A
s
e
p
2
0
0
7
6
2
2
2
p
d

乙
y
G
你
e
s
t

哦
n
0
7
S
e
p
e
米
乙
e
r
2
0
2
3

Demographic Changes, Saving, and Current Account

希金斯 (1988) argues that the demographic “center for gravity” for investment de-
mand should be earlier in the age distribution than that for savings supply. Invest-
ment demand is related positively to the share of the youth, or labor-force growth,
whereas savings supply is most closely related to the share of mature adults (close
to retirement ages). This implies that when the youth dependency rate rises initially,
investment demand can increase, while saving declines. 然后, as the age distribu-
tion shifts toward the middle age, saving increases and investment demand de-
克莱恩斯, thereby pushing current account into surplus. 最终, an increase in the
elderly dependency rate would cause saving to fall, surpassing investment demand
衰退, thereby pushing the current account into a deªcit.

The magnitude of the current account adjustment to saving and investment imbal-
ance depends critically on the extent of the country’s openness. In a closed economy,
aggregate national saving and aggregate domestic investment move together.1 A fall
in the saving rate would lead to a shortage of funds in the ªnancial market, 和
would raise the price of funds, or the real interest rate. 反过来, the real interest rate
fall would decrease investments. 因此, no change would occur to the current ac-
count balance. In a small open economy with perfect capital mobility, a fall in saving
rate would lead to the current account deªcit as the increase in the gap between do-
mestic saving and investment is ªlled by an increase in net capital inºows. In inter-
mediate cases (例如, a large open economy or an open economy with imper-
fect capital mobility), real interest rate increases, investment falls, and current
account worsening may be observed altogether.

4. Empirical links between demographic change, saving, and current account

This section brieºy reviews the existing empirical literature analyzing the impact of
demographic changes on saving and current account balances.2

Many studies have investigated empirically the impacts of population age structure
on saving and current accounts. Most of the empirical literature in this area has
adopted four distinct approaches: 1) regressions using a repeated cross-section data;
2) regressions using macroeconomic time-series data for a broad cross-section of
国家; 3) simulations of a multi-region general equilibrium model; and a 4) VAR
方法.

1 If ªnancial markets are imperfect, agents may need to save more in order to undertake

lumpy physical capital investment in the future. See Liu and Wing (1994) for the discussion
of this “investment-motivated saving” hypothesis in the context of Taiwan’s experience.

2 Part of this survey is drawn from Kim and Lee (2007). See Bosworth, Bryant, and Burtless

(2004) and Helliwell (2004) for detailed surveys.

Asian Economic Papers

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

6
2
2
2
1
6
8
2
3
4
2
A
s
e
p
2
0
0
7
6
2
2
2
p
d

乙
y
G
你
e
s
t

哦
n
0
7
S
e
p
e
米
乙
e
r
2
0
2
3

Demographic Changes, Saving, and Current Account

A number of empirical studies on demographics and saving have used repeated
cross-section data observed at different points of time to create the saving proªles of
synthetic cohorts over time. Several country studies based on household survey
data ªnd that household saving remains unexpected among the elderly population,
contradicting the life-cycle hypothesis. Deaton and Paxson (1997) construct a syn-
thetic cohort-based measure of the age proªle of saving for four countries including
美国, 英国, 台湾, and Thailand, and ªnd the data for Britain and
Thailand do not coincide well with the predictions of the life-cycle model. 这可能
support the precautionary or bequest motives of saving. 实际上, several studies such
as Bernheim, Skinner, and Weinberg (2001) and Banks, Blundell, and Tanner (1998)
ªnd that people tend to decrease consumption expenditures substantially when
retired.

There has been skepticism of the use of household survey data. The old-age house-
holds that are counted as independent samples in household surveys are mostly
wealthy elderly who do not need to live with younger household members. 这
tends to generate an upward bias of the elderly saving rate when income and saving
are positively correlated. Demery and Duck (2006) construct a sample that adjusts
household saving into saving rates of individual household members in Britain and
ªnd evidence of a pronounced hump-shaped pattern of individual saving rates,
supporting the life-cycle theory. Microeconomic survey studies are not always con-
sistent in treating accumulation and decumulation of saving in a pension program.
The estimate of saving depends critically on whether household income includes
the contributions to the pension system as well as the beneªt from it. 例如,
many studies based on household surveys ignore contributions to a pension plan
when workers are active and include a beneªt (often annuity) as retirement income.
Jappelli and Modigliani (1998) and other researchers argue that this treatment is in-
correct and that contributions to a pension plan should be counted as saving and the
annuity beneªt must be excluded from retirement income. This new treatment con-
tributes to a pronounced hump-shaped pattern in saving rates over the life cycle.

Most empirical studies based on an econometric approach using macroeconomic
time-series data have found a strong negative link between dependency rates and
saving rates (Leff 1969; 韦尔 1994; Masson, Beyoumi, and Samiei 1998; 爱德华兹
1995; and Higgins 1988). 例如, Weil uses a panel data set of 14 工业的
countries over the period between 1960 和 1985, and ªnds evidence supporting
that both the share of young-age and that of the elderly population have strong and
statistically signiªcant negative effects on the private saving rates. He adopts the
ªxed-effect estimation technique that controls for the inºuences from unobserved
country-speciªc factors on the saving rates.

Asian Economic Papers

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

6
2
2
2
1
6
8
2
3
4
2
A
s
e
p
2
0
0
7
6
2
2
2
p
d

乙
y
G
你
e
s
t

哦
n
0
7
S
e
p
e
米
乙
e
r
2
0
2
3

Demographic Changes, Saving, and Current Account

In an empirical study utilizing a panel data set of more than 100 国家, 希金斯
(1988) ªnds that the youth and old age dependency rates have a strong negative ef-
fect on the national saving rate, domestic investment, and the current account bal-
安斯. The estimated demographic effects are quite large. 例如, the favorable
demographic swing in Japan over the period between the early 1950s and the early
1980s was found to be responsible for the increase in Japanese national saving by
5.6 percent and that of current accounts by 7.1 percent of GDP, 分别. 超过
next two decades until 2025, he projects that the major industrial countries including
Japan will experience substantial current account surpluses as the decline in invest-
ment demand exceeds the drop in saving. Higgins also ªnds that the estimated de-
mographic effects are larger for more open economies, which is consistent with the
implication of the Feldstein–Horioka proposition.

Chinn and Prasad (2003) and Luhrman (2003) use cross-country panel data to inves-
tigate the determinants of the current account. Chinn and Prasad, using a large sam-
ple for 18 industrial and 71 developing countries between 1971 和 1995, ªnd a
higher dependency rate. 尤其, the youth dependency rate has a signiªcant
negative effect on current accounts among developing countries.

Luhrmann also ªnds that countries with higher youth and elderly dependency rates
tend to import more capital from abroad, 那是, current account deªcits. She also
ªnds evidence that anticipated demographic changes affect current account bal-
ances signiªcantly. 尤其, future declines in youth dependency rates are
strongly associated with anticipated capital outºows or current account surpluses.
The projections based on the estimates show that due to the future demographic
changes in the world, 日本, 法国, and the United States are all expected to experi-
ence negative current account balance swings between 2000 和 2020, which will be
mostly driven by the anticipated future declines in youth dependency rate in devel-
oping regions.

The econometric approach based on macroeconomic data investigates the relation-
ship between demographic variables and current account separately, rather than as
a part of the integrated economic system. To address this issue, recent studies have
used calibrated general equilibrium models to examine the relationship between
demographic changes and current account.

Several studies focus on Japan, the fastest-aging society to assess the impact of
demographic changes on saving, 投资, and current account (Horioka 1992;
Faruqee and Mühleisen 2003; Dekle 2004; McKibbin and Nguyen 2004; 和
McKibbin 2005). 例如, Dekle uses a small open economy Ramsey-type
growth model to project the impact of demographic changes on Japanese saving and

Asian Economic Papers

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

6
2
2
2
1
6
8
2
3
4
2
A
s
e
p
2
0
0
7
6
2
2
2
p
d

乙
y
G
你
e
s
t

哦
n
0
7
S
e
p
e
米
乙
e
r
2
0
2
3

Demographic Changes, Saving, and Current Account

capital ºows. As population aging depresses saving, Japan is expected to import
capital amounting to 15 percent of GDP in 2015. Contrastingly, McKibbin, by using a
multi-country general-equilibrium model (MSG3), forecasts that global demo-
graphic changes will contribute to current account surpluses in Japan between 2005
和 2050, although the surplus declines over time. In this global general equilibrium
模拟, Japanese capital outºows are stimulated by the strong demand for capi-
tal from developing countries where labor force increases and rates of return to capi-
tal remain high.

There have been quite a few studies that use multi-region calibrated general equilib-
rium models to examine the effects of global demographic changes. Feroli (2003)
simulates a multi-region overlapping generations (OLG) 模型, which is calibrated
to match the demographic differences among the major industrialized economies
过去 50 年. Feroli ªnds demographic changes can explain the substantial
part of long-term capital ºows or current account trends experienced by the G-7 na-
系统蒸发散, such as the size and timing of the U.S. current account deªcits and Japanese
current account surpluses.

布鲁克斯 (2003) analyzes the global impact of population aging, using a calibrated
multi-region OLG model with perfect capital mobility to simulate the general equi-
librium effects of projected population trends on international capital ºows. 他
ªnds that retirement saving by aging baby boomers will raise the supply of savings
substantially above investment demand in both North America and the European
联盟, making both regions substantial capital exporters to other regions by 2010.
超过 2010, 然而, population aging will eventually make both regions become
capital importers.

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

6
2
2
2
1
6
8
2
3
4
2
A
s
e
p
2
0
0
7
6
2
2
2
p
d

IMF (2004) introduces simulation results from a multiregional macroeconomic
model—the INGENUE world model—which explicitly incorporates the age struc-
ture of the population and a pension system in each region. The results suggest
that the demographic changes projected over the next 50 years will decrease aggre-
gate saving rates in Europe and Japan. Pension systems contribute signiªcantly to
this decline due to the increase in additional pension expenditures to the retirees.
The saving decline causes substantial deterioration of the current account in these
地区.

乙
y
G
你
e
s
t

哦
n
0
7
S
e
p
e
米
乙
e
r
2
0
2
3

The multi-country general equilibrium models, 然而, often provide different
projections, which are the result of the differences in assumptions and model struc-
特雷斯. 例如, the simulation results with the MSG model in Batini, Callen, 和
McKibbin (2006) show that in most of the industrialized countries population aging

Asian Economic Papers

Demographic Changes, Saving, and Current Account

will lead to an improvement in current accounts because investment declines more
sharply than saving. 相似地, Bryant (2004) presents a simulation result of a two-
region model, supporting that the country with lower dependency rates exports
capital to the country with the more active population. These predictions contrast to
those of other simulation studies.

Cross-country regressions, although data-based, are mostly static in their nature. 在
另一方面, the simulation approach, although showing dynamics, is hard to
view as data-oriented empirical evidence. Kim and Lee (2007) adopt another
method of a panel VAR model to analyze demographic effects on saving rate and
current account balances in the G-7 countries. This framework can combine the nice
features of previous empirical methodologies; the empirical evidence provided here
is data-oriented and dynamic in its nature. It also allows incorporating interactions
between the demographics and macro variables in an integrated framework. 这
estimation results from the panel VAR model show that an increase in dependency
rate signiªcantly lowers saving rates, especially public saving rates, 导致
deteriorating current account balances in the G-7 countries.

5. Empirical methodology and data

5.1 Panel VAR modeling
We examine the effects of an increase in the dependency rate on the components of
the current account using panel VAR models. VAR models provide a useful method-
ology to investigate the issue. 第一的, dynamic effects can be inferred from VAR mod-
这. 第二, some interactions between the demographics and macro-variables can
be allowed in the VAR model, although most past studies assume an exogenous
process for the demographics. Demographic changes may not depend much on the
current economic activity level, but demographic change may be inºuenced by
lagged economic activity level. 例如, prolonged recession may result in a
high death ratio or low birth rate. 第三, VAR models are relatively free of ad hoc
identifying assumptions, so that data-oriented empirical results can be provided.
另一方面, we developed the panel model for East Asian countries, 反而
of constructing the model for each country, because the data points are relatively
short and the VAR model for each country would suffer from the degree of freedom
问题.

Let us assume that an economy i (我 (西德:2) 1,2,. . .,10) is described by the following struc-
tural form equation:

G L y
)
(

我
t

(西德:2)

我

(西德:3)

我
t

(1)

Asian Economic Papers

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

6
2
2
2
1
6
8
2
3
4
2
A
s
e
p
2
0
0
7
6
2
2
2
p
d

乙
y
G
你
e
s
t

哦
n
0
7
S
e
p
e
米
乙
e
r
2
0
2
3

Demographic Changes, Saving, and Current Account

i is an m (西德:3) 1 data vector, dt
where G(L) is a matrix polynomial in the lag operator L, y t
and di are m (西德:3) 1 constant matrices, m is the number of variables in the model, and e t
denotes a vector of structural disturbances. By assuming that structural distur-
我 ) can be denoted by (西德:4), which is a diagonal
bances are mutually uncorrelated, var( e t
matrix where diagonal elements are the variances of structural disturbances. 在-
dividual ªxed effect, 的, is introduced to control for the country-speciªc factors that
are not included in the model but affect saving rates, the current account, 和别的
macroeconomic variables. The time ªxed effect, dt, is introduced to control for the
global factors that are not included in the model but affect the current account and
its component.

我

We pooled the data and estimated the following reduced form panel VAR with the
time ªxed and individual ªxed effects:

我
t

(西德:2)

(西德:3)

我

(西德:3)

B L y
)
(

(西德:3)(西德:4)

我
t

你

我
t

(2)

where ct and ci are m (西德:3) 1 constant matrices, 乙(L) is a matrix polynomial in the lag
operator L, and var(u t

我) (西德:2) (西德:5).

To recover the parameters in the structural form equation from the estimated pa-
rameters in the reduced form equation, popular recursive restrictions on contempo-
raneous structural parameters are imposed, which is proposed by Sims (1980). 为了
more details on the methodology, refer to Kim and Lee (2007). 最后, note that our
statistical inference is not affected by the presence of non-stationary data because we
follow a Bayesian inference (see Sims 1988 and Sims and Uhlig 1991).3

5.2 Basic model
我, 是 {DEP, RGDP, SAV, CUR, RIR} where DEP is
In the basic model, the data vector, y t
the dependency rate (the percentage of a sum of population aged 65 and above and
aged 14 and below in the total population), RGDP is a log of real GDP per capita,
SAV is the national saving rates (the sum of private and government saving, 百分
of GDP), CUR is the current account (percent of GDP), and RIR is the (ex post) 真实的
interest rate (百分). A constant is included and two lags were selected based on
Schwartz criterion.

We included DEP, SAV, and CUR because they are the main variables of our inter-
埃斯特. A change in the population age structure is expected to inºuence saving and

3 特别是, we generate the standard error bands based on a Bayesian method, as described

in the RATS Manual.

Asian Economic Papers

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

6
2
2
2
1
6
8
2
3
4
2
A
s
e
p
2
0
0
7
6
2
2
2
p
d

乙
y
G
你
e
s
t

哦
n
0
7
S
e
p
e
米
乙
e
r
2
0
2
3

Demographic Changes, Saving, and Current Account

current account. Because current account is, by deªnition, equal to the saving–
investment imbalance, domestic investment is excluded in the model, but its change
can be inferred from the net of the changes in saving and current account. 注意
we ªrst focus on the effects of total dependency rate shocks, instead of analyzing the
effects of elderly dependency rate shocks and the effects of youth dependency rate
shocks separately, because the theory is better established for the effects of total de-
pendency rate changes. The extended model analyzes the effects of elderly depend-
ency rate shocks and the effects of youth dependency rate shocks separately.

Real GDP (RGDP) is included because it is the single best measure of economic
活动, and it is regarded as an important determinant of saving and investment
费率. 此外, the effects of demographic changes on RGDP may help us to infer
how demographic changes affect saving and current account. The real interest rate
(RIR) is included because it is likely to affect various components of current account,
such as saving and investments. 此外, RIR may also provide an important
clue on the transmission mechanism of demographic changes.

Our basic model assumes a recursive structure on the contemporaneous relation
among variables. The ordering is {DEP, RGDP, SAV, CUR, RIR}, where contempora-
neously exogenous ones are ordered ªrst. This assumption is not completely unrea-
sonable. A change in the dependency rates is likely to affect economic variables con-
temporaneously. 另一方面, the economic variables might affect the DEP in
the end, but not contemporaneously. Economic conditions may affect birth rates, 但
the effect will take at least a year as the gestation period is approximately 9 月.
It may be more controversial whether economic conditions would affect the death
rate and, 反过来, affect the elderly dependency rate contemporaneously.

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

6
2
2
2
1
6
8
2
3
4
2
A
s
e
p
2
0
0
7
6
2
2
2
p
d

然而, a temporary change in economic conditions may not affect the death rate
much, although a prolonged change in economic conditions may. 所以, 这
contemporaneous effects of the economic variables on the dependency rate may be
minor. 更远, the changes in the elderly dependency rate are likely to be the re-
sults of the population aging process (such as longer life expectancy in general),
rather than the results of immediate death rate changes due to changes in economic
状况. At least we can say that this assumption of contemporaneous exogeneity
is weaker than the assumption of most previous studies. (Most past studies as-
sumed that demographic variables are fully exogenous to other economic variables.)
最后, notice that the ordering of the other variables does not matter when we ex-
amine the effects of shocks to the DEP.4

乙
y
G
你
e
s
t

哦
n
0
7
S
e
p
e
米
乙
e
r
2
0
2
3

4 Refer to Christiano, Eichenbaum, and Evans (1999).

Asian Economic Papers

Demographic Changes, Saving, and Current Account

5.3 Extended model
We constructed extended models to examine the effects on other relevant variables
and to infer further the transmission of demographic changes on saving and current
帐户. The extended model adds a new variable to the basic model: {DEP, RGDP,
SAV, CUR, RIR, X}, where X is the new variable. 第一的, we included private and pub-
lic saving one by one in order to examine the effects on the component of the saving
X. 第二, private consumption and government consumption are included one by
one in order to further infer the private consumer decision and government behav-
ior. 最后, the real exchange rate is included in order to infer the transmission
mechanism and the consequence of the demographic changes on external balances.

We also constructed another extended model to infer the effects of changes in two
dependency rates, the elderly dependency rate and the youth dependency rate, sep-
arately. With these models, we would like to examine whether the effects of the
changes in the two dependency rates are different and whether the results in the
basic model are robust. The extended model is constructed as {DEPO, DEPY, RGDP,
SAV, CUR, RIR}, where DEPO is the elderly dependency rate and DEPY is the youth
dependency rate. There is no deªnitive theory on the ordering between DEPO and
DEPY. 所以, we also experimented with the alternative ordering of {DEPY,
DEPO, RGDP, SAV, CUR, RIR}.

5.4 数据
We consider 10 East Asian countries: 香港, 台湾, 韩国, 新加坡,
Malaysia, 印度尼西亚, The Philippines, 日本, 中国, and Thailand. Annual data are
用过的. The estimation period is 1981–2003.

Most data series including total population, youth dependency rate (人口,
ages 0–14, 占总数的百分比), elderly dependency rate (人口, 年龄 65 以上,
占总数的百分比), GDP in the current domestic currency unit and constant domestic
currency unit, gross investment in current domestic currency unit, private consump-
tion in constant domestic currency unit, government consumption in constant do-
mestic currency unit, (ex post) real interest rates, and real effective exchange rate are
obtained from the World Bank, World Development Indicator. Investment rate is con-
structed by dividing gross investment in current domestic currency unit by GDP in
current domestic currency unit. Gross national saving rates (gross national saving/
GDP), private saving rates (private saving/GDP), and public saving rates (治理-
ment saving/GDP) were obtained from the IMF, World Economic Outlook. 电流-
rent account balances (as a ratio to GDP) are constructed by the saving rates sub-
tracted by investment rates (那是, the ratio of gross investment in current domestic
currency unit to GDP in current domestic currency unit).

Asian Economic Papers

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

6
2
2
2
1
6
8
2
3
4
2
A
s
e
p
2
0
0
7
6
2
2
2
p
d

乙
y
G
你
e
s
t

哦
n
0
7
S
e
p
e
米
乙
e
r
2
0
2
3

Demographic Changes, Saving, and Current Account

Some exceptions follow. The real effective exchange rate for Hong Kong, 韩国,
and Taiwan is obtained from Bank for International Settlements. The real effective
exchange rate for Indonesia and Thailand is constructed by weighting the real
exchange rates against the United States and Japan (using trade weights), 哪里的
real exchange rates are constructed by using the nominal exchange rate and GDP
deºator. The real interest rates for Hong Kong, 印度尼西亚, Malaysia, and Taiwan are
constructed by subtracting the GDP deºator inºation rate from the nominal interest
速度. The nominal interest rates are obtained from the Web site of Asian Development
Bank. Commercial Bill rate is used for Hong Kong whereas a 12-month time deposit
rate is used for other countries. Other data series of Taiwan are obtained from
National Statistics.

6. Estimation results

6.1 Basic results
数字 9 reports the basic results and the impulse responses to shocks to the depend-
ency rate (DEP) over a 25-year period with 90 percent probability bands for the basic
ªve-variable VAR model. The responding variable names are denoted at the top of
each graph. The responses of RGDP are in terms of percent changes whereas others
are in terms of percentage point changes. Note that we also calculate the responses
of investment, by subtracting the current account from the saving.5 From the re-
sponses of DEP, we can infer the nature of the shocks. The point estimate shows that
the dependency rate increases by approximately 0.07 percentage point in a year, fur-
ther increases to approximately 0.38 percentage point in 10 年, and then de-
creases over time, returning to the initial level in approximately 23 years.6

In response to such DEP shocks, real per capita GDP decreases over time. The maxi-
mum negative effect, approximately a 1.5 percent drop, is found in approximately
10 years and reverts to the initial level in approximately 25 年. The drop in RGDP
最多 10 years is different from zero with at least 95 percent probability. The de-
crease in real per capita GDP is consistent with the theoretical prediction; 作为
dependency rate increases and fewer people earn (劳动) 收入, the real GDP per
capita is likely to drop.7

5 For each draw, investment rate responses are calculated by subtracting current account from

saving, and then the error bands are constructed.

6 The temporary nature of the dependency rate shocks is not surprising because the empirical

model controls the global factor by introducing the time ªxed effect.

7 The multi-country simulation models predict the decreased working population will cause
substantial declines in real GDP of advanced countries. 例如, Batini, Callen, 和
McKibbin (2006) forecast that the level of Japanese GDP in 2050 will be 30 percent lower than
would have been the case without the impact of demographic transition.

Asian Economic Papers

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

6
2
2
2
1
6
8
2
3
4
2
A
s
e
p
2
0
0
7
6
2
2
2
p
d

乙
y
G
你
e
s
t

哦
n
0
7
S
e
p
e
米
乙
e
r
2
0
2
3

Demographic Changes, Saving, and Current Account

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

6
2
2
2
1
6
8
2
3
4
2
A
s
e
p
2
0
0
7
6
2
2
2
p
d

乙
y
G
你
e
s
t

哦
n
0
7
S
e
p
e
米
乙
e
r
2
0
2
3

我
e
d
哦
米

C
我
s
A
乙

:
s
k
C
哦
H
s

e
t
A
r
y
C
n
e
d
n
e
p
e
d
哦
t

s
e
s
n
哦
p
s
e
r

e
s
我
你
p
米

我

.
9

e
r
你
G
我
F

Asian Economic Papers

Demographic Changes, Saving, and Current Account

Both saving and investment rates decline. The decrease in saving rates is especially
强的. Saving rates decrease over time. The maximum effect of approximately
0.4 percentage point decline is found in 10 年, and then reverts to the initial level
in approximately 23 年. The decrease in the saving rates from 4- to 14-year hori-
zons is different from zero with a 95 percent probability.

Investment rates also decline, but the size and the persistence of the decline are
smaller than that of the saving rate. The decline is signiªcantly different from zero
only at a 4-year horizon with 95 percent probability.

The current account also declined, because the saving rate fall is larger and more
persistent than the investment rate. The current account decreases over time. 这
maximum decline of 0.4 percentage point is found in about 14 年. 然后, it in-
creases back to the initial level in 25 年. 然而, the decline is not signiªcantly
different from zero at any horizon with 95 percent probability, but it is signiªcantly
different from zero from 10- to 15-year horizons with 90 percent probability.8

The estimates show that a substantial demographic effect exists on the saving and
current account. 例如, Japan and Singapore had a faster increase in the total
dependency rate in the 1990s (compared to other countries and its own past), 哪个
would have had a similar magnitude of a saving and current account rate fall (com-
pared to other countries and its own past).

These responses of saving, 投资, and the current account are consistent with
理论. The life-cycle theory (and the ªscal arrangements in these countries)
predicts the decline in saving rates as the dependency rate increases. The current
account deterioration can be also found when saving rates decline in the economy
relative to investment rate with internationally mobile capital.

The real interest rate decreases on impact but the decrease is not signiªcantly
different from zero. From the second year after the shock, the real interest rate in-
creases over time, and the maximum effect of 0.3 percent is found in approximately
12 年. The rises in the real interest rate from 5 到 16 years horizon are different
from zero with at least 95 percent probability. The rise in the real interest rate may
be understood because of a decrease in the supply of loanable funds in the economy,
which is shown by the drop in the saving rate.

These responses to the shocks to dependency rate in East Asian countries are quali-
tatively similar to those in G-7 countries, reported by Kim and Lee (2007). Kim and

8 The result is available from the authors.

Asian Economic Papers

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

6
2
2
2
1
6
8
2
3
4
2
A
s
e
p
2
0
0
7
6
2
2
2
p
d

乙
y
G
你
e
s
t

哦
n
0
7
S
e
p
e
米
乙
e
r
2
0
2
3

Demographic Changes, Saving, and Current Account

桌子 1. Forecast error variance decomposition: contribution of dependency rate shocks

Horizon

DEP

RGDP

SAV

CUR

RIR

INV

5 年
英石. Error

10 年
英石. Error

15 年
英石. Error

25 年
英石. Error

94.8
2.6

81.4
7.0

68.0
10.3

52.8
13.0

4.4
3.0

8.4
5.9

11.5
8.6

14.1
9.9

2.3
1.8

8.1
5.5

12.7
8.4

14.8
9.0

0.9
1.0

3.3
3.0

7.4
6.3

10.7
8.3

1.4
1.2

3.5
2.3

5.8
3.7

7.2
4.6

2.2
1.9

3.6
3.1

4.8
3.7

7.1
5.3

Lee found that real GDP, saving rate, and investment rate decrease in response to a
positive shock to dependency rate, and that the current account also decreases be-
cause saving rate decreases more than investment rate. 然而, the response of the
real interest rate is qualitatively different. The real interest rate does not change
much in G-7 countries, but the real interest rate increases signiªcantly in East Asian
国家. G-7 countries have been more open to international ªnancial markets
than East Asian countries, which may explain the differences in the results.

最后, we report the forecast error variance decomposition results in Table 1 to in-
fer the role of the dependency rate shocks when explaining ºuctuations of macro-
economic variables. For the long horizons, the dependency rate shocks explain sub-
stantial parts of real GDP, saving, and current account ºuctuations. The dependency
rate shocks explain over 14 percent of saving rate and real GDP ºuctuations and
超过 10 percent of current account ºuctuations for the 25-year horizon.

6.2 Results of the extended model
数字 10 reports the impulse responses of other macro variables of interests for the
extended model. As discussed in section 5, an additional variable is added in the ba-
sic model. The responses of the private saving rate (PRIV SAV, percent of GDP),
government saving rate (GOV SAV, percent of GDP), real exchange rate (RER), 真实的
government consumption per capita (GOV CONS), and real private consumption
per capita (PRIV CONS) are reported. Notice that the responses of the private sav-
ing rate and government saving rates are in terms of percentage point changes,
whereas the additional variables are in terms of percentage changes.9

Both private and government saving decrease. The decreases are not different from
zero with 95 percent probability, but they are at a 6- and 7-year horizon with at least

9 Note that the impulse responses of the original ªve variables in the extended models are sim-

ilar to those in the basic ªve-variable model.

Asian Economic Papers

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

6
2
2
2
1
6
8
2
3
4
2
A
s
e
p
2
0
0
7
6
2
2
2
p
d

乙
y
G
你
e
s
t

哦
n
0
7
S
e
p
e
米
乙
e
r
2
0
2
3

Demographic Changes, Saving, and Current Account

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

6
2
2
2
1
6
8
2
3
4
2
A
s
e
p
2
0
0
7
6
2
2
2
p
d

乙
y
G
你
e
s
t

哦
n
0
7
S
e
p
e
米
乙
e
r
2
0
2
3

我
e
d
哦
米
d
e
d
n
e
t
X
e

:
s
k
C
哦
H
s

e
t
A
r
y
C
n
e
d
n
e
p
e
d
哦
t

s
e
s
n
哦
p
s
e
r

e
s
我
你
p
米

我

.
0
1

e
r
你
G
我
F

Asian Economic Papers

Demographic Changes, Saving, and Current Account

84 percent probability. The maximum declines are found in about 10 年. The max-
imum decline of private saving is approximately 0.22 percentage point whereas
that of government saving is approximately 0.17 percentage point. Real government
consumption increases over time, and the maximum increase of approximately
0.6 percent is found in about 11 年. As dependency rate increases, 政府
needs to spend more to support dependents, which may lead to an increase in
government consumption. This increase in government consumption, 反过来, leads
to the decrease in government saving. 另一方面, real private consumption
decreases sharply. The maximum decrease of approximately 1.8 percent will be seen
in about 10 年.

These results are somewhat different from the results of G-7 in Kim and Lee (2007).
In G-7 countries, the saving decline is mostly driven by the decline in government
saving (instead of the decline in private saving). This may be related to the social
security system. In G-7 countries that have a good social security system, the gov-
ernment can take over the burden of decline in income due to an increase in the de-
pendency rate, which may result in a decline in government saving but no changes
in private saving. 另一方面, in many East Asian countries, which do not
have a good social security system, the government cannot take over the entire bur-
这, and as a result, private saving also declines.

最后, the real exchange rate tends to appreciate, which may be explained by the
increase in the real interest rate. This result is also different from the long run depre-
ciation results of G-7 countries found in Kim and Lee (2005).

数字 11 reports impulse responses in the extended model that both the elderly de-
pendency rate and the youth dependency rate are included in the model, by replac-
ing the total dependency rate. The ªrst and the second columns show the responses
to the elderly dependency rate shocks and the youth dependency rate shocks, 关于-
spectively, 在模型中 {DEPO, DEPY, RGDP, SAV, CUR, RIR} and the third and the
fourth columns show the responses to the youth dependency rate shocks and the el-
derly dependency rate shocks, 分别, in the model of {DEPY, DEPO, RGDP,
SAV, CUR, RIR}. In addition to the responses of the six variables, we also report the
responses of the total dependency rate by adding the responses of the elderly de-
pendency rate and the youth dependency rate. The responding variables are de-
noted at the far left with the name of shocks and the model names reported at the
顶部. In order to compare easily the four cases, the same scale is applied to all graphs
in the same row.

Elderly dependency rate shocks have stronger effects on most variables than youth
dependency rate shocks. Elderly dependency rate shocks substantially decrease real

Asian Economic Papers

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

6
2
2
2
1
6
8
2
3
4
2
A
s
e
p
2
0
0
7
6
2
2
2
p
d

乙
y
G
你
e
s
t

哦
n
0
7
S
e
p
e
米
乙
e
r
2
0
2
3

Demographic Changes, Saving, and Current Account

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

6
2
2
2
1
6
8
2
3
4
2
A
s
e
p
2
0
0
7
6
2
2
2
p
d

乙
y
G
你
e
s
t

哦
n
0
7
S
e
p
e
米
乙
e
r
2
0
2
3

我
e
d
哦
米
d
e
d
n
e
t
X
e
:
s
k
C
哦
H
s
e
t
A
r
y
C
n
e
d
n
e
p
e
d
H
t
你
哦
y
d
n
A
e
t
A
r
y
C
n
e
d
n
e
p
e
d
y
我
r
e
d
我
e
哦
t

s
e
s
n
哦
p
s
e
r

e
s
我
你
p
米

我

.
1
1

e
r
你
G
我
F

Asian Economic Papers

Demographic Changes, Saving, and Current Account

GDP, saving, and the current account. The point estimate shows that over approxi-
马特利 20 年, an increase of the elderly dependency rate by approximately
0.25 percentage point is associated with a decrease of saving rate by 0.8 percentage
point and current account balance by 0.8 percentage point. 另一方面, 青年
dependency rate shocks have only small effects on those variables. These results are
quite different from the results of G-7 countries in Kim and Lee (2005). The elderly
dependency rate shocks mostly drive the changes in the real GDP, saving rate, 和
the current account in East Asian countries whereas both elderly and youth depend-
ency rate shocks are responsible for the changes in those variables. This result seems
to be partly due to the persistence of the shocks; elderly dependency rate shocks are
more persistent than youth dependency rate shocks. 此外, the results may be
related to different cultural and economic environments. 例如, income dur-
ing retirement is low in East Asia, which may lead to a larger effect of elderly de-
pendency rate on private saving. As another example, parents in East Asia may be
more eager to save money for their children, which may lead to a smaller effect of
youth dependency rate on private saving.

7. Concluding remarks

This paper examined the effects of demographic changes on saving and current ac-
count in 10 East Asian economies between 1981 和 2003. The panel VAR estima-
tions show that an increase in dependency rate has a strong association with saving
decline and current account deterioration. Our ªnding suggests that the rapidly ag-
ing population and subsequent rise in dependency rates in East Asia over the next
decade might cause substantial deterioration in saving rates and current account.
然而, the precise magnitude of a future impact would depend on the adjust-
ments that private agents and national governments could make in anticipation of
the demographic changes, 例如, an increase in private saving, pension re-
形式, a migration to more productive countries, and extension of retirement age.
尤其, East Asian economies would eventually have to face the two problems
that are now confounding the advanced countries: how to reform the pension
schemes and long-term health-care systems. The East Asian adjustments to these
demographic changes will change future saving rates, especially future public sav-
ing rates, and future current account balances.

From a global perspective, such aging-related pressures on the current accounts in
East Asia might be at least partly offset by the favorable effect of the increased
working population in other regions. 全球范围, the net capital inºows in East Asia as
well as advanced countries should be met by the net capital outºows in other devel-
oping countries. 然而, because the pace and timing of population aging will

Asian Economic Papers

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

6
2
2
2
1
6
8
2
3
4
2
A
s
e
p
2
0
0
7
6
2
2
2
p
d

乙
y
G
你
e
s
t

哦
n
0
7
S
e
p
e
米
乙
e
r
2
0
2
3

Demographic Changes, Saving, and Current Account

vary much among countries, the consequences of global demographic changes on
international capital markets would be more complicated. It might happen that fast-
aging nations decrease the supply of global saving substantially below global in-
vestment demand in the world. This shortage of global supply of loanable funds
would cause a higher cost of capital and subsequent drop in investment and real
输出. 清楚地, much interesting research remains to be done in future research on
the impact of demographic changes on current account imbalances across countries
and regions.

参考

Banks, James, Richard Blundell, and Sarah Tanner. 1998. Is There Retirement-Savings Puzzle?
American Economic Review 91(4), 769–788.

Batini, Nicoletta, Tim Callen, and Warwick McKibbin. 2006. The Global Impact of Demo-
graphic Change. IMF Working Paper No. wp/06/9. 华盛顿, 直流: IMF.

Bernheim, D ., J. Skinner, 和S. Weinberg. 2001. What Accounts for the Variations in Retire-
ment Wealth among U.S. Households? American Economic Review 91, PP. 832–857.

Bosworth, Barry, Ralph C. Bryant, and Gary Burtless. 2004. The Impact of Aging on Financial
Markets and the Economy: 调查. Working Paper No. CRR WP 2004–23. 波士顿: Center for
Retirement Research at Boston.

布鲁克斯, 罗宾. 2003. Population Aging and Global Capital Flows in a Parallel Universe. IMF
Staff Papers 50 (2):200–221.

Bryant, Ralph C. 2004. Cross-Border Macroeconomic Implications of Demographic Change.
Brookings Discussion Papers in International Economics No. 166. 华盛顿, 直流: 这
布鲁金斯学会.

卡罗尔, Christopher D. 1997. Buffer Stock Saving and the Life Cycle/Permanent Income Hy-
pothesis. Quarterly Journal of Economics 112:1–55.

Chinn, Menzie D., and Eswar S. Prasad. 2003. Medium-term Determinants of Current Account
in Industrial and Developing Countries: An Empirical Exploration. Journal of International Eco-
经济学 59 (1):47–76.

Christiano, L。, 中号. Eichenbaum, 和C. 埃文斯. 1999. Monetary Policy Shocks: What Have We
Learned and to What End? In Handbook of Macroeconomics, 卷. 1A, edited by John B. 泰勒
and Michael Woodford, PP. 65–148. 阿姆斯特丹: North-Holland.

Deaton, Angus, and Christina Paxson. 1997. The Effects of Economic and Population Growth
on National Saving and Inequality. Demography 34:97–114.

Dekle, 罗伯特. 2004. Financing Consumption in an Aging Japan: The Roles of Foreign Capital
Inºows and Immigration. NBER Working Paper No. 10781. 剑桥, 嘛: 国家局
经济研究部.

Demery, D ., and N. 瓦. Duck. 2006. Savings-age Proªles in the U.K. Journal of Population
Ecomomics 19 (3), PP. 521–541.

爱德华兹, Sebastian. 1995. Why Are Saving Rates so Different Across Countries? An Interna-

Asian Economic Papers

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

6
2
2
2
1
6
8
2
3
4
2
A
s
e
p
2
0
0
7
6
2
2
2
p
d

乙
y
G
你
e
s
t

哦
n
0
7
S
e
p
e
米
乙
e
r
2
0
2
3

Demographic Changes, Saving, and Current Account

tional Comparative Analysis. NBER Working Paper No. 507. 剑桥, 嘛: 国家局
经济研究部.

Eichengreen, Barry. 2005. The Blind Men and the Elephant. Paper presented at The Tokyo Club
Foundation’s Macroeconomic Research Conference on the Future of International Capital
Flows, 21–22 November, Kyoto.

Faruqee, Hamid, and Martin Mühleisen. 2003. Population Aging in Japan: Demographic Shock
and Fiscal Sustainability. Japan and the World Economy 15:185–210.

Feroli, 迈克尔. 2003. Capital Flows among the G-7 Nations: A Demographic Perspective, Fi-
nance and Economics Discussion Series 2003-54. 华盛顿, 直流: Board of Governors of the Fed-
eral Reserve System.

Helliwell, 约翰·F. 2004. Demographic Changes and International Factor Mobility. NBER
Working Paper No. 10945. 剑桥, 嘛: 国家经济研究局.

希金斯, 马修. 1988. Demography, National Saving, and International Capital Flows. 国际米兰-
national Economic Review 39:343–369.

Horioka, 查尔斯. 1992. Future Trends in Japan’s Saving Rate and the Implications Thereof for
Japan’s External Imbalance. Japan and the World Economy 3:307–330.

International Monetary Fund (IMF). 2004. World Economic Outlook. 九月. 华盛顿,
直流: International Monetary Fund.

Jappelli, Tullio, and Franco Modigliani. 1998. The Aging-Saving Proªle and the Life-Cycle Hy-
pothesis. CSEF Working Paper No. 09. 萨勒诺, 意大利: Centre for Studies in Economics and Fi-
南斯.

Kim, Soyoung, and Jong-Wha Lee. (2007). Demographic Changes, Saving, and Current Ac-
数数: An Analysis of Panel VAR Model. Japan and the World Economy forthcoming.

Leff, 氮. H. 1969. Dependency Rates and Saving Rates. American Economic Review 59, PP.
886–896.

刘, Liang-Yn, and Wing Thye Woo. 1994. Saving Behavior under Imperfect Financial Markets
and the Current Account Consequences. The Economic Journal 104 (可能):512–527.

Luhrmann, Melanie. 2003. Demographic Change, Foresight and International Capital Flows.
Mannheim Research Institute for the Economics of Aging? MEA Discussion Paper Series No.
03038. University of Mannheim, 德国.

McKibbin, Warwick. 2005. Global Demographic Change and Japanese Macroeconomic Perfor-
曼斯. Brookings Discussion Papers in International Economics No. 169. 华盛顿, 直流: 这
布鲁金斯学会.

McKibbin, W., 和 J. 阮. 2004. Modelling Global Demographic Change: Results for Japan.
Paper presented at the Conference on the International Collaborations Project, 东京.

Masson, Paul R., Tamim Bayoumi, and Hossein Samiei. 1998. International Evidence on the
Determinants of Private Saving. World Bank Economic Review 12:483–501.

Modigliani, Franco. 1970. The Life-Cycle Hypothesis of Saving and Intercountry Differences in
Saving Ratios. In Induction, 生长, and Trade, edited by W. A. Eltis, PP. 197–225. 伦敦: Clar-
endon Press.

Asian Economic Papers

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

6
2
2
2
1
6
8
2
3
4
2
A
s
e
p
2
0
0
7
6
2
2
2
p
d

乙
y
G
你
e
s
t

哦
n
0
7
S
e
p
e
米
乙
e
r
2
0
2
3

Demographic Changes, Saving, and Current Account

西姆斯, 克里斯托弗·A. 1980. Macroeconomics and Reality. Econometrica 48:1–48.

西姆斯, 克里斯托弗·A. 1988. Bayesian Skepticism on Unit Root Econometrics. Journal of Economic
Dynamics and Control 12:463–474.

西姆斯, Christopher A., and Harald Uhlig. 1991. Understanding Unit Rooters: A Helicopter Tour.
Econometrica 59:1591–1599.

韦尔, David N. 1994. The Saving of the Elderly in Micro and Macro Data. Quarterly Journal of
经济学 109:55–81.

我

D
哦
w
n
哦
A
d
e
d

F
r
哦
米
H

t
t

:
/
/

d
我
r
e
C
t
.

米

我
t
.

e
d
你
A
s
e
p
A
r
t
我
C
e
–
p
d

我

F
/

6
2
2
2
1
6
8
2
3
4
2
A
s
e
p
2
0
0
7
6
2
2
2
p
d

乙
y
G
你
e
s
t

哦
n
0
7
S
e
p
e
米
乙
e
r
2
0
2
3

Asian Economic Papers
下载pdf