Description
The rebalancing of investments is the action of bringing a portfolio that has deviated away from one's target asset allocation back into line. Under-weighted securities can be purchased with newly saved money; alternatively, over-weighted securities can be sold to purchase under-weighted securities.
FINANCIAL STUDY ON REBALANCING GROWTH IN CHINA: AN
INTERNATIONAL PERSPECTIVE
TABLE OF CONTENTS
Non-technical summary . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4Résumé non
technique . . . . . . . . . . . . . . . . . . . . . . . . . . .
5Résumé court
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
61. Introduction
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
72. China's
saving rate under scrutiny . . . . . . . . . . . . . . . . . . . . . 10
3. The model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3.1. General overview . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3.2. Households . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3.3. Production, investment and nominal rigidities . . . . . . . . . . . . . . . 15
3.4. Government . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.5. Trade . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
4. Structural policies in China . . . . . . . . . . . . . . . . . . . . . . . . 19
4.1. A pension reform . . . . . . . . . . . . . . . . . . . . . . . . . . 19
4.2. Other structural reforms . . . . . . . . . . . . . . . . . . . . . . . 23
4.3. Structural reforms in China under a ZIRP in the US . . . . . . . . . . . . 25
5. The impact of a monetary reform in China . . . . . . . . . . . . . . . . . . 26
5.1. A Pension "reverse" shock . . . . . . . . . . . . . . . . . . . . . . 28
5.2. Reserve accumulation shock . . . . . . . . . . . . . . . . . . . . . . 30
6. A third country: the Euro area . . . . . . . . . . . . . . . . . . . . . . . 32
6.1. Impact of a pension reform in China in a three-country setting . . . . . . . . 33
6.2. Impact of a diversification of China's official reserves . . . . . . . . . . . . 34
6.3. Impact of a diversification of China's official reserves under a ZIRP . . . . . . 34
7. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
A. Details of the model . . . . . . . . . . . . . . . . . . . . . . . . . 41
B. Simulations of alternative structural reforms in China . . . . . . . . . . . . 45
C. Robustness of the model . . . . . . . . . . . . . . . . . . . . . . . 49
List of working papers released by CEPII . . . . . . . . . . . . . . . . . . . . 51
2
N O N - T E C H N I C A L S U M M A RY
Since the early 2000s, China has been the center of repeated criticism by the international community
concerning its export-oriented growth strategy based on an artificially weak currency. However China has
responded to this criticism through structural reforms rather than monetary reforms. This paper aims at
analyzing the complementarity or substitutability of structural and monetary reforms in China, for the
domestic economy as for the global rebalancing agenda. Specifically, we study the impact of a several
reforms that all involve a reduction in Chinese saving rate, under various regimes for capital ?ows and the
exchange rate; reciprocally, we analyze the rebalancing impact of a change in China's monetary regime, absent
structural reforms, hence in a situation of cumulated surpluses in China.
To this end, we build a micro-founded macroeconomic model with overlapping generations, nominal
rigidities and financial constraints, for two countries: China, and the United States. The structural re- forms
successively considered are an increase in the generosity of China's pension system, a financial reform that
facilitates the access of private companies to debt markets while removing capital subsidies, and an increase
in the government debt target. As for monetary regimes, they include the status quo (a fixed exchange rate
backed by capital controls and foreign-exchange interventions), a relaxation of capital controls, and a move
to a ?exible exchange-rate regime. We successively consider the impact of a structural reform under
different monetary regimes, and the impact of a monetary regime change in itself, in a context of global
imbalances. Finally, we introduce a third country (the Euro area) in the model in order to study the
asymmetries stemming from China's bilateral peg on the dollar and from its preference for dollar-denominated
assets.
We find that a fall in China's saving rate would contribute to global rebalancing whatever the exchange rate
regime, provided international capital ?ows do react to interest-rate differentials (which implies less-than-
complete capital controls in China). Under a ?exible exchange-rate regime, the reduction of the bilateral
current-account imbalance between the United States and China is quicker (because the renminbi
appreciates in nominal terms against the US dollar), but it is not stronger than under a fixed exchange-rate
regime. Still, only under a ?exible exchange-rate regime is China able to control in?ation stemming from more
dynamic domestic demand. This feature provides strong incentive for China to move away from its fixed
peg, although the advantage for the United States of such move is only short lived.
We also find that, should the United States refrain from hiking its interest rate when global savings are
reduced, then the rebalancing would be muted and the exchange-rate regime of China would become crucial
both for global rebalancing and for the domestic Chinese economy. As for the Euro area, despitethe
asymmetric exchange-rate regime of China (where the renminbi is pegged on the dollar), it reacts in a
similar way as the United States to structural reforms in China. The reason is high capital mobility
3
between the United States and the Euro area, that tends to equalize real interest rates between the two
countries.
Absent structural reforms reducing the aggregate saving rate of China, a move to a ?exible renminbi is
found unable to rebalance the global economy, except if the People's Bank of China decides to sell off its
accumulated reserves. The reason is that high savings in China, combined with capital controls, result in a
relatively low policy interest rate: removing capital controls while allowing the Chinese currency to ?oat
could then result in a sudden capital out?ows and depreciation of the renminbi in the short run, although in the
longer run the Chinese currency would appreciate.
On the whole, our results suggest that the contribution of China to global rebalancing should primarily rely
on structural policies aiming at reducing aggregate savings in China and on a relaxation of capital controls.
The role of the exchange rate regime would be minor under standard monetary policies in advanced
economies, although more important if monetary policies are constrained. Finally, relying only on a
change in China's monetary regime could end up in delaying rather than accelerating the rebalancing,
depending on China's policy regarding accumulated reserves.
A B S T R AC T
Based on simulations of an original DGE model of the US, Chinese and Euro area economies with
financial frictions and various monetary regimes, the paper shows that the contribution of China in global
rebalancing should primarily rely on structural policies aiming at reducing aggregate savings in China. The
role of the exchange-rate regime would be minor under standard monetary policies, although more important if
monetary policies in advanced countries are constrained, as they are today. Finally, relying only on a change
in China's monetary regime (without structural reforms) could end up in delaying rather than accelerating the
rebalancing, depending on China's policy regarding accumulated reserves.
JEL Classification: F32, F42, F47.
Keywords: Global imbalances, exchange rate regimes, capital controls, China.
4
RÉQUILIBRAGE DE LA CROISSANCE EN CHINE: UNE PERSPECTIVE INTERNATIONALE
RÉSUME NON TECHNIQUE
Depuis le début des années 2000, la Chine s'est trouvée au centre de critiques répétées, de la part de la
communauté internationale, au sujet de sa stratégie de croissance extravertie fondée sur une mon- naie
sous-évaluée. La Chine a répondu à ces critiques par des réformes structurelles plutôt que mo- nétaires.
Nous étudions ici dans quelle mesure les deux types de réformes sont complémentaires ou substituables,
pour l'économie chinoise et pour le rééquilibrage de l'économie mondiale. Plus spécifi- quement, nous
étudions l'impact de différentes réformes structurelles qui toutes conduisent à une baisse du taux d'épargne
agrégé en Chine, selon différents régimes en matière de ?ux de capitaux et de taux de change. A l'inverse, nous
analysons l'impact d'une réforme monétaire sur les déséquilibres mondiaux,
en l'absence de réformes structurelles, donc dans une situation d'accumulation d'excédents en Chine.
A cette fin, nous construisons un modèle à génération imbriquées avec fondements microéconomiques,
rigidités nominales et contraintes financières, pour deux pays : la Chine et les États-Unis. Les réformes
structurelles successivement envisagées sont une augmentation de la générosité du système de retraite par
répartition, une réforme financière permettant aux entreprises un accès plus facile aux financements externes,
les subventions au capital étant simultanément supprimées, et, enfin, un relèvement de l'objec- tif de dette
publique. Trois régimes monétaires sont successivement étudiés : le statu quo (taux de change fixe, contrôles de
capitaux, interventions de change), un relâchement des contrôles de capitaux et un ré- gime de ?ottement sans
contrôles de change. On étudie successivement l'effet des réformes structurelles selon les trois régimes
monétaires, puis l'effet d'une réforme monétaire seule, dans un contexte d'ex- cédents accumulés. Enfin, on
introduit un troisième pays (la zone euro) dans le modèle afin d'analyser les asymétries découlant de l'ancrage
bilatéral du renminbi sur le dollar et de la préférence de la banque centrale pour les réserves en dollars.
Les résultats montrent que, quelque soit le régime de change, une baisse du taux d'épargne en Chine est de
nature à contribuer au rééquilibrage mondial, pourvu que les ?ux internationaux de capitaux réagissent aux
écarts de taux d'intérêt, ce qui suppose un relâchement des contrôles de capitaux en Chine. En régime de
change ?exible, la réduction du déséquilibre bilatéral entre la Chine et les États-Unis est plus rapide (grâce à
l'appréciation nominale du renminbi par rapport au dollar) mais elle n'est pas plus marquée qu'en change
fixe. Cependant, seul le régime de change ?exible permet à la Chine de contrôler son taux d'in?ation face à la
hausse de la demande interne. Cette particularité constitue une puissante incitation, pour la Chine, à
abandonner le régime de change fixe, alors que les bénéfices de la ?exibilité sont de courte durée pour les
Etats-Unis. Cependant, si les Etats-Unis rechignent à relever leur taux d'intérêt lorsque l'épargne mondiale
diminue, alors le rééquilibrage mondial est faible et le régime de change en Chine en devient un élément
essentiel. En dépit de l'asymétrie du régime de change chinois (le renminbi est ancré sur le dollar, non sur
l'euro), la zone euro réagit de façon très similaire aux Etats-Unis à une réforme en Chine. La raison en est la
forte mobilité des capitaux entre Etats-Unis et zone euro, qui tend à égaliser les taux d'intérêt réels dans les
deux zones.
5
En l'absence de réformes structurelles réduisant le taux d'épargne en Chine, le passage au ?ottement ne
permet pas, en lui-même, de réaliser le rééquilibrage mondial, sauf si la Banque du Peuple Chinois décide de
vendre ses réserves accumulées. En effet, le haut niveau de l'épargne en Chine, combiné aux contrôles de
capitaux, conduit à un taux d'intérêt relativement faible dans ce pays. Dans ces conditions, la levée des
contrôles de capitaux concomitante au passage au ?ottement conduit à une sortie nette decapitaux, donc à
une dépréciation du renminbi à court terme, même si à plus long terme la monnaie chinoise s'apprécie bien.
Finalement, nos résultats suggèrent que la contribution de la Chine au rééquilibrage mondial devrait re- poser
essentiellement sur des réformes structurelles réduisant le taux d'épargne agrégé de ce pays et sur un
relâchement des contrôles de capitaux. Le rôle du régime de change est secondaire lorsque les poli- tiques
monétaires des pays avancés sont standard, même s'il devient plus important lorsque les politiques monétaires
sont contraintes. Enfin, fonder le rééquilibrage mondial seulement sur une modification du régime de change
chinois pourrait retarder plutôt qu'accélérer le rééquilibrage, selon la politique suivie par la Chine quant au
devenir de ses réserves accumulées.
R É S U M É C O U RT
A partir d'un modèle dynamique d'équilibre général à trois pays (Etats-Unis, Chine, Zone euro) avec frictions
financières et plusieurs régimes monétaires possibles, nous montrons que la contribution de la Chine au
rééquilibrage mondial devrait reposer principalement sur des réformes structurelles ayant pour effet une
baisse du taux d'épargne agrégé en Chine et de relâcher les contrôles de capitaux. Le rôle du régime de
change est secondaire en présence de politiques monétaires standard dans les pays avancés mais devient
plus important si les politiques monétaires des pays avancés sont contraintes par un taux d'intérêt plancher.
Enfin, miser sur une simple modification du régime monétaire chinois (sans réformes structurelles) pourrait
retarder au lieu d'accélérer le rééquilibrage, selon la politique suivie au regard des réserves accumulées.
Classification JEL : F32, F42, F47.
Mots clés : Déséquilibres mondiaux, régime de change, contrôle des capitaux, Chine.
6
-
1. I N T RO D U C T I O N
Since the early 2000s, China has been the center of repeated criticism by the international com-
munity concerning its export-oriented growth strategy based on an artificially weak currency. In
2005, Ben Bernanke famously pointed the aggregate trade surplus of emerging countries as one
key explanation for the growing US deficit: corresponding capital in?ows in the US was preventing
upward adjustment of US savings rates through maintaining real interest rates at a low level.
2
Although few economists and officials did endorse Bernanke's "savings glut" ap- proach in full
(Bernanke himself recognized the own responsibility of the United States), the G7 consistently
called for a reduction in current-account surpluses especially in emerging Asia, and its leaders
queued up in Beijing to try to persuade Chinese officials to allow the renminbi to appreciate.
Global imbalances temporary retreated during the 2007-2009 global crisis, as a consequence of
the trade collapse and financial de-globalization. When the global economy started to pick up in
mid-2009, it however became clear that the crisis had not wiped up the problem which was
deemed to soon re-emerge (see Blanchard & Milesi-Ferretti (2010)). The G20, which at that time
had replaced the G7 as the main forum for international coordination, refocused the debate on current-
account imbalances themselves, as opposed to the G7 insistence on exchange rates.
3
China
responded by announcing far-reaching structural reforms aiming, in particular, at im- proving
social safety nets, which would allow Chinese households to reduce their savings rates. However
China failed to satisfy advanced economies through letting its currency appreciate, despite
continuous pressure.
1
We are grateful to the participants of the 7 december 2010 at CEPII and the 8 March 2011 RES seminar at the IMF,
for useful comments. All errors remain ours. This paper has been prepared for the Bruegel-CEPII research project
"Reform options for the global reserve currency system and their implications for the EU economy" for the European
Commission (contract No. ECFIN/220/2010/573686). Financial support by DG ECFIN of the European Commission is
gratefully acknowledged.
7
This paper aims at analyzing the complementarity or substitutability of structural and monetary
reforms in China, for the domestic economy as for the global rebalancing agenda. Specifically, we
study the impact of a several reforms that all involve a reduction in Chinese saving rate, under
various regimes for capital ?ows and the exchange rate; reciprocally, we analyze the rebalancing
impact of a change in China's monetary regime, absent structural reforms, hence in a situation of
cumulated surpluses in China.
Our analysis follows-up on the pre-crisis literature on global imbalances and adjustment sce-
narios. Two groups of explanations have been suggested to global imbalances (Bernanke, 2005;
Dooley et al., 2003; Mendoza et al., 2007; Caballero et al., 2008): (i) macroeconomic and
structural factors, such as high saving rates in emerging countries triggered by energy wind- falls,
deficient social safety nets, or financial under-development, low investment rates in some emerging
countries following the Asian crisis, or lax monetary policy in the United States; and (ii) the
international monetary system (IMS hereafter) itself, through the key role of the United States as a
supplier of international reserve assets, the lack of trust in multilateral financial safety nets which led
emerging countries especially in Asia to self-insure through reserve accumula- tion, and the success
of export-oriented growth strategies that encouraged fixed or quasi-fixed exchange-rate regimes.
Contrasting with this two-pillar understanding of global imbalances, researches on rebalancing
scenarios have tended to focus on real exchange-rate adjustment, es- pecially for the United States,
abstracting from the IMS (Obstfeld & Rogoff, 2005; Blanchard et al., 2005). Few studies have
been devoted to the interconnection between real adjustments and monetary regimes. One
exception is Blanchard and Giavazzi (2006) mentioning the needs to combine a nominal
appreciation of the renminbi with a fall in China's saving rate, in order to secure internal balance
while reducing the current-account surplus. Another one is Faruqee et al. (2007) who contrast the
impact of a fiscal adjustment in the US depending on China's monetary regime, within a four-
region dynamic general equilibrium (DGE) model.
The present paper is interested in the interconnection between global rebalancing and monetary
regimes. The main case in point being the Chinese economy, where excess savings result in
official reserve accumulation due to the fixed exchange-rate regime, we choose to focus on this
country, while acknowledging that China is by no way the only contributor to global imbal-
ances. According to Prasad (2009), China accounted for 62 percent of gross national savings in
Asia excluding Japan in 2008, which provides an additional motivation for focusing on Chi- nese
savings. We start by brie?y reviewing the literature that has tried to explain the very high gross
saving rate in China (50% of GDP on average between 2005 and 2008, according to Ma & Yi
(2010)). We then build a DGE model of two countries (China and the United States) with
overlapping generations, nominal rigidities and financial constraints, and simulate differ- ent
structural reforms in China that all result in declining saving rates, under different monetary regimes.
The reforms successively considered are an increase in the generosity of China's pen- sion system, a
financial reform that facilitates the access of private companies to debt markets, and an increase in the
government debt target. As for monetary regimes, they include the status quo (a fixed exchange rate
backed by capital controls and foreign-exchange interventions), a
8
relaxation of capital controls, and a move to a ?exible exchange-rate regime. We successively
consider the impact of a structural reform under different monetary regimes, and the impact of a
monetary regime change in itself, in a context of global imbalances. Finally, we introduce a third
country (the Euro area) in the model in order to study the asymmetries stemming from China's
bilateral peg on the dollar and from its preference for dollar-denominated assets.
We find that a fall in China's saving rate would contribute to global rebalancing whatever the
exchange rate regime, provided international capital ?ows do react to interest-rate differentials (which
implies less-than-complete capital controls in China). Under a ?exible exchange-rate regime, the
reduction of the bilateral current-account imbalance between the United States and China is quicker
(because the renminbi appreciates in nominal terms against the US dollar), but it is not stronger than
under a fixed exchange-rate regime. Still, only under a ?exible exchange- rate regime is China able to
control in?ation stemming from more dynamic domestic demand, consistent with the intuition of
Blanchard & Giavazzi (2006). This feature provides strong incentive for China to move away
from its fixed peg, although the advantage for the United States of such move is only short lived.
We also find that, should the United States refrain from hiking its interest rate when global
savings are reduced, then the rebalancing would be muted and the exchange-rate regime of China
would become crucial both for global rebalancing and for the domestic Chinese economy. As for the
Euro area, despite the asymmetric exchange-rate regime of China (where the renminbi is pegged on
the dollar), it reacts in a similar way as the United States to structural reforms in China. The reason
is high capital mobility between the United States and the Euro area, that tends to equalize real
interest rates between the two countries.
Absent structural reforms reducing the aggregate saving rate of China, a move to a ?exible
renminbi is found unable to rebalance the global economy, except if the People's Bank of China decides
to sell off its accumulated reserves. The reason is that high savings in China, combined with capital
controls, result in a relatively low policy interest rate: removing capital controls while allowing the
Chinese currency to ?oat could then result in a sudden capital out?ows and depreciation of the
renminbi in the short run, although in the longer run the Chinese currency would appreciate.
On the whole, our results suggest that the contribution of China to global rebalancing should
primarily rely on structural policies aiming at reducing aggregate savings in China. The role of the
exchange-rate regime would be minor under standard monetary policies, although more important if
monetary policies in advanced countries are constrained. Finally, relying only on a change in
China's monetary regime could end up in delaying rather than accelerating the rebalancing,
depending on China's policy regarding accumulated reserves.
The remaining of the paper is organized as follows. In Section 2, we review the literature on the
determinants of China's high savings rate. Section 3 presents the model. In Section 4, we
simulate structural reforms in China under different monetary regimes. Section 5 studies
9
Figure 1 - Saving
a
rates
e
in
0
selected countries
Saving, verag 2 05-2007
50
40
30
20
10
Gov. 6,40
0
-10
Source: Ma & Yi (2010)
the impact of a monetary reform in China after the latter has accumulated a large amount of
official reserves. Section 6 adds the Euro area to the analysis and discusses the impact of a
diversification of China's official reserves. Section 7 concludes.
2. CHINA'S SAVING RATE UNDER SCRUTINY
The very high saving rate in China has attracted much attention from both policy-makers and
researchers. Ma & Yi (2010) note that gross national savings reached 54 percent of GDP in 2008,
up from 39 percent in 1990 and 36 percent in 2000. They argue that what is special in China is
the conjunction of high saving rates in the three sectors of the economy: households, corporations,
and the government (see Figure 1). Hence, explanations for the high level of the aggregate saving
rate must address the three sectors simultaneously in a consistent way.
Many explanations have been suggested for high saving rates in China.
4
They can be summa-
rized as follow:
• Households: The steady decline in the share of households in national income during the
2000s has been a major contribution to the fall in the consumption-to-GDP ratio and sub-
sequent rise in the saving-to-GDP one. It can be related to the downward pressure on labor
income (especially in farming and unincorporated activities), a reduction in social transfers, a low
level of capital income (low interest rates), and growing taxes and social contribu- tions. As
for the rise in households saving rates (as a percentage of disposable income), it
4
See, e.g., Ma & Yi (2010), Guo & N'Diaye (2010), Jha et al. (2009), Horioka & Wan (2007), Feng et al. (2009),
Chamon & Prasad (2010), Baker & Orsmond (2010), Cui & Aziz (2007), Chamon et al. (2010), Prasad (2009),
Bayoumi et al. (2010), Huang & Wang (2010).
10
H
o
u
s
e
h
o
l d
s
:
2
1
,
5
0
%
o
f
G
D
P
C
o
r
p
o
r
a
t
i o
n
s
:
2
0
,
4
0
U
S
T
a
i
w
a
n
P
h
i
l
F
r
a
n
c
e
G
e
r
m
a
n
y
J
a
p
a
n
K
o
r
e
a
I
n
d
i
a
C
h
i
n
a
is generally attributed to the attrition of the social safety nets (the retreat of enterprise-based
safety nets having not been taken over by the public system), together with a general rise in
uncertainty due to structural change and the rising share of education expenditures falling on
families. The rise in private home ownership and the 1997 pension reform (with a re- duction
in its generosity) have also played a role, in combination with the delayed effect of the one-
child policy (and the subsequent lack of support expected for the old age). All these factors
were amplified by financial underdevelopment, which has limited the access of households to bank
loans as well as the diversification opportunities for their assets. In con- trast, the impact of ageing
is generally downplayed by the literature, saving rates displaying an uncommon U-shape along
the life cycle.
• Corporations: as a mirror effect, corporations have reaped a growing share of national in-
come during the 2000s. Their high profitability has also come from subsidized inputs such as land,
energy and environment. State-owned enterprises were not required to pay dividends on their
profits before 2008. As for private firms, they may have faced financial constraints that have
obliged them to accumulate savings to self-finance their investments. Although corporations do
have seen their share in national income increase over the decade, the reality of excess savings
has however been challenged in the recent literature based on firm-level data.
• Government: the shares of the government in national income and national saving rose more
than those of the corporate sector in the 2000s. One major reason was the surge in tax receipts that
was not matched by equal increase in expenditures. In fact, government consumption declined
as a proportion of GDP, since the priority was (and local incentives were) given to public
investment and the building-up of public-pension assets (consistent with the switch decided in
1997 in favor of a partially-funded pension assets), and taxes were not much redistributed to
local governments that were in charge of social spending.
In the following, we propose a model that allows for some of the main causes of high saving
rates to be mitigated:
• A pension reform: according to Chamon et al. (2010), "rising income uncertainty and pen-
sion reforms can account for over half of the increase in the urban saving rate in China since the
mid-1990s" (p.1). Consistently, we simulate a pension reform that would increase the
generosity of the pay-as-you-go system. By extension, this simulation covers any reform of the
social safety net that would reduce uncertainty about the ability of households to cover "big
ticket" expenses (such as health care).
• A financial reform: much of the literature (see, e.g., Cui & Aziz (2007); or Ma & Yi (2010))
stresses the declining share of households' share in national GDP as a major cause of the rise in
China's aggregate saving rate. Consistently, we simulate a financial reform in China that relies on
three pillars: (i) enterprises get the opportunity to massively increase leverage, which raises the
value of the firms; (ii) simultaneously, the subsidy on corporate investment is removed; (iii) the
combination of higher firm value and lower incentive to invest leads
11
to a transfer of income from the corporate sector to the households one, possibly transiting
through a public body.
• A government spending reform: we finally assume a permanent increase in public spending
that is engineered through a rise in the public-debt target. Subsequent public transfers are
channelled to old households through the PAYG pension system. In the long run, the tax rate
increases to stabilize the debt ratio at its new in?ated target level.
These various simulations are run under different regimes in terms of exchange-rate regime
(either fixed or ?exible) and capital ?ows (restricted or freed), and the results are systematically
compared. It can be argued however that the problem today is less that of carrying out structural
reforms that will reduce China's saving rate and ultimately contribute to global rebalancing, than to
allow the exchange rate to correct accumulated disequilibria. Consistently, we run a reverse
simulation where a reduction in the generosity of the pension system (like after the 1997 reform) or an
increase in the official reserve target that both raise China's saving rate and trade balance, under fixed
exchange-rate regime. After a few years, the exchange rate is allowed to ?oat freely, and we study the
impact of the sole monetary reform on the curent account and the economy.
3. THE MODEL
3.1. General overview
The world is divided into two countries: the United-States (u) and China (c). In an extension of the
model, we include a third region: the Euro area (e). Countries trade both goods and financial assets.
The model explicitly takes into account the saving behavior of the three main economic
sectors: households, firms and the public sectors.
5
The model is depicted in Figure 2, and detailed in Appendix, together with its standard cali-
bration. Each country is populated with overlapping generations of households à la Blanchard
(1985) and Yaari (1965). On aggregate, households supply labor, receive wages and the pro-
ceeds of their savings (interests and dividends), pay taxes on their labor endowment and receive
transfers from the government. Depending on their age, the ?ow of funds differs: as they be- come
older, households accumulate financial wealth (hence they receive more dividend and interest
payments), their labor endowment decreases (labor income decreases), and they receive rising
transfers from the PAYG pension system. They also benefit from a complete set of contin- gent
securities (perfect risk sharing among alive households). In addition, when they die, their bequest
(financial assets) is redistributed to surviving households through an insurance scheme. The pension
system is introduced as the main policy tool that in?uences aggregate saving of households: the
higher the replacement rate, the lower the saving rate for a given real interestrate. As already
mentioned, this stylized pension system can account for any insurance scheme (e.g. health
insurance) that will reduce the incentive for precautionary savings.
5
We
do not distinguish the financial from the households sector: firms are assume to borrow directly from domes-
tic or foreign households.
12
There are two categories of firms. The first category hires labor L and capital K to produce goods
Y that are indifferently used for consumption and investment. These goods are differen- tiated and
sold to domestic and foreign customers under monopolistic competition. The firms face a price
rigidity à la Calvo (1983): at each period, each firm has a fixed probability of being unable to reset its
price. Wages are also sticky with staggered contracts.
The second category of firms specializes in accumulating capital K and renting it to the first
category of firms. Capital accumulation can be financed either by diverting dividends (internal
financing) or by borrowing Bf on the bond market (external financing). In China, firms' borrow- ings are
constrained by financial frictions. The saving behavior of capital firms is in?uenced by these
constraints: absent external financing, firms have to self-finance their investments, hence they pay low
dividends and their saving rate is high. With financial liberalization, firms can bor- row more, hence
they save less and distribute higher dividends. In China, the savings behavior
of capital firms is also in?uenced by public subsidies on capital.
6
The government receives contributions from households, positive or negative transfers from the
central bank, issues debt, provides subsidies to capital firms and pays pensions to old house- holds.
The pension level is determined by the government budget constraint, with exogenous debt target,
tax rate and, in the case of China, capital subsidies.
Finally, the central bank sets the interest rate according to a modified Taylor rule. In the United
States, the interest rate simply reacts to in?ation deviations from target.
7
In China, the rule is
augmented with an impact of official reserve accumulation. To (partially) sterilize its interven- tions,
the central bank issues sterilization bonds Bc. Depending on the amount of sterilization and on
return differentials between foreign-exchange reserves (which yield the foreign policy rate) and
sterilization bonds (which yield the Chinese policy rate), the central bank generates a positive or
negative profit that is immediately transferred to the pension system, hence indirectly transferred to old
households.
In the benchmark model, there are capital controls in China: foreign investors cannot buy as
many yuan-denominated assets as they would like, and Chinese investors cannot buy as many
foreign-denominated assets as they would like: the uncovered interest parity does not hold. This
grants the Chinese central bank some independence in the conduct of monetary policy even
when the exchange rate is fixed. Alternatively, capital controls can be relaxed, in which case UIP
does hold, impeding China from monitoring domestic in?ation unless the renminbi is allowed to ?oat.
The key equations are presented in the next sub-sections, the model being detailed in Appendix.
6
In our model, investment is restricted to the corporate sector. Alternatively, capital firms could be viewed as
intermediaries investing in real estate and renting their capital to the households. Aggregating households and capital
firms would then reproduce the impact of financial frictions on households' saving, in relation with housing.
7
Including the output-gap in the monetary rule does not change the results qualitatively, although it complicates the
resolution since the model then needs to be simulated at first with nominal rigidities in order to derive potential output.
13
Figure 2 - Diagram of the model
Bh *
B ond
market
FR
Central bank
(r,FR,Bc)
div
Pens. syst.
(tax,pens)
Bc tax pe ns
Bg
Households
Bh *
B ond
market
Bh
Households
(C,w)
I
Bk
Capital firms
(I,Bk)
K
div
Prod. Firms
L
Product
market
X,M Product
market
Y (p)
C
Foreign Home
3.2. Households
The representative household of the cohort born at time a and still living at time t consumes C
a,t
,
has a time-endowment for work of L
t
÷
a
hours, and effectively works L
a,t
.
8
At a given period, ¯
each previously-born household has a constant probabilityu
g
to survive, and new households
appear at rate (1 ÷u
g
)N , where N is the population, which thus is constant. Let 0<|< 1
¯ ¯
denote the subjective discount factor. The inter-temporal utility function of household a at time
t, U
a,t
, is increasing in consumption and in leisure:
·
U
a,t
=
s=0
(u
g
|)
s
(1÷k) log(C
a,t
+
s
) +k log(L
t
+s÷
a
÷ L
a,t
+
s
) , ¯ (1)
with 0<k< 1.
In each country, the portfolio A
t
of households consists in (i) domestic equities (ii) a complete
set of domestic contingent claims and (iii) foreign assets in the form of short-term bonds.
9
Life-
insurance companies à la Yaari (1965) pay a premium to alive households against the promise of
inheriting their assets when they are dead. We assume that insurance companies always accept
contingent claims and bonds (whether domestic or foreign), but are reluctant to insure
equities. The sharev
E
of equities eligible to life insurance is the first indicator of financial-
market development. It leads households to discount future dividends using a discount factor
8
The retirement schedule and associated pension are assumed inverse, geometric functions of the remaining life-
time labor endowment of the household.
9
The complete set of contingent claims are traded only among domestic households so that, when aggregating
households, the net holding is a domestic bond.
14
lowered by the factoru
E
=v
E
+ (1÷v
E
)u
g
< 1. Non-insured equities of dead households are assumed to
be distributed to the newborns. The budget constraint, the first-order conditions at
the cohort level, aggregate consumption and labor supply are detailled in Appendix A.1 (page 41).
The international portfolio of households consists in foreign bonds only: equities and contingent
claims are assumed not to be internationally tradable.
10
Imperfect capital mobility is taken into
account by assuming that domestic households of country j hold an amount Bh
jj
of foreign
bonds (of country j ) such that their share in the portfolio is proportional to the excess expected
return compared to the uncovered interest-rate parity:
E
jj
Bh
j
tj t
A
jt
Share in domestic portfolio
¸
jj
= 1 ÷¸
jj
E
t
jj
u
t
t+1
,j
R
j
t E
t
jj
E
t
+1
UIP deviation
÷1
(2)
where u
t
t+1
,j
is the stochastic discount factor of country j, R
j
t is the nominal interest rate on
foreign bonds, E
jj
is the nominal exchange rate between j and j , and 0<¸
jj
< 1 is a
proxy for capital mobility between the two countries. If¸ ÷ 0, the model reproduces capital controls:
even when the excess expected return is high, the ability of foreign households to buy domestic
assets (or of domestic households to sell foreign assets) is limited. On the contrary, ¸ ÷ 1 denotes
an absence of capital control as a tiny excess return motivates large capital ?ows between the two
countries. In the following, we assume that¸ may take one of the two following values:¸ = 0.5 (low
capital mobility) or¸ = 0.9 (high capital mobility).
3.3. Production, investment and nominal rigidities
Wage rigidities. Unions are used as a way to introduce monopolistic competition and wage
stickiness in the labor market. Specifically, each union is assumed to hire hours from households on a
competitive market. In turn, it rents these hours to the firms on a monopolistic competition market
where labor demand depends on the relative wage charged by the union with an elasticity
of substitution denoted
w
. Wages are sticky à la Calvo, with 1 ÷u
w
denoting the probability
for the union to be able to reset the wage level. The first-order condition on wage setting and
the derivation of aggregate-wage in?ation are derived in Appendix A.2 (page 42).
Production and price rigidities. The representative production firm i hires labor N
t
(i) from the
unions and capital K
t
(i) from capital firms to produce a differentiated good along a Cobb- Douglas
production function Y
t
(i) = A
t
K
t
(i)
o
N
t
(i)
1
÷
o
, where 0<o< 1 and A
t
denotes the
10
This assumption avoids perfect international risk-sharing. Since the stochastic discount factor will then be region- specific,
it is convenient to also assume that equities are not internationally traded in order to properly define the program of the
firms.
15
exogenous level of total factor productivity. The good is sold under monopolistic competition. The
elasticity of substitution of demand is denoted
p
. Calvo-type price stickiness is introduced
so that, at each period, the representative firm is able to reset its price with a probability 1 ÷
u
p
. Factor demand and optimal price setting and aggregate in?ation dynamics are derived in
Appendix A.3.
Investment. Capital accumulation is carried out by specialized firms that rent it to producing firms
on a competitive market. The level of capital K
t
(k) made available by a capital firm k at
the end of period t for the next production period depends on the capital stock at the end of pe-
riod t÷1 and on I
t
(k), the gross investment during period t: K
t
(k) = K
t
÷
1
(k)u
K
Itt÷(k()k
)
÷o , 1
whereo is the depreciation rate andu is a non-decreasing concave function withu(0) =
u (0) = 1 encompassing real rigidities on capital accumulation. Although the market of physi- cal capital
is perfectly competitive, capital firms can make profit due to entry barrier formed by their initial level
of capital.
Due to out-of-model financial frictions, the level of borrowing Bf
t
(k) (negative when actually
the firm borrows) is constrained by the expected value of the firm at the date the loan is paid
back (t+1). This leads to:
ç
÷Bf
t
(k) s
borrowing at t
1֍
E
t
u
t
t
+1
Vk
t
+1
(k)
expected value of the firm at repayment time
, (3)
where Vk
t
(k) is the market value of the capital firm and 0<ç< 1.
11
First-order conditions
on capital accumulation and the financial structure of the firm (debt or equities) are derived in
appendix A.4 (page 44).
3.4. Government
The role of the government is to run the PAYG pension system, distribute capital subsidies
(China only), and to conduct the monetary/exchange-rate policy.
Monetary and exchange-rate policies. In the United States (and, in the extension of the
model, in the Euro area), monetary policy is modeled through a simple interest-rate rule:
R
t
R
t
÷1
log =µ
R
log + (1 ÷µ
R
)o
t
log(t
t
), (4)
R
-
R
-
where R
t
is the nominal interest rate, R
-
its steady-state value,t
t
the production-price in?ation,
o
t
> 1 the long run reaction of interest rate to in?ation andµ
R
= 0.8 a smoothing parameter.
12
11
At
the steady-state,ç will be equal to the share of capital financed by external funds and so it is a second indicator
for financial development.
12
Interest rates and in?ation are expressed as one plus the rate itself.
16
Note that, consistent with the Taylor rule (Taylor (1993)), the real interest rate rises whenever
in?ation increases.
13
Since the model is not used to study cost-push shocks, we can safely drop
the output-gap from the Taylor rule, since the output-gap will be perfectly correlated with
in?ation.
14
This simple feedback rule is augmented in the case of China to account for the fixed exchange-
rate regime.
15
Specifically, the Chinese peg is maintained through foreign-exchange reserve
accumulation in the form of dollar-denominated bonds Bc
u
(and, in the three-country exten- t
sion of the model, of euro-denominated bonds Bc
e
t). These interventions are partially sterilized
through the issuance of sterilization bonds. Denoting by FR
t
> 0 the total amount of official
reserves at the end of period t and by Bc
c
t< 0 the stock of central bank liabilities in terms of
sterilization bonds, we have: Bc
c
t = ÷v
steril
FR
t
, where 0<v
steril
< 1 is the degree of steril-
ization of foreign-exchange interventions. According to Greenwood (2008), the sterilization of
official interventions in China during the 2000s was also channeled through a rise of reserve
requirements of commercial banks. Here the banking sector is not modeled, hence steriliza- tion
is performed only through the issuance of sterilization bonds that are directly (rather than indirectly)
purchased by households. Following Greenwood's evidence of a high degree of ster-
ilization, we setv
steril
= 0.9.
16
However, if China removes capital controls while maintaining its
peg, the amount of interventions becomes very large and it is unlikely to be widely sterilized.
In this case, we setv
steril
= 0.1.
To the extent that official interventions are not fully sterilized, they feed money creation, which
in turn exerts downward pressure on the interest rate. This effect is accounted for here through
adding a second term in the interest-rate rule for China:
R
t
R
t
÷1
log =µ
R
log + (1 ÷µ
R
) [o
t
log (t
t
) ÷ (1 ÷v
steril
)o
FR
(FR
t
÷ FR
-
)] , (5)
R
-
R
-
with FR
-
the target level for official reserves, ando
FR
> 0.
17
In order for the central bank's balance sheet to stabilize in the long run, we assume the cost
(or benefit) from official interventions to be indirectly transferred to households through the
13
For
14
We
the sake of simplicity, it is assumed here that the in?ation target is zero.
have checked the robustness of our results when introducing the output gap in the Taylor rule
15
The ability of central banks to peg exchange rate with partial capital mobility has been discussed by Mundell (1963)
and Obstfeld (1980).
16
This figure also corresponds to Ouyang et al. (2010) who find that almost 90% of reserves accumulation is sterilized
in China.
17
Note that this second term in the interest-rate rule also allows China's net foreign asset position to stabilize in the long run:
the downward pressure on the interest rate when the central bank accumulates foreign-exchange reserves reduces the
incentives of households to save, hence reduces their accumulation of foreign assets. Here, we set
o
FR
= 0.1.
17
pension system. Denoting by DCBt the amount of this positive or negative transfer, we have:
DCBt = E
u
(Bc
u
÷
1
R
u
÷ Bc
u
) + (Bc
c
t÷
1
R
c
t ÷ Bc
c
t÷
1
)
t t t t
(6)
where E
u
denotes the exchange rate of the renminbi against the dollar, R
u
the US interest rate
t t
and R
ct
the Chinese one.
18
Taxes, subsidies and the PAYG pension system. In each country, the government sets the tax
rate on labor endowments (t
t
)
19
and issues domestic-currency denominated risk-free bonds,
the stock of which is denoted Bg
t
at the end of period t, with a negative sign whenever the
government is a debtor. The debt level Bg
t
is set so that the debt ratio adjusts progressively to
a target level Bg
target
:
Bg
t
=µ
Bg
Bg
t
÷
1
+ (1 ÷µ
Bg
)Bg
target
P
t
Y
t
(7)
where P
t
Y
t
represents nominal GDP, andµ
Bg
is set to 0.95. In China under a fixed exchange-rate
regime, the pension system also receives positive or negative transfers from the central bank,
DCBt, and subsidizes capital accumulation (see Appendix A.4). Finally, the level of pensions,
H
t
, is set so as to abide with the budget constraint:
H
t
=t
t
W
t
L
t
+ Bg
t
+ DCBt ÷ R
t
÷
1
Bg
t
÷
1
÷ (t
k
÷ 1)rk
t
K
t
÷1 (8)
3.5. Trade
Domestic demand is the sum of households' consumption and capital firms' investment. This
demand in country j is distributed across domestic goods (D
jt
) and imported goods from country
j (M
t
jj ) according to a Dixit-Stiglitz CES. Imports and domestic demand write:
M
t
jj
=
q
jj
E
jj
P
t
j t
÷|
(I
j
t
+
C
j
) t
and
D
jt
=
1÷
q
jj
P
t
j
÷|
(I
j
+ C
j
),
Pc
jt j =j
Pc
jt
t t
where I
j
, C
j
and Pc
jt
denote investment, consumption and the consumption price index of
t t
country j, respectively, P
t
j
is the producer price index of country j, P
t
j
that of country j ,
18
In
the three-country extension of the model, a third term is introduced on the right-hand side of the equation to
account for earnings on euro-denominated reserves.
19
Taxing labor endowment rather than hours worked allows us to neglect the efficiency cost of taxation. Here, the tax has
no impact on the labor market.
18
|> 0 andq
jj
> 0 represents trade openness of j vis-ï¿
1
-vis j .
20
The trade balance writes, in 2
domestic currency:
TB
jt
=
j =j
P
t
j
M
t
j
j
÷ E
jj
P
t
j
M
t
jj t
We now turn to two series of simulations of the model. In Section 4, we study the impact of
structural policies in China leading to a rise in the national savings rate, under different
monetary regimes. In Section 5, we assume that China stays on a high-saving path and study the
impact of a monetary reform.
4. S T RU C T U R A L P O L I C I E S I N C H I N A
In this section, we study different structural reforms presented in Section 2, all of which are
deemed to impact negatively on China's aggregate savings rate. We contrast the impact of these reforms
depending on China's monetary regime. Specifically, we study three monetary regimes,
successively:
StatQuo The exchange rate is fixed, with capital controls and almost complete sterilization of
interventions;
CapMob Capital controls are relaxed, the nominal exchange rate is kept constant but there is
almost no sterilization;
Flex The renminbi is allowed to ?oat freely with no capital controls, no interventions, hence
no sterilization.
The corresponding key parameters are reported in Table 1.
Table 1 - Monetary and financial parameters under the three regimes
Regime Exchange-rate regime Capital mobility¸ Sterilizationv
steril
StatQuo Fix 0.5 0 .9
CapMob Fix 0.9 0 .1
Flex Flex 0.9 0
4.1. A pension reform
We first consider an increase in the generosity of the Chinese PAYG pension system (a 6.5 pp
increase in the replacement rate) financed through an increase in the tax rate on labor en-
dowment. In our overlapping-generation model, the reform induces a net transfers from future
generations (that will bear a share of higher future pensions through labor income taxation) to
20
We
present here the most general specification of the model that applies whatever the number of countries.
19
alive generations (that benefit from higher future pensions). Hence, aggregate consumption is
boosted in the short run.
21
In autarky, the fall in aggregate saving in the long run would induce a
higher real interest rate, hence lower capital, GDP, income, and consumption. To the extent that
capital controls are not complete, this effect is mitigated by international capital in?ows and the
Chinese net foreign asset position (NFA hereafter) falls in the long run. Higher capital mobility
allows the capital stock to fall less in the long run, thanks to a larger decline in the
NFA reduction, as depicted in Table 2.
22
Dynamic reaction of the Chinese economy. Consumption in China surges following the
shock, but the extent of this surge depends crucially on the monetary and financial regime (see Figure
3). In the status quo regime (currency-peg, low capital mobility), the Chinese interest rate is largely
independent from the US one in the short run: it rises sharply as a reaction to higher in?ation. This
increase in the interest rate mitigates the impact of the reform on consumption (+1% of GDP) and
drives investment downwards (-0.3% of GDP). GDP and employment in- crease in the short run
but the hike is short-lived: higher consumption induces households to reduce their labor, which
leads to a wage increase that feeds in?ation (+0.3 pp); hence the real exchange rate appreciates and
the trade balance deteriorates (-0.3% of GDP). The rise in China's interest rate also attracts foreign
capital in?ows (and reduces incentives for China's residents to invest abroad). Despite low capital
mobility, this change in net capital ?ows exceeds the de- terioration of China's trade balance, so the
central bank slightly increases its foreign-exchange reserves (+1% of GDP) to keep the nominal
exchange rate constant, which in turn mitigates the rise in the interest rate, despite the sterilization
policy. Due to higher in?ation, the real exchange rate appreciates during the first ten quarters, before
falling back to its baseline level.
When capital controls are removed while the peg is maintained (CapMob regime), large capital
in?ows limit the reaction of the interest rate to the shock in the short run. This feeds higher
in?ation (+0.6 pp), a limited fall in investment (-0.06% of GDP) and a large increase in con-
sumption (+2.2% of GDP). The trade balance deteriorates more (-0.6% of GDP). Net capital
in?ows are now much larger. In the short term, this is compensated by larger reserve accumu-
lation. After ten quarters however, the NFA falls much more rapidly than under the StatQuo
regime.
In a ?exible exchange rate regime (with no capital controls), the central bank no longer accu-
mulates foreign-exchange reserves whenever there is a surge in capital in?ows (or a drop in
capital out?ows). Rather, it allows the nominal exchange rate to appreciate. Hence, following
21
The
reform also induces transfers across alive households since elder households benefit immediately without
bearing its cost, whereas younger households bear the bulk of the tax. However, in our overlapping-generation model,
this redistribution drives no change in the aggregate saving rate as the propensity to consume wealth is independent
from age. The transitory increase in aggregate consumption then comes from the fact that part of the cost of the reform
will be borne by future generations.
22
Equity stocks are not reported in the table since equities are not supposed to be traded internationally here.
20
Table 2 - Long-run impact of the pension reform on China's balance sheet
Bonds supply Bonds demand
StatQuo CapMob Flex StatQuo CapMob Flex
Firms -3.0 -2.6 -2.5 Households -4.7 -9.4 -10.3
Government -0.2 -0.1 -0.1
Central bank -0.6 0.1 0 Central bank -0.7 0.8 0
Nation -3.8 -2.6 -2.6 Nation -5.4 -8.6 -10.3
NFA -1.6 -6.0 -7.7
Source: Author calculations.
a rise in the generosity of the Chinese pension system (and the subsequent decrease in house-
holds' savings in China), there is both a rise in the domestic interest rate and an immediate
appreciation of the renminbi.
23
Such policy is successful in curbing in?ation, at the expense of
consumption that grows less than under the CapMob regime (but as much as under the StatQuo one).
Because net exports are also affected by the short-run appreciation of the renminbi, GDP no longer
increases in the short run, neither does employment.In all simulations, employment varies like
GDP, hence this graph is omitted in the Figures. Interestingly, the fall in Chinese exports is
deeper in the short run than under a peg, but after six quarters the level of exports is the same as
under the CapMob regime, and the dynamics of the NFA position is similar in the two regimes. This
is consistent with the one-off appreciation of the renminbi, contrasting with progressive
appreciation through cumulated in?ation differentials under a fixed peg.
If Chinese authorities' objective function is close to that of the households, hence if it weighs
higher consumption as a positive outcome, then the best regime is the CapMob one since this is the
regime yielding the highest increase in consumption. Conversely, if Chinese authorities are mainly
concerned with in?ation, then the pension reform should be carried out under the Flex regime.
Finally, if Chinese authorities consider employment as the main objective, as it has been the case
since the mid 1990s, then our simulations suggests that the peg should be maintained while the
pension reform is being implemented.
Impact on the US economy. The transmission of the shock to the United States comes from
increased imports by Chinese households (which boosts US GDP) and from higher returns on
Chinese assets (which offers an incentive for US households to invest in China and reduces the
incentive of Chinese residents to invest in the US). Since both channels are the two sides of the
same coin, the transmission is magnified when capital is more mobile. Importantly, the rise in US
GDP comes along with a rebalancing of the US economy since both consumption and
investment fall in this country due to a higher interest rate. But only with high capital
23
The
renminbi subsequently depreciates, consistent with the interest-rate differential together with rational expec-
tations.
21
Figure 3 - Impact of a pension reform in China
Impact on China
2.5
2
1.5
1
0.5
0
÷0.5
pibchiC
0
÷0.02
÷0.04
÷0.06
÷0.08
÷0.1
÷0.12
÷0.14
÷0.16
÷0.18
÷0.2
pibchiI
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
÷0.1
perchiinfp
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo
CapMob
Flex
StatQuo
CapMob
Flex
StatQuo
CapMob
Flex
0
÷0.5
÷1
÷1.5
÷2
÷2.5
÷3
÷3.5
÷4
÷4.5
÷5
Consumption
pibchiNFA
1.6
1.4
1.2
1
0.8
0.6
0.4
0.2
0
÷0.2
÷0.4
Investment
perchiRER
0.25
0.2
0.15
0.1
0.05
0
In?ation perchiR
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo CapMob Flex StatQuo CapMob Flex StatQuo CapMob Flex
Net foreign assets perchiY Real exchange rate pibchiBC Nominal interest rate pibchibask
2 0.2 1.5
0.1
11.5 0
÷0.10.5
1 ÷0.2
÷0.3 0
0.5 ÷0.4
÷0.5÷0.5
0 ÷0.6÷1
÷0.7 per
c
hi
C
÷0.5 ÷0.8 ÷1.5
5 10 15 20 25 30 352.540 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo CapMob Flex StatQuo CapMob Flex StatQuo CapMob Flex
0.25
0.2
0.15
0.1
0.05
0
÷0.05
÷0.1
Real GDP
perusaY
2
1.5
1
0.5
0
0.2
0.1
0
÷0.1
÷0.2
÷0.3
÷0.4
÷0.5
÷0.6
÷0.7
Trade balance
Impact on the US
pibusaC
0.045
0.04
0.035
0.03
0.025
0.02
0.015
0.01
0.005
0
Official reserves
perusaR
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo
CapMob
Flex
StatQuo
CapMob
Flex
StatQuo
CapMob
Flex
GDP
÷0.5
5
10 Consumption
25
15 20 30 35
Nominal interest rate 40
StatQuo CapMob Flex
22
i
n
p
e
r
c
e
n
t
t
o
G
D
P
t
o
G
D
P
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
t
o
G
D
P
i
n
p
e
r
c
e
n
t
t
o
G
D
P
t
o
G
D
P
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
t
o
G
D
P
mobility is such rebalancing significant. Under a ?exible regime, the immediate appreciation of the
renminbi stimulates US exports and GDP in the very short run. Simultaneously, the fall in
consumption is larger in the very short run than under a fixed peg, because the interest rate itself
increases more rapidly. After five quarters, however, US exports, consumption and GDP are similar
under the Flex and the CapMob regimes. Hence, except in the short run, a ?exible exchange rate in
China does not make much difference for the United States to the extent that capital controls are
removed.
In the longer run, GDP and consumption fall below the baseline due to the permanent fall in the
investment rate (thus in productive capital). Hence, the positive impact of the Chinese pension
reform is only short lived for the US economy, although it has a permanent effect on the NFA
position.
In Appendix C, robustness checks are performed by comparing the impact of the pension reform
with only one monetary regime (StatQuo) but different values of the key parameters of the
model.
4.2. Other structural reforms
We now simulate the impact of the two other structural reforms described in Section 2, which both
have the direct consequence of reducing aggregate savings in China. The impact on the Chinese
and US economies is depicted in Figures B.1 and G.2 in Appendix B, for the three monetary
regimes.
A financial reform. Chinese firms are allowed to increase their borrowings by 45 percent of GDP,
while loosing a capital subsidy equivalent to 8% percent. The two measures have opposite effects on
investment: on the one hand, it becomes easier to finance fixed capital formation; on the other one,
the expected return is lowered due to the removal of capital subsidies. On the whole, the
investment rate increases by only 0.15 percent of GDP in the short run. The remain- ing borrowings
are channeled back to households through dividends. This one-shot transfer is triggered by a
sudden increase in the value of capital firms. It boosts consumption in the short run (+0.5 to +1% of
GDP, depending on the monetary regime).
24
The interest rate rises to op- pose in?ationary
pressures, which mitigates the rise in consumption and especially investment. The NFA position
declines due to net capital in?ows and consistent with the fall in the trade balance.
Interestingly, the positive impact of the reform on investment is permanent, which allows GDP
and consumption to increase in the long run (contrasting with the pension reform). As detailed in
Table 3, the stock of corporate bonds increases by 37 percent of GDP in the long run. Most
24
In
our model, the propensity to consume is the same for labor as for capital income. Hence, we do not consider
horizontal redistributive effects amongst households.
23
Table 3 - Long-run impact of the financial reform on China's balance sheet
Bonds supply Bonds demand
StatQuo CapMob Flex StatQuo CapMob Flex
Firms 37.1 37.2 37.2 Households 37.1 36.0 36.0
Government 0.2 0.2 0.2
Central bank 0 0 0 Central bank -0.1 0.2 0
Nation 37.3 37.4 37.4 Nation 37 36.2 36
NFA -0.3 -1.2 -1.4
Source: Author calculations.
of it is held by Chinese households, but the NFA position is reduced by 0.3 to 1.4 percent of GDP,
depending on the monetary regime.
Like for the pension reform, there is more rebalancing when capital controls are removed, while
exchange-rate ?exibility allows for less in?ation and a faster rebalancing effect. In fact, the main
advantage of exchange-rate ?exibility is its stabilizing impact on China's consumption and in?ation.
The effects of the shock on the US economy are similar to those obtained with a pension reform
in China: higher exports and GDP in the short run, lower consumption and investment due to a
higher interest rate, hence a rebalancing of the economy which very much depends on the extent
of capital mobility. Again, GDP falls in the long run. Like for the pension reform, the US
economy enjoys higher foreign demand more quickly under a ?exible regime than under a fixed peg
in China. The evolution of the NFA position of the US is similar under the CapMob and the Flex
regimes, except in the short run where exchange-rate ?exibility allows for quicker rebalancing. This is
paid in the US by a sharper short-term drop in consumption under the Flex regime.
In terms of current-account rebalancing, this reform is much less effective than the pension
reform, as can be seen by comparing Tables 3 and 2. In order for the financial reform to have the
same long-run impact on the NFA position than with the 6.5 pp pension reform described above,
capital firms would need to almost triple the debt-to-GDP ratio. However, since this reform has
positive impact on GDP and consumption in the long run, contrasting with the pension reform
that depresses income and consumption in the long run, it could make sense for China to combine the
two.
Government spending reform. We finally study the impact of a government spending re- form.
Specifically, the target debt-to-GDP ratio of China is assumed to be increased from zero to 15
percent of the steady-state GDP. Since in our model the government budget constraint determines
the level of pensions, the latter increase, so this reform is close to an unfunded pen-
24
Table 4 - Long-run impact of the government spending reform on China's balance sheet
Bonds supply Bonds demand
StatQuo CapMob Flex StatQuo CapMob Flex
Firms -0.6 -0.5 -0.5 Households 11.1 10.0 10.0
Government 12 12 12
Central bank -0.1 0 0 Central bank -0.1 0 0
Nation 11.3 11.5 11.5 Nation 11.0 10.0 10.0
NFA -0.3 -1.5 -1.5
Source: Author calculations.
sion reform. Indeed, contrasting with the pension reform studied above, the tax rate on labor
endowment does not increase. In the longer run, the stabilization of the debt ratio is obtained
through a downward adjustment of pensions (see Figure B.2 in Appendix B).
Not surprisingly, the impact of the shock for China is very similar to that of a pension reform:
rise in consumption, fall in investment, surge in in?ation, rebalancing. The main difference is that
the adjustment spreads over a longer period. Hence, differences across regimes are more marked
and long-lived. In the long run (see Table 4), the government debt increases by 12 percent of
GDP
25
, most of this new debt being held by Chinese households. The NFA position declines by 0.3
to 1.5 percent of GDP, depending on the monetary regime (again, there is more rebalancing when
capital ?ows are freed).
The ?exible exchange rate regime in China benefits immediately to US GDP, and the gain
compared to the MobCap regime lasts seven quarters (instead of five in the case of the pension
reform). The interest rate rises with some delay compared with the pension reform case.
On the whole, this reform is also less powerful than a funded reform of the PAYG pension
system to achieve current-account adjustment. This is because the rise in pensions is not per-
manent, hence young household cannot anticipate higher pensions for their old age, although they
will receive interests on their government bonds holdings. For the government spending reform to
have the same impact as the 6.5 pp pension reform on the long-run NFA position, the target debt ratio
would need to be increased by over 56 percent of GDP.
4.3. Structural reforms in China under a ZIRP in the US
In the above simulation, one impact of Chinese reforms is channeled to the US through the
federal Reserve's policy rate, the increase of which lowers aggregate demand. However, the
Federal Reserve may oppose such interest-rate increase if it is constrained by the zero bound:
25
Remember
that the shock is given as a percentage of the steady-state GDP: due to the price increase, the rise in
the debt-to-GDP ratio is only 12% ex post.
25
if the initial policy rate is above its Taylor level due to the zero bound, then an increase in US
in?ation does not immediately lead to monetary tightening. To account for this possibility, we
assume that, for the next eight quarters following the shock, US monetary policy follows a
modified rule where the interest rate smoothing parametero
R
is set to 0.98 instead of 0.8 in the
benchmark model.
The results are reported in Figure 4 for the same, three monetary regimes in China as previ-
ously: a fixed exchange rate with limited capital mobility (StatQuo), a fixed exchange rate with more
capital mobility (CapMob), and a ?exible exchange rate regime with high capital mobility (Flex).
Since the ZIRP only binds in the short and medium run, the long-run impact of the pen- sion reform
is the same as before, but the dynamics differ. Comparing Figure 4 with Figure 3, it appears clearly
that the US policy rate increases much less with the ZIRP than with a Taylor rule during the first eight
quarters of the simulation. Consistently, US consumption increases instead of declining during this
period. Interestingly then, the amount of rebalancing now is much moredependent on China's
monetary regime, as evidenced by the evolution of China's trade balance: under the StatQuo regime, the
pension reform in China produces almost no rebalancing in the short run; and the amount of
rebalancing is now doubled if the renminbi is not allowed to ?oat, compared to the CapMob and Flex
regime.
Turning to China, the main difference between Figures 4 (Taylor rule in the US) and 3 (ZIRP in
the US) occurs under the CapMob regime, i.e. when capital controls are removed but the nom- inal
exchange rate is kept constant. In this case, Chinese monetary policy is dependent on the US one.
Other things equal, the increase in the Chinese interest rate is muted, which produces more in?ation,
a larger fall in the real interest rate, hence a higher increase in consumption and a slower decrease in
investment. Consequently, the increase in real GDP is doubled compared with the Taylor-rule case:
the policy trade-off between employment and in?ation is twisted in favor of the former. Under the
other two regimes (StatQuo and Flex), the results for China are close whether the Fed follows a
Taylor rule or a ZIRP because China's monetary policy is isolated through capital controls (StatQuo)
or a ?exible exchange rate (Flex).
On the whole, our simulation under a ZIRP in the United States tends to mitigate our previous
conclusion that the regime for capital ?ows is more important than the exchange rate regime as far as
global rebalancing is concerned. We find that, under a ZIRP, the exchange rate regime matters
much more than under a standard, Taylor-rule type monetary policy.
5. THE IMPACT OF A MONETARY REFORM IN CHINA
One limitation of the simulations presented in Section 4 is that they analyze the impact of Chi- nese
reforms as deviations from a baseline which is a balanced steady-state. This feature hardly matches
the debate on global imbalances that blames accumulated imbalances and discusses how these
imbalances could unfold. Here we tackle this issue by studying the impact of a monetary reform
in China when the economy is not at the steady state but rather displays large current-account
surpluses.
26
Figure 4 - Impact of a pension reform in China with ZIRP in the United-States
Impact on China
3.5
3
2.5
2
1.5
1
0.5
0
÷0.5
pibchiC
0
÷0.02
÷0.04
÷0.06
÷0.08
÷0.1
÷0.12
÷0.14
÷0.16
÷0.18
÷0.2
pibchiI
1.2
1
0.8
0.6
0.4
0.2
0
÷0.2
perchiinfp
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo
CapMob
Flex
StatQuo
CapMob
Flex
StatQuo
CapMob
Flex
0
÷1
÷2
÷3
÷4
÷5
÷6
Consumption
pibchiNFA
3.5
3
2.5
2
1.5
1
0.5
0
÷0.5
Investment
perchiRER
0.25
0.2
0.15
0.1
0.05
0
In?ation perchiR
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo CapMob Flex StatQuo CapMob Flex StatQuo CapMob Flex
3
2.5
2
1.5
1
0.5
0
÷0.5
Net foreign assets
perchiY
0.4
0.2
0
÷0.2
÷0.4
÷0.6
÷0.8
÷1
÷1.2
Real exchange rate
pibchiBC
per
c
hi
C
Nominal interest rate
pibchibask
2
1.5
1
0.5
0
÷0.5
÷1
÷1.5
5 10 15 20 25 30 35 440 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo CapMob Flex StatQuo CapMob Flex StatQuo CapMob Flex
3.5
3
2.5
2
1.5
1
0.5
0
÷0.5
Real GDP
perusaY
3.5
3
2.5
2
1.5
1
0.5
0
2
1.5
1
0.5
0
÷0.5
Trade balance
Impact on the US
pibusaC
0.05
0.045
0.04
0.035
0.03
0.025
0.02
0.015
0.01
0.005
0
Official reserves
perusaR
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo
CapMob
Flex
StatQuo
CapMob
Flex
StatQuo
CapMob
Flex
GDP
÷0.5
5
10 Consumption
25
15 20 30 35
Nominal interest rate 40
StatQuo CapMob Flex
27
i
n
p
e
r
c
e
n
t
t
o
G
D
P
t
o
G
D
P
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
t
o
G
D
P
i
n
p
e
r
c
e
n
t
t
o
G
D
P
t
o
G
D
P
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
t
o
G
D
P
We proceed in two steps. First, we simulate an "imbalance" shock in the form of (i) a structural
shock whereby China's savings rate is permanently increased, or (ii) a shock on the target level of
China's official reserves. Second, we assume that, after 15 quarters, China decides to allow its
exchange rate to ?oat (and it subsequently keeps the level of official reserves constant). The choice to
undertake the monetary reform after 15 quarters results from the necessity to have the reform take
place once the economy has already accumulated significant imbalances. It does not change the
results qualitatively to select a shorter or a longer period.
5.1. A Pension "reverse" shock
First, we consider a reduction in the generosity of China's pension system, consistent with the
1997 pension reform. Symmetrically to the above pension reform, we assume that the
replacement rate is reduced by 6.5 pp. The results are reported in Figure 5. The grey line
corresponds to the simulation with a fixed exchange rate (StatQuo); the dotted line assumes a
?exible exchange rate from the start of the simulation (Flex); finally, the plain, black line shows the
path of the economy when switching from StatQuo to Flex after 15 quarters.
Not surprisingly, the restrictive pension shock leads to a fall in Chinese consumption in the short
run. The interest rate decreases due to de?ationary pressures. This has two consequences: (i) a rise in
investment, and (ii) a net out?ow of capital. Consistently, the trade balance surges and the NFA
position rises until it stabilizes at a higher level (+2.3% of GDP in the StatQuo regime, with low
capital mobility, and +9.7% of GDP in the Flex regime, with high capital mobility). The real
exchange rate depreciates during several quarters before appreciating towards a long run
equilibrium that lies above its baseline level (+0.2% in the StatQuo regime and +2% in the Flex
regime). Under a fixed peg, the central bank of China initially reduces its official reserves to counter-
balance the (limited) out?ow of capital. After two quarters, it starts accumulating reserves to
ensure the equilibrium of the balance of payment with higher trade surplus and lower private capital
out?ows due to the increase in China's interest rate towards its initial level. Undera ?exible regime,
official reserves stay constant and the nominal exchange rate depreciates in the short run before
appreciating steadily along the adjustment path of the economy. There is less de?ation in this
regime but the increase in the trade balance is doubled compared to the status quo regime, due to
capital mobility.
The switch from a fixed to a ?exible regime freezes official reserves at an in?ated level com-
pared to the steady state.
26
The switch takes place while the nominal, policy interest rate is still
depressed in China relative to the United States. Since the monetary change includes a
relaxation of capital controls, there is a sudden out?ow of private capital, and the renminbi
initially depreciates by 0.5% (although it appreciates in the long run by 2.5% in real terms).
27
26
Alternatively,
we could have assumed that the PBoC progressively reduces its excess reserves when switching to
a ?exible regime. However, this would have amounted to piling up two shocks: a change in the monetary regime, and a
portfolio shock. The results would have been less clear cut.
27
One may wonder why moving to exchange-rate ?exibility while the exchange rate is undervalued leads to an
28
Figure 5 - Flexibilization of China's exchange rate after a "reverse" pension reform
Impact on China
0.4
0.2
0
÷0.2
÷0.4
÷0.6
÷0.8
÷1
÷1.2
÷1.4
perchiC
1.6
1.4
1.2
1
0.8
0.6
0.4
0.2
0
perchiI
0.1
0.05
0
÷0.05
÷0.1
÷0.15
÷0.2
÷0.25
÷0.3
÷0.35
÷0.4
perchiinfp
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
Flex StatQuo Mon. Ref. Flex StatQuo Mon. Ref. Flex StatQuo Mon. Ref.
Consumption pibchiNFA Investment perchiRER In?ation perchiR
5 0.4 0
4.5 0.2
4 0 ÷0.05
3.5 ÷0.2
3 ÷0.4 ÷0.1
2.5 ÷0.6
2 ÷0.8 ÷0.15
1.5 ÷1
1 ÷1.2 ÷0.2
0.5 ÷1.4
0 ÷1.6 ÷0.25
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
Flex StatQuo Mon. Ref. Flex StatQuo Mon. Ref. Flex StatQuo Mon. Ref.
0.6
0.4
0.2
0
÷0.2
÷0.4
÷0.6
÷0.8
Net foreign assets
perchiY
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
÷0.1
÷0.2
Real exchange rate
pibchiBC
per
c
hi
C
nominal interest rate
pibchibask
1.5
1
0.5
0
÷0.5
÷1
÷1.5
÷2
5 10 15 20 25 30 350.440 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
Flex StatQuo Mon. Ref. Flex StatQuo Mon. Ref. Flex StatQuo Mon. Ref.
0.1
0.05
0
÷0.05
÷0.1
÷0.15
÷0.2
÷0.25
real GDP
perusaY
0.2
0
÷0.2
÷0.4
÷0.6
÷0.8
÷1
÷1.2
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
÷0.1
÷0.2
Trade balance
Impact on the US
perusaC
0
÷0.005
÷0.01
÷0.015
÷0.02
÷0.025
÷0.03
÷0.035
÷0.04
official reserves
perusaR
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
Flex
StatQuo
Mon. Ref.
Flex
StatQuo
Mon. Ref.
Flex
StatQuo
Mon. Ref.
GDP
÷1.4
5
10 Consumption
25
15 20 30 35
Nominal interest rate 40
Flex StatQuo Mon. Ref.
29
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
t
o
G
D
P
i
n
p
e
r
c
e
n
t
t
o
G
D
P
t
o
G
D
P
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
Due to lower purchasing power, consumption declines. In?ation is subsequently reduced. On the
whole, the regime shift triggers an increase in the trade surplus, before accompanying the economy
back to its balanced path through a strong appreciation of the real exchange rate.
When the regime shift takes place, US GDP is negatively affected (because the dollar appreci-
ates) but US consumption is boosted (due to the higher purchasing power of US households). This
mirror image of China confirms the failure of the regime shift to rebalance the US econ- omy.
The lack of rebalancing following a switch of China from a pegged regime to a ?exible one
arises from the relaxation of capital controls and from the reluctance of the PBoC to sell its
accumulated stock of reserves. It should be kept in mind that the relaxation of capital controls is a
precondition for the switch to a ?exible exchange-rate regime, since the proper functioning of the
foreign-exchange market necessitates both residents and non-residents to be able to trade the
renminbi for other currencies. What the model says is that switching to a ?exible regime (with free
capital mobility) in a situation of depressed demand would put downward rather than upward
pressure on the renminbi in the short run unless (i) structural reforms are simultaneously (or previously)
carried out to reduce aggregate savings in China and/or (ii) the PBoC decides to sell excess
reserves. Of course, this stylized result may be mitigated by the initial return differential between
China and the United States. It should be reminded here that only bonds yielding the policy rate
are traded between the two countries: in our model, relaxing capital controls does not induce US
investors to buy high-return Chinese stocks, which may put upward pressure on the RMB. Still, the
model highlights the complementarity between monetary and structural reforms to engineer a fast
rebalancing of the economy.
5.2. Reserve accumulation shock
Even though a "reverse" pension shock may have contributed to the high saving rate in China and
the growing imbalances, this shock is not sufficient to explain official reserve accumulation. As argued
by Cova et al. (2009), the rise in emerging Asia's current account surplus after 2001 may also be
related to an increase in the desired net foreign asset position. In the case of China, due to capital
controls, this suggests an increase in the target level for official reserves. Here, we assume that this
target increases progressively by 38 percent of GDP in 20 years. Accordingly, the interest rate rule of
the Chinese central bank is shifted upward in order to be consistent with the constant exchange rate
constant (see Equation (5)). Like in the previous exercise, China is assumed to shift to a ?exible
regime after 15 quarters.
The results are displayed in Figure 6.
28
To the extent that this policy is well anticipated, the
immediate depreciation rather than an appreciation. The reason is the uncovered interest parity with rational
expectations: the exchange rate initially jumps to a path where it can then appreciate at a pace that is consistent with the
interest-rate differential. Before the monetary reform takes place, the exchange-rate is undervalued compared
to its long-run, equilibrium level, not compared to its short-term, market level
28
Here, we do not report the results when China is running a Flex regime during the whole period since, by
30
Figure 6 - Flexibilization of China's exchange rate after a reserve accumulation shock
Impact on China
pibchiC pibchiI perchiinfp
1 0 0.5
0.5
÷0.02
0 0
÷0.5 ÷0.04
÷1 ÷0.5÷0.06
÷1.5
÷0.08÷2 ÷1
÷2.5 ÷0.1
÷3 ÷1.5
÷0.12
÷3.5
÷4 ÷0.14 ÷2
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
MonRef StatQuo MonRef StatQuo MonRef StatQuo
25
20
15
10
5
0
Consumption
pibchiNFA
2
1
0
÷1
÷2
÷3
÷4
÷5
÷6
Investment
perchiRER
0.6
0.4
0.2
0
÷0.2
÷0.4
÷0.6
÷0.8
In?ation perchiR
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
MonRef StatQuo MonRef StatQuo MonRef StatQuo
1
0.5
0
÷0.5
÷1
÷1.5
÷2
÷2.5
÷3
Net foreign assets
perchiY
2.5
2
1.5
1
0.5
0
÷0.5
÷1
Real exchange rate
pibchiBC
per
c
hi
C
Nominal interest rate Test
reserves
30
Target reserves
25
20
15
10
5
0
÷5
5 10 15 20 25 30 35 140 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
MonRef StatQuo MonRef StatQuo
0.6
0.4
0.2
0
÷0.2
÷0.4
÷0.6
÷0.8
Real GDP
perusaY
0
÷1
÷2
÷3
÷4
2.5
2
1.5
1
0.5
0
÷0.5
÷1
Trade balance
Impact on the US
pibusaC
0.1
0.05
0
÷0.05
÷0.1
÷0.15
÷0.2
÷0.25
Official reserves
perusaR
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
MonRef StatQuo MonRef StatQuo MonRef StatQuo
GDP
÷5
5
10 Consumption
25
Mon Ref
15
31
20 30
StatQuo
35
Nominal interest rate 40
i
n
p
e
r
c
e
n
t
t
o
G
D
P
t
o
G
D
P
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
t
o
G
D
P
i
n
p
e
r
c
e
n
t
t
o
G
D
P
t
o
G
D
P
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
t
o
G
D
P
real interest rate is expected to be higher until the level of reserves has reached its new target. As a
consequence, consumption initially drops (intertemporal substitution effect). In?ation decreases
in the short run, the real interest rate increases but the nominal interest rate starts by a large fall to
mitigate the decline in Chinese in?ation. Despite the fixed nominal exchange rate, the real exchange
rate depreciates. Due to nominal rigidities, the depreciation is spread over several quarters.
The Chinese economy progressively accumulates trade surpluses, hence assets on the US econ-
omy in the form of official reserves. Symmetrically, consumption increases in the US thanks to a
lower expected level of nominal interest rate in the medium run, whereas GDP is lowered as
domestic demand shifts to imported goods. After a few years, when a sizable amount of official
reserves has been accumulated, the real interest rate in China starts declining, increasing con-
sumption and reducing the current account surplus. The real exchange rate appreciates through a
recovery of in?ation in China (by 15% from its value in quarter 15 to its long run equilibrium value).
At this stage, a monetary reform that encompasses the relaxation of capital controls, a shift to
a ?exible exchange rate regime and the end of official reserve accumulation (that are frozen at
their in?ated, current value, +11.5% of GDP) induces a large appreciation of the renminbi, by 10%,
and simultaneously removes any in?ationary pressure in China. It also rebalances the world economy
as consumption drops in the US, the nominal interest rate increases and GDP recovers.
On the whole, these two successive simulations suggest that the key factor of rebalancing is less
the ?exibility of the exchange rate by itself than its implications in terms of official reserves, and
that a exchange-rate ?exibility can usefully complement but not be a substitute for structural reforms.
6. A THIRD COUNTRY: THE E U RO A R E A
Finally, we add the Euro area to the model which therefore becomes a three-country model. The
model of the Euro area is the same as for the United States, except for the share of their respec- tive
currencies in Chinese official reserves. Here, we assume that foreign exchange reserves in China
amount to 60% of Chinese GDP, dollar-denominated assets represent 80% of the total and euro-
denominated assets the remainder 20%.
29
We also assume that euro and dollar are not perfect
substitutes: capital mobility between the United States and the Euro area is however high, with¸ = 0.9.
This three-country model is then used to perform three series of simulations: (i) a pension
definition, there is no reserve accumulation under a Flex.
29
At the end of 2010, foreign reserves reached USD 2800 bn whereas Chinese GDP amounted to USD 5100 bn. The
currency composition of Chinese reserves is not released, but the share of the euro is generally estimated around 20%.
32
reform in China (as in Section 4.1), (ii) a diversification of China's official reserves, and (iii) a
diversification of China's official reserves under a ZIRP in both the United States and the Euro area.
6.1. Impact of a pension reform in China in a three-country setting
The impact of an increase in the generosity of China's pension system is similar for the Euro
area as for the United States (see Figure 7). The minor differences come from the higher share
of the dollar in Chinas' reserves.
30
Figure 7 - Impact of a pension reform in China, three-country model
Impact on the US and the euro area
0.3
0.25
0.2
0.15
0.1
0.05
0
÷0.05
Y
0.1
0
÷0.1
÷0.2
÷0.3
÷0.4
÷0.5
÷0.6
C
4
3
2
1
0
÷1
X
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo CapMob Flex StatQuo CapMob Flex StatQuo CapMob Flex
0.04
0.035
GDP
infp
0.035
0.03
Consumption R
0.6
0.5
Exports
pereurNER
0.03
0.025
0.02
0.015
0.01
0.005
0
5
10
3
2.5
15
2
20
25
30
35
40
0.025
per
c
hi
C
0.02
0.015
0.01
0.005
0
5
10
15
20
25
30
35
40
0.4
0.3
0.2
0.1
0
÷0.1
5
10
15
20
25
30
35
40
StatQuo CapMob Flex StatQuo CapMob Flex StatQuo CapMob Flex
0.6
0.5
0.4
0.3
0.2
0.1
0
÷0.1
In?ation BC
1.5
1
0.5
0
Nominal interest rate NFA
5
4
3
2
1
0
0.06
0.04
0.02
0
÷0.02
÷0.04
÷0.06
euro/dollar nominal
pereurRER
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo CapMob Flex StatQuo CapMob Flex StatQuo CapMob Flex
Trade Balance ÷0.5
5 10
15
Net foreign asset
20 25 30 35
40
euro/dollar real
StatQuo CapMob Flex ? Euro area
The Chinese reform is a symmetric shock for the United States and the Euro area. The rebal-
ancing occurs both between China and the United States and between China and the Euro area.
30
In
the steady state, the PBoC holds official reserves equivalent to 60% of GDP. In the simulations performed
under a Flex regime, this amount stays constant but the asymmetric distribution of existing Chinese reserves across the two
currencies produces some asymmetric valuation effects.
33
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
However, as long as the Chinese peg is maintained (StaQuo and CapMob regimes), there is more
rebalancing in the United States than on the Euro area, as the drop in Chinese demand for
reserves falls relatively more on the dollar than on the euro. As a result, the interest rate
increases more in the US, lowering investment and stimulating household saving. The differ-
ences between the US and Euro area are benign such that the EUR/USD exchange rate is not
significantly affected by the Chinese reform. The large sale of dollar compared to euro by the
Chinese central bank triggers a depreciation of the dollar against the euro that induces private
agents in the US and in the euro area, who enjoy free capital mobility, to buy more dollars. This
stabilizing effect contributes to the absence of euro-dollar reaction in the first period. We now turn to
simulating the impact of a change in PBoC's portfolio allocation.
6.2. Impact of a diversification of China's official reserves
We now simulate a diversification of China's foreign exchange reserves out of the dollar into the
euro, with the renminbi still being pegged on the dollar. Specifically, the share of the euro in China's
reserves is assumed to increase to 30% from 20%. As a first step, monetary policy is assumed to
follow a Taylor rule in both the Euro area and the United States. The results are displayed in
Figure 8, where the plain line represents the status quo regime (limited capital mobility) and the
dotted one shows the same simulation when capital controls are relaxed in China.
This diversification induces large capital in?ows into the Euro area, hence a sizable apprecia-
tion of the euro and a fall in Euro area's growth rate. However potential growth increases in the
medium run due to a lower interest rate. The impact is opposite in the US: the dollar deprecia- tion
promotes growth in the short run, but the potential is reduced in the long run due to lower capital
accumulation.
6.3. Impact of a diversification of China's official reserves under a ZIRP
Our last exercise is to study the impact of the same diversification of Chinese reserves under a zero
interest-rate policy in the Euro area and in the United States (see Figure 9). Due to the zero
bound, the ECB is unable to cut its interest rate, whereas monetary tightening in the United States
is also muted. Hence, the fall in European GDP is larger than under a standard Taylor-rule type
monetary policy, whereas the increase in GDP is magnified in the United States. The bilateral
exchange rate of the euro vis-a-vis the dollar appreciates much more than under unconstrained
monetary policy.
7. CONCLUSION
We have constructed a two-country model to study different scenarios of global rebalancing
depending on (i) structural reforms decided in China, (ii) the monetary regime of China, (iii) the
constraints on monetary policy in the United States and (iv) the initial level of cumulated
34
Figure 8 - Impact of a diversification of China's official reserves, three country model
Impact on China
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
perchiC 10x10
8
6
4
2
0
÷2
÷3
pibchiI 0.15
0.1
0.05
0
÷0.05
perchiinfp
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo CapMob StatQuo CapMob StatQuo CapMob
0.1
0.05
0
÷0.05
÷0.1
÷0.15
Consumption
pibchiNFA
0.4
0.3
0.2
0.1
0
÷0.1
Investment
perchiRER
0.04
0.03
0.02
0.01
0
÷0.01
÷0.02
In?ation perchiR
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo CapMob StatQuo CapMob StatQuo CapMob
Net foreign assets Real exchange rate Nominal interest rate
Y Impact on the US and the euro area (?1.5
C ) X
0.15 0.4
0.1 0.3 1
0.05 0.2 0.5
0 0.1 0
÷0.05 0 ÷0.5
÷0.1 ÷0.1 ÷1
÷0.15 ÷0.2 ÷1.5
÷0.2 ÷0.3 ÷2
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo CapMob StatQuo CapMob StatQuo CapMob
0.02
0.01
GDP
infp
0.02
0.015
Consumption R
1.4
1.2
Exports
pereurNER
0
÷0.01
÷0.02
÷0.03
0.35
0.3
0.01
per
c
hi
C
0.005
0
÷0.005
÷0.01
÷0.015
1
0.8
0.6
0.4
0.2
5
10
0.25
15
20 25 30 35 40 ÷0.02 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40
StatQuo CapMob StatQuo CapMob StatQuo CapMob
0.3
0.2
0.1
0
÷0.1
÷0.2
÷0.3
÷0.4
In?ation BC
0.2
0.15
0.1
0.05
Nominal interest rate NFA
3
2
1
0
÷1
÷2
1.2
1
0.8
0.6
0.4
0.2
0
÷0.2
euro/dollar nominal
pereurRER
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo CapMob StatQuo CapMob StatQuo CapMob
Trade
0
Balance
5
10
15
Net foreign asset
20 25 30 35
40
euro/dollar real
StatQuo CapMob ? Euro area
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
t
o
G
D
P
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
t
o
G
D
P
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
35
Figure 9 - Impact of a diversification with a ZIRP in the United states and the Euro Area
Impact on China
perchiC ÷3 pibchiI perchiinfp 0.4 14x10 0.35
0.35 12 0.3
0.3 10 0.25
0.25 8 0.2
0.2 6 0.15
0.15 4 0.1
0.1 2 0.05
0.05 0 0
0 ÷2 ÷0.05
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo ZIRP StatQuo ZIRP StatQuo ZIRP
0.25
0.2
0.15
0.1
0.05
0
÷0.05
÷0.1
÷0.15
Consumption
pibchiNFA
0.4
0.3
0.2
0.1
0
÷0.1
Investment
perchiRER
0.1
0.08
0.06
0.04
0.02
0
÷0.02
In?ation perchiR
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo ZIRP StatQuo ZIRP StatQuo ZIRP
Net foreign assets Real exchange rate Nominal interest rate
1
0.5
Y
Impact on the US and the euro area (?)3C
0.4
2
0.2
1
X
0
÷0.5
÷1
0
÷0.2
÷0.4
0
÷1
÷2
÷3
÷4
÷1.5 ÷0.6 ÷5
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo ZIRP StatQuo ZIRP StatQuo ZIRP
0.3
0.2
GDP
infp
0.02
0.015
Consumption R
3.5
3
Exports
pereurNER
0.1
0
÷0.1
÷0.2
÷0.3
÷0.4
÷0.5
0.4
0.35
0.3
0.01
per
c
hi
C
0.005
0
÷0.005
÷0.01
÷0.015
÷0.02
2.5
2
1.5
1
0.5
0
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo ZIRP StatQuo ZIRP StatQuo ZIRP
In?ation
0.25 BC
Nominal interest rate NFA euro/dollar nominal
pereurRER
0.4 3 2.5
0.2
0
÷0.2
÷0.4
÷0.6
0.2
0.15
0.1
0.05
2
1
0
÷1
÷2
÷3
2
1.5
1
0.5
0
÷0.5
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo ZIRP StatQuo ZIRP StatQuo ZIRP
Trade
0
Balance
5
10
15
Net foreign asset
StatQuo
20 25
36
30
ZIRP
35
40
euro/dollar real
? Euro area
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
t
o
G
D
P
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
t
o
G
D
P
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
imbalances. Our simulations suggest that, if monetary policy follow a standard, Taylor rule in the
United States, then structural reforms in China that reduce aggregate savings are a powerful driver of
global rebalancing, provided there is some relaxation of capital controls. A ?exibiliza- tion of the
renminbi can accelerate the rebalancing but has only a minor effect on its extent. For China,
switching to a ?exible exchange-rate regime during the reform raises a clear trade-off between
in?ation (which would be stabilized with a ?exible regime) and employment (which would suffer).
The possibility that China moves close to the "Lewis turning point" may change the terms of this
trade-off and could favor a ?exible exchange rate.
Now, if the US monetary policy is committed to a zero interest rate, then a move to a ?exible
exchange-rate regime in China has more impact on the amount of the rebalancing. In this
case, the Federal Reserve refrains from hiking its interest rate when global savings are reduced.
Hence the rebalancing can only come from a depreciation of the dollar against the renminbi,
hence from a more ?exible exchange-rate regime in China. The US ZIRP also modifies the policy
trade-off in China since removing capital controls while keeping a fixed peg on the dollar would
amount to adopting the US ZIRP, hence let Chinese in?ation would develop without control.
We then perform the reverse exercise where an initial reduction in the generosity of China's
pension system (as with the 1997 reform) raises the saving rate of households, triggering net
foreign asset accumulation under a fixed peg, before the renminbi is eventually allowed to ?oat.
To the extent that the switch from a peg to a ?oat is accompanied by a relaxation of capital
controls but not by structural reforms reducing the saving rate, and assuming that China's official
reserves are frozen at their observed level when switching to the ?oating regime (not sold out), then the
regime shift involves downward pressure on the renminbi, which delays rather than accelerating
the rebalancing. The reason is the large out?ow of private capital involved by the conjunction of a
low policy rate in China (related to excess savings) and the relaxation of capital controls.
Alternatively, if Chinese current-account surpluses are the result not of a depressed consump- tion
but of a rise in the official reserve target, then the move to a ?exible regime (with frozen reserves)
does produce the desired rebalancing features.
Finally, we introduce the Euro area as a third country in the model. We find here that despite the peg of
the renminbi on the dollar, a structural reform in China produces very similar effects on the Euro
area as on the US economy. Conversely, a diversification of China's official reserves depresses the
economy of the Euro area through an appreciation of the euro against both the dollar and the
renminbi. Interestingly, this detrimental effect is magnified by having a ZIRP in both the United
States and the Euro area.
We conclude that the contribution of China in global rebalancing should primarily rely on
structural policies that reduce aggregate savings in China and on simultaneous opening up of the
Chinese economy. The role of the exchange-rate regime would be minor under standard
37
monetary policies, although more important if monetary policies in advanced countries are con-
strained. Finally, relying only on a change in China's monetary regime could end up in delaying rather
than accelerating the rebalancing, depending on China's policy regarding accumulated reserves.
38
BIBLIOGRAPHY
Baker, M. & Orsmond, D. (2010). Household consumption trends in china. RBA Bulletin, (pp.
13-17).
Bayoumi, T., Tong, H., & Wei, S.-J. (2010). The Chinese Corporate Savings Puzzle: A Firm-
level Cross-country Perspective. NBER Working Papers 16432, National Bureau of Eco-
nomic Research, Inc.
Bernanke, B. S. (2005). The global saving glut and the u.s. current account deficit. Sandridge
Lecture, Virginia Association of Economics, Richmond, Virginia.
Blanchard, O. & Giavazzi, F. (2006). Rebalancing growth in china: A three-handed approach.
China & World Economy, 14(4), 1-20.
Blanchard, O., Giavazzi, F., & Sa, F. (2005). International investors, the u.s. current account,
and the dollar. Brookings Papers on Economic Activity, 36(2005-1), 1-66.
Blanchard, O. & Milesi-Ferretti, G. M. (2010). Global Imbalances: In Midstream? CEPR
Discussion Papers 7693, C.E.P.R. Discussion Papers.
Blanchard, O. J. (1985). Debt, deficits, and finite horizons. Journal of Political Economy, 93(2),
223-47.
Caballero, R. J., Farhi, E., & Gourinchas, P.-O. (2008). An equilibrium model of "global im-
balances" and low interest rates. American Economic Review, 98(1), 358-93.
Calvo, G. A. (1983). Staggered prices in a utility-maximizing framework. Journal of Monetary
Economics, 12(3), 383-398.
Chamon, M., Liu, K., & Prasad, E. (2010). Income uncertainty and household savings in China.
Imf working papers, International Monetary Fund.
Chamon, M. D. & Prasad, E. S. (2010). Why are saving rates of urban households in china
rising? American Economic Journal: Macroeconomics, 2(1), 93-130.
Cova, P., Pisani, M., & Rebucci, A. (2009). Global imbalances: The role of emerging asia.
Review of International Economics, 17(4), 716-733.
Cui, L. & Aziz, J. (2007). Explaining China's Low Consumption: The Neglected Role of House-
hold Income. IMF Working Papers 07/181, International Monetary Fund.
Dooley, M. P., Folkerts-Landau, D., & Garber, P. (2003). An Essay on the Revived Bretton
Woods System. NBER Working Papers 9971, National Bureau of Economic Research, Inc.
Faruqee, H., Laxton, D., Muir, D., & Pesenti, P. A. (2007). Smooth landing or crash? model-
based scenarios of global current account rebalancing. In G7 Current Account Imbalances:
Sustainability and Adjustment, NBER Chapters (pp. 377-456). National Bureau of Economic
Research, Inc.
Feng, J., He, L., & Sato, H. (2009). Public pension and household saving: Evidence from China.
BOFIT Discussion Papers 2/2009, Bank of Finland, Institute for Economies in Transition.
Guo, K. & N'Diaye, P. M. P. (2010). Determinants of China's Private Consumption: An Inter-
39
national Perspective. IMF Working Papers 10/93, International Monetary Fund.
Horioka, C. Y. & Wan, J. (2007). The determinants of household saving in china: A dynamic
panel analysis of provincial data. Journal of Money, Credit and Banking, 39(8), 2077-2096.
Huang, Y. & Wang, B. (2010). Cost distortions and structural imbalances in china. China &
World Economy, 18(s1), 1-17.
Jha, S., Prasad, E., & Terada-Hagiwara, A. (2009). Saving in Asia and Issues for Rebalancing
Growth. Working Papers 162, Asian Development Bank.
Ma, G. & Yi, W. (2010). China's High Saving Rate: Myth and Reality. Working Paper 312,
Bank for International Settlements.
Mendoza, E. G., Quadrini, V., & Rios-Rull, J.-V. (2007). Financial Integration, Financial Deep-
ness and Global Imbalances. NBER Working Papers 12909, National Bureau of Economic
Research, Inc.
Mundell, R. A. (1963). Capital mobility and stabilization policy under fixed and ?exible ex-
change rates. The Canadian Journal of Economics and Political Science / Revue canadienne
d'Economique et de Science politique, 29(4), pp. 475-485.
Obstfeld, M. (1980). Imperfect asset substitutability and monetary policy under fixed exchange
rates. Journal of International Economics, 10(2), 177-200.
Obstfeld, M. & Rogoff, K. S. (2005). Global current account imbalances and exchange rate
adjustments. Brookings Papers on Economic Activity, 36(2005-1), 67-146.
Ouyang, A. Y., Rajan, R. S., & Willett, T. D. (2010). China as a reserve sink: The evidence
from offset and sterilization coefficients. Journal of International Money and Finance, 29(5), 951-
972.
Prasad, E. S. (2009). Rebalancing Growth in Asia. NBER Working Papers 15169, National
Bureau of Economic Research, Inc.
Taylor, J. B. (1993). Discretion versus policy rules in practice. Carnegie-Rochester Conference
Series on Public Policy, 39(1), 195-214.
Yaari, M. E. (1965). Uncertain lifetime, life insurance, and the theory of the consumer. The
Review of Economic Studies, 32(2), 137-150.
40
APPENDIX
A. Details of the model
A.1. Households' program
Households at the cohort level In region j, the inter-temporal budget constraint of a household
born at time a and still living at time t writes (we skip the j superscript for clarity):
Pc
t
C
a,t
+ A
a,t
= W
t
L
a,t
÷t
t
W
t
L
t
÷
a
+ O
a,t
+ H
a,t
, ¯
(9)
where Pc
t
denotes the consumption price index, A
t,a
is the end-of-period asset holding, W
t
L
t,a
÷
t
t
W
t
L
t
÷
a
is the after-tax wage income,
31
O
a,t
represents beginning-of-period financial wealth, and H
a,t
¯
pensions received from the government. Asset holding consist in equities on domestic firms (Z
a,t
(f ),
where f is the firm index), a complete set of contingent claims in domestic currency (Bh
j
a,t
+1
) and ,j
risk-free nominal bonds in foreign currency (Bh
j
a,t ). The value of the portfolio A at the end of period t ,j
writes:
A
ja,t
= E
t
u
t
+1
,j
Bh
j
a,t
+1
+ ,j
t
E
j,j
Bh
j
a,t + Q
jt
(f )Z
j
(f )df ,j
j =j
t f a,t
(10)
Where u
t
+1
,j
is the stochastic discount factor and Q
jt
(f ) is the nominal equity price. t
Table A.1 - Households variables at the cohort level in country j
Name
C
a,t
Description
consumption in volume
Name
H
a,t
Description
pension received from the government
L
a,t
O
a,t
effective labor time
beginning of period financial wealth
L
t
÷a ¯
A
a,t
time endowment
end-of-period financial wealth
Bh
j
a,t
,j
end-of-period foreign bonds holding
Bh
j
a,t+1 ,j
end-of-period holding of contingent
(on date t+1 shocks realization) claims
Given the asset accumulation at period t ÷ 1, and conditionally on the survival of the household at t, the financial
wealth is augmented thanks to the life-insurance contract where bonds are fully eligible
whereas equities are partially eligible (a sharev
E
). Letu
E
=v
E
+ (1 ÷v
E
)u
g
.
O
ja,t
=
1
u
g
Bh
j
a,t + ,j
j =j
E
j,j
R
jt
÷
1
Bf
j
a,t÷
1
+ ,j
t
u
E
u
g
j
f
(Q
t
(f )
+ D
j
(f ))Z
j
÷
1
(f )df
(1 ÷
v
E
)Vk
j
/N
j
¯
t
t a,t
if a < t
if a = t
(11)
Where Vk
j
is the capital firm value and D
j
(f ) is the dividend received by households. Each household
t t
decides his level of consumption, labor supply and asset holdings so as to maximize his intertemporal
utility (1) under the budget constraint (9). The optimality conditions implied by the household program
are the following:
31
In
order to disentangle the redistributive from the allocation effect of labor taxation, we assume that the govern-
ment is able to tax time endowment rather than labor supply.
41
Euler equation
u
t
+1
=
t
|
+1
CC
a,t
t
c
t
a,t+1
Labor supply
L
a,t
= L ¯
a
÷
t
÷
1
÷k MRS
t
k C
a,t
Firm's stock price:
Q
t
= E
t
u
t
+1
(D
t+1
+ Q
t+1
) t
E
jj
Bh
jj
Foreign bonds holding
t a,t
=
¸
jj
t E
jj
t
÷1
A
j
a,t
1÷¸
jj
E
t
u
t
+1
,j
R
j
t
E
jj
t+1
It can be shown that, at the cohort level, consumption is a linear function of total wealth, i.e. financial
wealth (O) and human wealth HW (the future stream of labor an pension income) which is defined by:
·
HW
a,t
= u
t
+
s
u
s
(1÷t
t
+
s
)W
t
+
s
L
t
+s÷
a
+ H
a,t
+s ¯
s=0
t g
where H
a,t
+
s
represents the pension received by generation a at time t + s.
Table A.2 - Notations for regions and agents
symbol description possible values
j, j region u (the United States), c (China), e (euro area)
agent h (households), g (government), b (central bank), f (firms)
Aggregation The aggregate household behavior is described by two main relations: (i) aggregate labor
supply and (ii) aggregate consumption. The former is easily derived through direct aggregation of cohort-
specific labor supplies and is similar to the representative-agent case. The latter is proportional to total wealth,
i.e. human wealth, HW , (discounted sum of future net of tax labor income and pensions of already alive
households), plus financial wealth,O, (the market value of equities and bonds).
k C
t
L
t
= L ÷ 1 ÷k MRS , t
(12) Pc
t
C
t
= O
t
(HW
t
+ O
t
) (13)
To evaluate the nominal stochastic discount factor, u
t
+1
, we introduce a new variable, the relative t
consumption of the young rcy
t
, which is specific to overlapping-generation models. One then has:
u
t
+1
= 1 ÷ (1u÷|u )rcy PcPc
1
C
t
÷
1
,
where
t g g t t÷
t
C
t
(14)
rcy
t
= HW
t,t
+ O
t,t
(15)
HW
t
+ O
t
A.2. Unions and wage setting
The representative union allowed to resets its wage at period t chooses W
t
-
j
so as to maximize the
following objective function:
E
t
·
s=0
(u
g
u
w
)
s
u
t
+
s
(W
t
- t
÷ MRS
t
+
s
Pc
t
+
s
)
42
W
t
-j
W
t
j+s
÷
w
L
jt
+s
(16)
where M RS
t
+
s
denotes the marginal rate of substitution between consumption and leisure in period t+s.
The union receives W
t
-
j
from firms and pays MRS
t
Pc
t
to households. The first order condition of
the program defined by Equation (16) writes
W
t
-
= 1 ÷w
±
t
w W ,
where
w
Z
w
t
t
±
t
w
= E
t
·
s=0
(u
g
u
w
)
s
u
t
+s t
W
t
W
t
+s
÷1÷
w
Pc
t
+
s
MRS
t
+
s
L
t
+s
and
Z
w
= E
t
t
·
s=0
(u
g
u
w
)
s
u
t
+s t
W
t
W
t
+s
÷
w
L
t
+s
.
A.3. Production-firms and price setting
At time t, the production firm i chooses labor and capital inputs so as to maximize profit, given the
demand function Y
t
(i). Letting W
t
and rk
t
denote the nominal wage and the rent cost of capital, respec-
tively, the marginal cost MC
t
writes
1
1
o
1
1÷o
MC
t
= A t o 1÷o rk
o
W
t
1÷
o
. t
(17)
A firm that resets its price at period t will choose P
t
-
that maximizes its market value Vp
t
(i):
Vp
t
(i) = E
t
·
s=0
(u
p
u
E
)
s
u
t
+
s
[P
t
-
÷ MC
t
+
s
] t
P
t
-
P
t
+s
÷
p
Y
t
+
s
,
(18)
where Y
t
+
s
is aggregate demand at time t + s and P
t
+
s
the aggregate production-price level at the same
time.
The representative firm maximizes the objective function given by Equation (18) which leads to the
following first order condition
p
P
t
-
=
±
t
p P ,
where
1 ÷
p
Z
p
t t
±
t
p
= E
t
·
s=0
(u
g
u
p
)
s
u
t
+s t
P
t
P
t
+s
÷1÷
p
MC
t
+
s
Y
t
+s
and
Z
p
= E
t
t
·
s=0
(u
g
u
p
)
s
u
t
+s t
P
t
P
t
+s
43
÷
p
L
t
+s
.
A.4. Capital-firms and capital accumulation
Each capital firm k chooses investment I and external funding Bf in order to maximize its market value,
equal to the expected discounted sum of future dividends, taking as given the rental price of capital rk ,
the subsidies to capital accumulationt
k
, the price of investment Pi and the nominal interest rate on
loans R.
32
The capital-firm so chooses the level of capital K, the level of investment I and the amount
of external financing (Bf ) in order to maximize its market value:
·
(1 +t
k
)rk
t
+
s
K
t
+s÷
1
(k) ÷ Pi
t
+
s
I
t
+
s
(k)
Vk
t
(k) = E
t u
t
+
s
u
s
s=0
t E
+R
t
+s÷
1
Bf
t
+s÷
1
(k) ÷ Bf
t
+
s
(k)
. (19)
under the financing constraint (Equation (3), p.16) and the capital accumulation:
K
t
+
s
(k) = K
t
+s÷
1
(k)u
I
t
+
s
(k) ÷o .
K
t
+s÷
1
(k)
Because dividend ?ows are more discounted than bonds, the borrowing constraint (3) is always binding.
Thanks to this binding condition, the value of the capital firm can then be rewritten in a simpler way that can be
interpreted as the "book value" (see Box).
·
Vk
t
(k) = E
t
s=0
u
t
+
s
(ç + (1÷ç)u
E
)
s
[(1 +t
k
)rk
t
+
s
K
t
+s÷
1
(k) ÷ Pi
t
+
s
I
t
+
s
(k)] +R
t
÷
1
Bf
t
÷
1
(k). t
(20)
This last expression shows the positive effect of financial development on the value of capital firms, hence
on capital accumulation. At the steady-state, the marginal productivity of capital has to cover the
depreciation rateo and the internal financing cost, that equals R/u
E
in absence of external borrowing
(ç = 0). With developed financial markets, the internal financing cost drops to R/(ç + (1÷ç)u
E
) and
the level of capital rises.
Box: The value of the firm with borrowing constraints
Let Vk
1
(k) = Vk
0
(k) ÷ R
t
÷
1
Bf
t
÷
1
(k). Vk
1
(k) verifies the following recursive relation
t t t
Vk
1
(k) = (1 +t
k
)rk
t
K
t
÷
1
(k) ÷ Pi
t
I
t
(k) ÷ (1 ÷u
E
)Bf
t
(k) + E
t
u
t
+1
u
E
Vk
1
÷
1
(k)
t t t
Asu
E
< 1, the borrowing constraint (3) is binding which leads to
ç
÷Bf
t
(k) = 1 ÷
ç
E
t
u
t
+1
Vk
0
+1(k) ,
t t
i.e.
÷ Bf
t
(k) =çE
t
u
t
+1
Vk
1
+1(k) ,
Vk
1
(k) is rewritten as follows: t
t t
Vk
1
(k) = (1 +t
k
)rk
t
K
t
÷
1
(k) ÷ Pi
t
I
t
(k) [(1 ÷u
E
)ç +u
E
] E
t
u
t
+1
Vk
1
+1(k) .
t t t
One can then derive Equation (20)
32
At
date t, dividends are the residual of the ?ow of funds: previous borrowing enters negatively as it has to be
reimbursed whereas fresh borrowing enters positively as it can increases investment or... dividends. Capital firms as also
assumed not to be able to sell their entire capital stock. This assumption does not modify their optimization program.
44
We find two first order conditions
µ
t
=
Pi
t
,
u (
K
Itt÷
1
)
I
t+1
(k)
µ
t
= E
t
u
t
+1
((1 ÷u
E
)ç +u
E
) (1 +t
k
)rk
t+1
+ µ
t+1
ut
where µ
t
is the investment's Lagrange-multiplier.
We can write the cost of capital at the steady-state
K
t
(k)
÷ IK ((kk))
ut+1
t
I
t+1
(k)
K
t
(k)
.
rk = 1 Pit (1 ÷u Rç +u ÷ (1 ÷o) .
+
k
E
) E
A.5. Market equilibria
Good markets equilibrium
Balance of payments
jj
¬j,
Y
t
j
= D
j
+ t
j =j
M
t
j j
(21)
¬j,
j =j
E
t
Bh
j
t
j
+ Bc
j
tj
j holding of j' assets
÷ Bh
j
t
j
+ Bc
j
t
j
=
j' holding of j assets
E
jj
t
Bh
j
tj
1
+ Bc
j
tj
1
R
j
t÷
1
÷ Bh
j
t÷j
1
+ Bc
j
t÷j
1
R
jt
÷
1
+ TB
jt
j
÷ ÷
(22)
Bond markets equilibrium For the record, Bh
i,j
and Bg
i,j
denote the level of bonds denominated t
in the currency of region j and owned by the household sector and the government sector of region i
respectively. Bf
jt
is the (opposite of) the total borrowing of capital-firms in region j. The following
equilibrium holds:
¬j, Bh
j
t
j
+ Bg
j
t
j
+ Bc
j
j
+ Bf
jt
= 0 (23)
j
A.6. Calibration of the model
See Table A.3 page 46.
B. Simulations of alternative structural reforms in China
See Figure B.1 and B.2 pages 47 and 48.
45
Table A.3 - Main parameters
Name Description Value
u
g
Probability to survive 0.966
| Discount factor 0.99
k Leisure preference 0.3
w
p
v
E
o
ç
o
t
µ
R
o
F
R
v
steril
µ
Bg
Bg
target
q
|
u
w
u
p
t
k
¸
jj
o
Elasticity of substitution on the labor market
Elasticity of substitution on the goods market
Share of equities eligible to life-insurance
Share of capital in the production
Share of capital financed by external funds (at the steady-state)
In?ation parameter in the Taylor rule
smoothing parameter in the Taylor rule
Official reserves parameter in the Chinese augmented Taylor rule
Degree of sterilization
Smoothing parameter for government debt level
Government debt target
Share of import in consumption
Elasticity of substitution in trade
Wages rigidity parameter
Prices rigidity parameter
Rate of subsidy on corporate investment
Proxy of capital mobility between country j and j
Depreciation rate of the capital
77
0.9
0.3
0.6
a
2
0.8
0.1
0.1 or 0.9
0.95
0.6
b
0.1
c
2
0.75
0.67
0.08
d
0.5 or 0.9
e
0.02
a
ç
b
0
= 0.3 in China before the financial liberalization shock. in China
before the government spending reform.
c
In the three countries model, imports are shared equally between the two other regions.
d
In China before the financial liberalization shock,t = 0 otherwise. k
e
0.5 for low capital mobility and 0.9 for high capital mobility.
46
Figure B.1 - Financial liberalization
Impact on China
0.45
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
pibchiC
0.25
0.2
0.15
0.1
0.05
0
pibchiI
0.16
0.14
0.12
0.1
0.08
0.06
0.04
0.02
0
÷0.02
perchiinfp
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo
CapMob
Flex
StatQuo
CapMob
Flex
StatQuo
CapMob
Flex
0
÷0.2
÷0.4
÷0.6
÷0.8
÷1
÷1.2
÷1.4
Consumption
pibchiNFA
0.4
0.3
0.2
0.1
0
÷0.1
÷0.2
÷0.3
Investment
perchiRER
0.07
0.06
0.05
0.04
0.03
0.02
0.01
0
In?ation perchiR
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo CapMob Flex StatQuo CapMob Flex StatQuo CapMob Flex
0.5
0.45
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
Net foreign assets
perchiY
0.05
0
÷0.05
÷0.1
÷0.15
÷0.2
Real exchange rate
pibchiBC
per
c
hi
C
Nominal interest rate
pibchibask
0.4
0.3
0.2
0.1
0
÷0.1
÷0.2
÷0.3
÷0.4
÷0.5
5 10 15 20 25 30 0.4540 35 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo CapMob Flex StatQuo CapMob Flex StatQuo CapMob Flex
0.06
0.05
0.04
0.03
0.02
0.01
0
÷0.01
÷0.02
Real GDP
perusaY
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0.1
0.05
0
÷0.05
÷0.1
÷0.15
Trade balance
Impact on the US
pibusaC
0.012
0.01
0.008
0.006
0.004
0.002
0
Official reserves
perusaR
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo
CapMob
Flex
StatQuo
CapMob
Flex
StatQuo
CapMob
Flex
GDP
0
5
10 Consumption
25
15 20 30 35
Nominal interest rate 40
StatQuo CapMob Flex
47
i
n
p
e
r
c
e
n
t
t
o
G
D
P
t
o
G
D
P
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
t
o
G
D
P
i
n
p
e
r
c
e
n
t
t
o
G
D
P
t
o
G
D
P
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
t
o
G
D
P
Figure B.2 - Government spending reform
Impact on China
pibchiC pibchiI perchiinfp
0.3 0 0.09
0.08
0.25 ÷0.005
0.07
0.2 ÷0.01 0.06
0.050.15
÷0.015
0.04
0.1 ÷0.020.03
0.05 ÷0.025 0.02
0.01
0 ÷0.03
0
÷0.05 ÷0.035 ÷0.01
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo
CapMob
Flex
StatQuo
CapMob
Flex
StatQuo
CapMob
Flex
0
÷0.1
÷0.2
÷0.3
÷0.4
÷0.5
÷0.6
÷0.7
÷0.8
÷0.9
Consumption
pibchiNFA
0.3
0.25
0.2
0.15
0.1
0.05
0
÷0.05
Investment
perchiRER
0.04
0.035
0.03
0.025
0.02
0.015
0.01
0.005
0
In?ation perchiR
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo CapMob Flex StatQuo CapMob Flex StatQuo CapMob Flex
0.2
0.15
0.1
0.05
0
÷0.05
Net foreign assets
perchiY
0.02
0
÷0.02
÷0.04
÷0.06
÷0.08
÷0.1
Real exchange rate
pibchiBC
per
c
hi
C
Nominal interest rate
pibchibask
0.2
0.15
0.1
0.05
0
÷0.05
÷0.1
÷0.15
÷0.2
5 10 15 20 25 30 350.340 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo CapMob Flex StatQuo CapMob Flex StatQuo CapMob Flex
Real GDP
0.25
Trade balance
Impact on the US
Official reserves
0.03
0.025
0.02
0.015
0.01
0.005
0
÷0.005
÷0.01
perusaY 0.2
0.15
0.1
0.05
0
0.02
0
÷0.02
÷0.04
÷0.06
÷0.08
÷0.1
pibusaC
6x10
5
4
3
2
1
0
÷1
÷3
perusaR
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo
CapMob
Flex
StatQuo
CapMob
Flex
StatQuo
CapMob
Flex
GDP
÷0.05
5
10 Consumption
25
15 20 30 35
Nominal interest rate 40
StatQuo CapMob Flex
48
i
n
p
e
r
c
e
n
t
t
o
G
D
P
t
o
G
D
P
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
t
o
G
D
P
i
n
p
e
r
c
e
n
t
t
o
G
D
P
t
o
G
D
P
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
t
o
G
D
P
C. Robustness of the model
We now test the robustness of our model by changing some parameters. To do so, we simulate the pen- sion
reform shock with three different sets of parameters, successively, and compare these simulations with the
"StatQuo" simulation of Section 4.
Elasticity of labor supply. First, we changek (the share of leisure in utility, see Equation (1)) from 0.3 to 0.2.
According to the labor supply (Equation (12)) the elasticity of labor supply to wage is given
by:
k C
L
t
,W
t
= 1 ÷k MRSt L tt
Thus, an increase ink involve a decrease in the elasticity of labor supply. As we can see in Figure C.3, the
increase in the real wage involve a less important increase of the labor supply in the case wherek is greater,
but the differences remain limited.
3.5
3
chi
C
Figure C.3 - Pension reform with different elasticities of labor supply
Impact on China
1.2
1
0.8
0.6
0.4
0.2
0
÷0.2
perchiL 2.5
2
1.5
1
0.5
0.25
0.2
0.15
0.1
0.05
0
÷0.05
÷0.1
÷0.15
÷0.2
÷0.25
perchiw
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
÷0.05
perchiinfp
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo (k=0.3) k=0.2 StatQuo (k=0.3) k=0.2 StatQuo (k=0.3) k=0.2
Labor supply
0
5 10 Real wage
15 20 25 30 35 40 In?ation
StatQuo (k = 0.3) k = 0.2
Elasticity of substitution between goods. The calibrated elasticity of substitution between goods
produced in China and goods produced in the US (|) is set to 2. Figure C.4 shows the impact of changing this
parameter to 5.
A rise in this parameter (both in China and USA) involves more competition between goods of the two
countries. Thus, the pension reform shock involves a larger decrease in China's trade balance and a more limited
real appreciation of the renminbi.
49
l
e
v
e
l
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
chi
C
3.5
Figure C.4 - Pension
3
eform with different elasticities of substitution of goods r
Impact on China
0.2
0.1
0
÷0.1
÷0.2
÷0.3
÷0.4
÷0.5
÷0.6
2.5pibchiBC
2
1.5
1
0.5
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
÷0.1
÷0.2
perchiRER
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo (o=2) o=5 StatQuo (o=2) o=5
Trade Balance
10
0
5
15 20 Real exchange rate
25 30 35 40
StatQuo (o = 2) o=5
Share of imports in consumption The calibrated trade openness of the US and China is set to 10%.
Figure C.5 shows the impact of changing this parameter to 20%. The pension reform has now
more limited impact on the real exchange rate and
chi
on the export level, but the results are qualitatively C
unchanged. 3.5
Figure C.5 - Pension
3
reform with different shares of imports in consumption
Impact on China
0.5
0
÷0.5
÷1
÷1.5
÷2
÷2.5
2.5
2
1.5
1
0.5
perchiX
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
÷0.1
÷0.2
perchiRER
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo (,C=0.1) ,C=0.2 StatQuo (,C=0.1) ,C=0.2
Exports
0
5
10 15 20
Real exchange rate
25 30 35 40
StatQuo (,
C
= 0.1) ,
C
= 0.2
50
l
e
v
e
l
i
n
p
e
r
c
e
n
t
t
o
G
D
P
l
e
v
e
l
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
LIST OF WORKING PAPERS RELEASED BY CEPII
51
doc_283837943.docx
The rebalancing of investments is the action of bringing a portfolio that has deviated away from one's target asset allocation back into line. Under-weighted securities can be purchased with newly saved money; alternatively, over-weighted securities can be sold to purchase under-weighted securities.
FINANCIAL STUDY ON REBALANCING GROWTH IN CHINA: AN
INTERNATIONAL PERSPECTIVE
TABLE OF CONTENTS
Non-technical summary . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4Résumé non
technique . . . . . . . . . . . . . . . . . . . . . . . . . . .
5Résumé court
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
61. Introduction
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
72. China's
saving rate under scrutiny . . . . . . . . . . . . . . . . . . . . . 10
3. The model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3.1. General overview . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3.2. Households . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3.3. Production, investment and nominal rigidities . . . . . . . . . . . . . . . 15
3.4. Government . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.5. Trade . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
4. Structural policies in China . . . . . . . . . . . . . . . . . . . . . . . . 19
4.1. A pension reform . . . . . . . . . . . . . . . . . . . . . . . . . . 19
4.2. Other structural reforms . . . . . . . . . . . . . . . . . . . . . . . 23
4.3. Structural reforms in China under a ZIRP in the US . . . . . . . . . . . . 25
5. The impact of a monetary reform in China . . . . . . . . . . . . . . . . . . 26
5.1. A Pension "reverse" shock . . . . . . . . . . . . . . . . . . . . . . 28
5.2. Reserve accumulation shock . . . . . . . . . . . . . . . . . . . . . . 30
6. A third country: the Euro area . . . . . . . . . . . . . . . . . . . . . . . 32
6.1. Impact of a pension reform in China in a three-country setting . . . . . . . . 33
6.2. Impact of a diversification of China's official reserves . . . . . . . . . . . . 34
6.3. Impact of a diversification of China's official reserves under a ZIRP . . . . . . 34
7. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
A. Details of the model . . . . . . . . . . . . . . . . . . . . . . . . . 41
B. Simulations of alternative structural reforms in China . . . . . . . . . . . . 45
C. Robustness of the model . . . . . . . . . . . . . . . . . . . . . . . 49
List of working papers released by CEPII . . . . . . . . . . . . . . . . . . . . 51
2
N O N - T E C H N I C A L S U M M A RY
Since the early 2000s, China has been the center of repeated criticism by the international community
concerning its export-oriented growth strategy based on an artificially weak currency. However China has
responded to this criticism through structural reforms rather than monetary reforms. This paper aims at
analyzing the complementarity or substitutability of structural and monetary reforms in China, for the
domestic economy as for the global rebalancing agenda. Specifically, we study the impact of a several
reforms that all involve a reduction in Chinese saving rate, under various regimes for capital ?ows and the
exchange rate; reciprocally, we analyze the rebalancing impact of a change in China's monetary regime, absent
structural reforms, hence in a situation of cumulated surpluses in China.
To this end, we build a micro-founded macroeconomic model with overlapping generations, nominal
rigidities and financial constraints, for two countries: China, and the United States. The structural re- forms
successively considered are an increase in the generosity of China's pension system, a financial reform that
facilitates the access of private companies to debt markets while removing capital subsidies, and an increase
in the government debt target. As for monetary regimes, they include the status quo (a fixed exchange rate
backed by capital controls and foreign-exchange interventions), a relaxation of capital controls, and a move
to a ?exible exchange-rate regime. We successively consider the impact of a structural reform under
different monetary regimes, and the impact of a monetary regime change in itself, in a context of global
imbalances. Finally, we introduce a third country (the Euro area) in the model in order to study the
asymmetries stemming from China's bilateral peg on the dollar and from its preference for dollar-denominated
assets.
We find that a fall in China's saving rate would contribute to global rebalancing whatever the exchange rate
regime, provided international capital ?ows do react to interest-rate differentials (which implies less-than-
complete capital controls in China). Under a ?exible exchange-rate regime, the reduction of the bilateral
current-account imbalance between the United States and China is quicker (because the renminbi
appreciates in nominal terms against the US dollar), but it is not stronger than under a fixed exchange-rate
regime. Still, only under a ?exible exchange-rate regime is China able to control in?ation stemming from more
dynamic domestic demand. This feature provides strong incentive for China to move away from its fixed
peg, although the advantage for the United States of such move is only short lived.
We also find that, should the United States refrain from hiking its interest rate when global savings are
reduced, then the rebalancing would be muted and the exchange-rate regime of China would become crucial
both for global rebalancing and for the domestic Chinese economy. As for the Euro area, despitethe
asymmetric exchange-rate regime of China (where the renminbi is pegged on the dollar), it reacts in a
similar way as the United States to structural reforms in China. The reason is high capital mobility
3
between the United States and the Euro area, that tends to equalize real interest rates between the two
countries.
Absent structural reforms reducing the aggregate saving rate of China, a move to a ?exible renminbi is
found unable to rebalance the global economy, except if the People's Bank of China decides to sell off its
accumulated reserves. The reason is that high savings in China, combined with capital controls, result in a
relatively low policy interest rate: removing capital controls while allowing the Chinese currency to ?oat
could then result in a sudden capital out?ows and depreciation of the renminbi in the short run, although in the
longer run the Chinese currency would appreciate.
On the whole, our results suggest that the contribution of China to global rebalancing should primarily rely
on structural policies aiming at reducing aggregate savings in China and on a relaxation of capital controls.
The role of the exchange rate regime would be minor under standard monetary policies in advanced
economies, although more important if monetary policies are constrained. Finally, relying only on a
change in China's monetary regime could end up in delaying rather than accelerating the rebalancing,
depending on China's policy regarding accumulated reserves.
A B S T R AC T
Based on simulations of an original DGE model of the US, Chinese and Euro area economies with
financial frictions and various monetary regimes, the paper shows that the contribution of China in global
rebalancing should primarily rely on structural policies aiming at reducing aggregate savings in China. The
role of the exchange-rate regime would be minor under standard monetary policies, although more important if
monetary policies in advanced countries are constrained, as they are today. Finally, relying only on a change
in China's monetary regime (without structural reforms) could end up in delaying rather than accelerating the
rebalancing, depending on China's policy regarding accumulated reserves.
JEL Classification: F32, F42, F47.
Keywords: Global imbalances, exchange rate regimes, capital controls, China.
4
RÉQUILIBRAGE DE LA CROISSANCE EN CHINE: UNE PERSPECTIVE INTERNATIONALE
RÉSUME NON TECHNIQUE
Depuis le début des années 2000, la Chine s'est trouvée au centre de critiques répétées, de la part de la
communauté internationale, au sujet de sa stratégie de croissance extravertie fondée sur une mon- naie
sous-évaluée. La Chine a répondu à ces critiques par des réformes structurelles plutôt que mo- nétaires.
Nous étudions ici dans quelle mesure les deux types de réformes sont complémentaires ou substituables,
pour l'économie chinoise et pour le rééquilibrage de l'économie mondiale. Plus spécifi- quement, nous
étudions l'impact de différentes réformes structurelles qui toutes conduisent à une baisse du taux d'épargne
agrégé en Chine, selon différents régimes en matière de ?ux de capitaux et de taux de change. A l'inverse, nous
analysons l'impact d'une réforme monétaire sur les déséquilibres mondiaux,
en l'absence de réformes structurelles, donc dans une situation d'accumulation d'excédents en Chine.
A cette fin, nous construisons un modèle à génération imbriquées avec fondements microéconomiques,
rigidités nominales et contraintes financières, pour deux pays : la Chine et les États-Unis. Les réformes
structurelles successivement envisagées sont une augmentation de la générosité du système de retraite par
répartition, une réforme financière permettant aux entreprises un accès plus facile aux financements externes,
les subventions au capital étant simultanément supprimées, et, enfin, un relèvement de l'objec- tif de dette
publique. Trois régimes monétaires sont successivement étudiés : le statu quo (taux de change fixe, contrôles de
capitaux, interventions de change), un relâchement des contrôles de capitaux et un ré- gime de ?ottement sans
contrôles de change. On étudie successivement l'effet des réformes structurelles selon les trois régimes
monétaires, puis l'effet d'une réforme monétaire seule, dans un contexte d'ex- cédents accumulés. Enfin, on
introduit un troisième pays (la zone euro) dans le modèle afin d'analyser les asymétries découlant de l'ancrage
bilatéral du renminbi sur le dollar et de la préférence de la banque centrale pour les réserves en dollars.
Les résultats montrent que, quelque soit le régime de change, une baisse du taux d'épargne en Chine est de
nature à contribuer au rééquilibrage mondial, pourvu que les ?ux internationaux de capitaux réagissent aux
écarts de taux d'intérêt, ce qui suppose un relâchement des contrôles de capitaux en Chine. En régime de
change ?exible, la réduction du déséquilibre bilatéral entre la Chine et les États-Unis est plus rapide (grâce à
l'appréciation nominale du renminbi par rapport au dollar) mais elle n'est pas plus marquée qu'en change
fixe. Cependant, seul le régime de change ?exible permet à la Chine de contrôler son taux d'in?ation face à la
hausse de la demande interne. Cette particularité constitue une puissante incitation, pour la Chine, à
abandonner le régime de change fixe, alors que les bénéfices de la ?exibilité sont de courte durée pour les
Etats-Unis. Cependant, si les Etats-Unis rechignent à relever leur taux d'intérêt lorsque l'épargne mondiale
diminue, alors le rééquilibrage mondial est faible et le régime de change en Chine en devient un élément
essentiel. En dépit de l'asymétrie du régime de change chinois (le renminbi est ancré sur le dollar, non sur
l'euro), la zone euro réagit de façon très similaire aux Etats-Unis à une réforme en Chine. La raison en est la
forte mobilité des capitaux entre Etats-Unis et zone euro, qui tend à égaliser les taux d'intérêt réels dans les
deux zones.
5
En l'absence de réformes structurelles réduisant le taux d'épargne en Chine, le passage au ?ottement ne
permet pas, en lui-même, de réaliser le rééquilibrage mondial, sauf si la Banque du Peuple Chinois décide de
vendre ses réserves accumulées. En effet, le haut niveau de l'épargne en Chine, combiné aux contrôles de
capitaux, conduit à un taux d'intérêt relativement faible dans ce pays. Dans ces conditions, la levée des
contrôles de capitaux concomitante au passage au ?ottement conduit à une sortie nette decapitaux, donc à
une dépréciation du renminbi à court terme, même si à plus long terme la monnaie chinoise s'apprécie bien.
Finalement, nos résultats suggèrent que la contribution de la Chine au rééquilibrage mondial devrait re- poser
essentiellement sur des réformes structurelles réduisant le taux d'épargne agrégé de ce pays et sur un
relâchement des contrôles de capitaux. Le rôle du régime de change est secondaire lorsque les poli- tiques
monétaires des pays avancés sont standard, même s'il devient plus important lorsque les politiques monétaires
sont contraintes. Enfin, fonder le rééquilibrage mondial seulement sur une modification du régime de change
chinois pourrait retarder plutôt qu'accélérer le rééquilibrage, selon la politique suivie par la Chine quant au
devenir de ses réserves accumulées.
R É S U M É C O U RT
A partir d'un modèle dynamique d'équilibre général à trois pays (Etats-Unis, Chine, Zone euro) avec frictions
financières et plusieurs régimes monétaires possibles, nous montrons que la contribution de la Chine au
rééquilibrage mondial devrait reposer principalement sur des réformes structurelles ayant pour effet une
baisse du taux d'épargne agrégé en Chine et de relâcher les contrôles de capitaux. Le rôle du régime de
change est secondaire en présence de politiques monétaires standard dans les pays avancés mais devient
plus important si les politiques monétaires des pays avancés sont contraintes par un taux d'intérêt plancher.
Enfin, miser sur une simple modification du régime monétaire chinois (sans réformes structurelles) pourrait
retarder au lieu d'accélérer le rééquilibrage, selon la politique suivie au regard des réserves accumulées.
Classification JEL : F32, F42, F47.
Mots clés : Déséquilibres mondiaux, régime de change, contrôle des capitaux, Chine.
6
-
1. I N T RO D U C T I O N
Since the early 2000s, China has been the center of repeated criticism by the international com-
munity concerning its export-oriented growth strategy based on an artificially weak currency. In
2005, Ben Bernanke famously pointed the aggregate trade surplus of emerging countries as one
key explanation for the growing US deficit: corresponding capital in?ows in the US was preventing
upward adjustment of US savings rates through maintaining real interest rates at a low level.
2
Although few economists and officials did endorse Bernanke's "savings glut" ap- proach in full
(Bernanke himself recognized the own responsibility of the United States), the G7 consistently
called for a reduction in current-account surpluses especially in emerging Asia, and its leaders
queued up in Beijing to try to persuade Chinese officials to allow the renminbi to appreciate.
Global imbalances temporary retreated during the 2007-2009 global crisis, as a consequence of
the trade collapse and financial de-globalization. When the global economy started to pick up in
mid-2009, it however became clear that the crisis had not wiped up the problem which was
deemed to soon re-emerge (see Blanchard & Milesi-Ferretti (2010)). The G20, which at that time
had replaced the G7 as the main forum for international coordination, refocused the debate on current-
account imbalances themselves, as opposed to the G7 insistence on exchange rates.
3
China
responded by announcing far-reaching structural reforms aiming, in particular, at im- proving
social safety nets, which would allow Chinese households to reduce their savings rates. However
China failed to satisfy advanced economies through letting its currency appreciate, despite
continuous pressure.
1
We are grateful to the participants of the 7 december 2010 at CEPII and the 8 March 2011 RES seminar at the IMF,
for useful comments. All errors remain ours. This paper has been prepared for the Bruegel-CEPII research project
"Reform options for the global reserve currency system and their implications for the EU economy" for the European
Commission (contract No. ECFIN/220/2010/573686). Financial support by DG ECFIN of the European Commission is
gratefully acknowledged.
7
This paper aims at analyzing the complementarity or substitutability of structural and monetary
reforms in China, for the domestic economy as for the global rebalancing agenda. Specifically, we
study the impact of a several reforms that all involve a reduction in Chinese saving rate, under
various regimes for capital ?ows and the exchange rate; reciprocally, we analyze the rebalancing
impact of a change in China's monetary regime, absent structural reforms, hence in a situation of
cumulated surpluses in China.
Our analysis follows-up on the pre-crisis literature on global imbalances and adjustment sce-
narios. Two groups of explanations have been suggested to global imbalances (Bernanke, 2005;
Dooley et al., 2003; Mendoza et al., 2007; Caballero et al., 2008): (i) macroeconomic and
structural factors, such as high saving rates in emerging countries triggered by energy wind- falls,
deficient social safety nets, or financial under-development, low investment rates in some emerging
countries following the Asian crisis, or lax monetary policy in the United States; and (ii) the
international monetary system (IMS hereafter) itself, through the key role of the United States as a
supplier of international reserve assets, the lack of trust in multilateral financial safety nets which led
emerging countries especially in Asia to self-insure through reserve accumula- tion, and the success
of export-oriented growth strategies that encouraged fixed or quasi-fixed exchange-rate regimes.
Contrasting with this two-pillar understanding of global imbalances, researches on rebalancing
scenarios have tended to focus on real exchange-rate adjustment, es- pecially for the United States,
abstracting from the IMS (Obstfeld & Rogoff, 2005; Blanchard et al., 2005). Few studies have
been devoted to the interconnection between real adjustments and monetary regimes. One
exception is Blanchard and Giavazzi (2006) mentioning the needs to combine a nominal
appreciation of the renminbi with a fall in China's saving rate, in order to secure internal balance
while reducing the current-account surplus. Another one is Faruqee et al. (2007) who contrast the
impact of a fiscal adjustment in the US depending on China's monetary regime, within a four-
region dynamic general equilibrium (DGE) model.
The present paper is interested in the interconnection between global rebalancing and monetary
regimes. The main case in point being the Chinese economy, where excess savings result in
official reserve accumulation due to the fixed exchange-rate regime, we choose to focus on this
country, while acknowledging that China is by no way the only contributor to global imbal-
ances. According to Prasad (2009), China accounted for 62 percent of gross national savings in
Asia excluding Japan in 2008, which provides an additional motivation for focusing on Chi- nese
savings. We start by brie?y reviewing the literature that has tried to explain the very high gross
saving rate in China (50% of GDP on average between 2005 and 2008, according to Ma & Yi
(2010)). We then build a DGE model of two countries (China and the United States) with
overlapping generations, nominal rigidities and financial constraints, and simulate differ- ent
structural reforms in China that all result in declining saving rates, under different monetary regimes.
The reforms successively considered are an increase in the generosity of China's pen- sion system, a
financial reform that facilitates the access of private companies to debt markets, and an increase in the
government debt target. As for monetary regimes, they include the status quo (a fixed exchange rate
backed by capital controls and foreign-exchange interventions), a
8
relaxation of capital controls, and a move to a ?exible exchange-rate regime. We successively
consider the impact of a structural reform under different monetary regimes, and the impact of a
monetary regime change in itself, in a context of global imbalances. Finally, we introduce a third
country (the Euro area) in the model in order to study the asymmetries stemming from China's
bilateral peg on the dollar and from its preference for dollar-denominated assets.
We find that a fall in China's saving rate would contribute to global rebalancing whatever the
exchange rate regime, provided international capital ?ows do react to interest-rate differentials (which
implies less-than-complete capital controls in China). Under a ?exible exchange-rate regime, the
reduction of the bilateral current-account imbalance between the United States and China is quicker
(because the renminbi appreciates in nominal terms against the US dollar), but it is not stronger than
under a fixed exchange-rate regime. Still, only under a ?exible exchange- rate regime is China able to
control in?ation stemming from more dynamic domestic demand, consistent with the intuition of
Blanchard & Giavazzi (2006). This feature provides strong incentive for China to move away
from its fixed peg, although the advantage for the United States of such move is only short lived.
We also find that, should the United States refrain from hiking its interest rate when global
savings are reduced, then the rebalancing would be muted and the exchange-rate regime of China
would become crucial both for global rebalancing and for the domestic Chinese economy. As for the
Euro area, despite the asymmetric exchange-rate regime of China (where the renminbi is pegged on
the dollar), it reacts in a similar way as the United States to structural reforms in China. The reason
is high capital mobility between the United States and the Euro area, that tends to equalize real
interest rates between the two countries.
Absent structural reforms reducing the aggregate saving rate of China, a move to a ?exible
renminbi is found unable to rebalance the global economy, except if the People's Bank of China decides
to sell off its accumulated reserves. The reason is that high savings in China, combined with capital
controls, result in a relatively low policy interest rate: removing capital controls while allowing the
Chinese currency to ?oat could then result in a sudden capital out?ows and depreciation of the
renminbi in the short run, although in the longer run the Chinese currency would appreciate.
On the whole, our results suggest that the contribution of China to global rebalancing should
primarily rely on structural policies aiming at reducing aggregate savings in China. The role of the
exchange-rate regime would be minor under standard monetary policies, although more important if
monetary policies in advanced countries are constrained. Finally, relying only on a change in
China's monetary regime could end up in delaying rather than accelerating the rebalancing,
depending on China's policy regarding accumulated reserves.
The remaining of the paper is organized as follows. In Section 2, we review the literature on the
determinants of China's high savings rate. Section 3 presents the model. In Section 4, we
simulate structural reforms in China under different monetary regimes. Section 5 studies
9
Figure 1 - Saving
a
rates
e
in
0
selected countries
Saving, verag 2 05-2007
50
40
30
20
10
Gov. 6,40
0
-10
Source: Ma & Yi (2010)
the impact of a monetary reform in China after the latter has accumulated a large amount of
official reserves. Section 6 adds the Euro area to the analysis and discusses the impact of a
diversification of China's official reserves. Section 7 concludes.
2. CHINA'S SAVING RATE UNDER SCRUTINY
The very high saving rate in China has attracted much attention from both policy-makers and
researchers. Ma & Yi (2010) note that gross national savings reached 54 percent of GDP in 2008,
up from 39 percent in 1990 and 36 percent in 2000. They argue that what is special in China is
the conjunction of high saving rates in the three sectors of the economy: households, corporations,
and the government (see Figure 1). Hence, explanations for the high level of the aggregate saving
rate must address the three sectors simultaneously in a consistent way.
Many explanations have been suggested for high saving rates in China.
4
They can be summa-
rized as follow:
• Households: The steady decline in the share of households in national income during the
2000s has been a major contribution to the fall in the consumption-to-GDP ratio and sub-
sequent rise in the saving-to-GDP one. It can be related to the downward pressure on labor
income (especially in farming and unincorporated activities), a reduction in social transfers, a low
level of capital income (low interest rates), and growing taxes and social contribu- tions. As
for the rise in households saving rates (as a percentage of disposable income), it
4
See, e.g., Ma & Yi (2010), Guo & N'Diaye (2010), Jha et al. (2009), Horioka & Wan (2007), Feng et al. (2009),
Chamon & Prasad (2010), Baker & Orsmond (2010), Cui & Aziz (2007), Chamon et al. (2010), Prasad (2009),
Bayoumi et al. (2010), Huang & Wang (2010).
10
H
o
u
s
e
h
o
l d
s
:
2
1
,
5
0
%
o
f
G
D
P
C
o
r
p
o
r
a
t
i o
n
s
:
2
0
,
4
0
U
S
T
a
i
w
a
n
P
h
i
l
F
r
a
n
c
e
G
e
r
m
a
n
y
J
a
p
a
n
K
o
r
e
a
I
n
d
i
a
C
h
i
n
a
is generally attributed to the attrition of the social safety nets (the retreat of enterprise-based
safety nets having not been taken over by the public system), together with a general rise in
uncertainty due to structural change and the rising share of education expenditures falling on
families. The rise in private home ownership and the 1997 pension reform (with a re- duction
in its generosity) have also played a role, in combination with the delayed effect of the one-
child policy (and the subsequent lack of support expected for the old age). All these factors
were amplified by financial underdevelopment, which has limited the access of households to bank
loans as well as the diversification opportunities for their assets. In con- trast, the impact of ageing
is generally downplayed by the literature, saving rates displaying an uncommon U-shape along
the life cycle.
• Corporations: as a mirror effect, corporations have reaped a growing share of national in-
come during the 2000s. Their high profitability has also come from subsidized inputs such as land,
energy and environment. State-owned enterprises were not required to pay dividends on their
profits before 2008. As for private firms, they may have faced financial constraints that have
obliged them to accumulate savings to self-finance their investments. Although corporations do
have seen their share in national income increase over the decade, the reality of excess savings
has however been challenged in the recent literature based on firm-level data.
• Government: the shares of the government in national income and national saving rose more
than those of the corporate sector in the 2000s. One major reason was the surge in tax receipts that
was not matched by equal increase in expenditures. In fact, government consumption declined
as a proportion of GDP, since the priority was (and local incentives were) given to public
investment and the building-up of public-pension assets (consistent with the switch decided in
1997 in favor of a partially-funded pension assets), and taxes were not much redistributed to
local governments that were in charge of social spending.
In the following, we propose a model that allows for some of the main causes of high saving
rates to be mitigated:
• A pension reform: according to Chamon et al. (2010), "rising income uncertainty and pen-
sion reforms can account for over half of the increase in the urban saving rate in China since the
mid-1990s" (p.1). Consistently, we simulate a pension reform that would increase the
generosity of the pay-as-you-go system. By extension, this simulation covers any reform of the
social safety net that would reduce uncertainty about the ability of households to cover "big
ticket" expenses (such as health care).
• A financial reform: much of the literature (see, e.g., Cui & Aziz (2007); or Ma & Yi (2010))
stresses the declining share of households' share in national GDP as a major cause of the rise in
China's aggregate saving rate. Consistently, we simulate a financial reform in China that relies on
three pillars: (i) enterprises get the opportunity to massively increase leverage, which raises the
value of the firms; (ii) simultaneously, the subsidy on corporate investment is removed; (iii) the
combination of higher firm value and lower incentive to invest leads
11
to a transfer of income from the corporate sector to the households one, possibly transiting
through a public body.
• A government spending reform: we finally assume a permanent increase in public spending
that is engineered through a rise in the public-debt target. Subsequent public transfers are
channelled to old households through the PAYG pension system. In the long run, the tax rate
increases to stabilize the debt ratio at its new in?ated target level.
These various simulations are run under different regimes in terms of exchange-rate regime
(either fixed or ?exible) and capital ?ows (restricted or freed), and the results are systematically
compared. It can be argued however that the problem today is less that of carrying out structural
reforms that will reduce China's saving rate and ultimately contribute to global rebalancing, than to
allow the exchange rate to correct accumulated disequilibria. Consistently, we run a reverse
simulation where a reduction in the generosity of the pension system (like after the 1997 reform) or an
increase in the official reserve target that both raise China's saving rate and trade balance, under fixed
exchange-rate regime. After a few years, the exchange rate is allowed to ?oat freely, and we study the
impact of the sole monetary reform on the curent account and the economy.
3. THE MODEL
3.1. General overview
The world is divided into two countries: the United-States (u) and China (c). In an extension of the
model, we include a third region: the Euro area (e). Countries trade both goods and financial assets.
The model explicitly takes into account the saving behavior of the three main economic
sectors: households, firms and the public sectors.
5
The model is depicted in Figure 2, and detailed in Appendix, together with its standard cali-
bration. Each country is populated with overlapping generations of households à la Blanchard
(1985) and Yaari (1965). On aggregate, households supply labor, receive wages and the pro-
ceeds of their savings (interests and dividends), pay taxes on their labor endowment and receive
transfers from the government. Depending on their age, the ?ow of funds differs: as they be- come
older, households accumulate financial wealth (hence they receive more dividend and interest
payments), their labor endowment decreases (labor income decreases), and they receive rising
transfers from the PAYG pension system. They also benefit from a complete set of contin- gent
securities (perfect risk sharing among alive households). In addition, when they die, their bequest
(financial assets) is redistributed to surviving households through an insurance scheme. The pension
system is introduced as the main policy tool that in?uences aggregate saving of households: the
higher the replacement rate, the lower the saving rate for a given real interestrate. As already
mentioned, this stylized pension system can account for any insurance scheme (e.g. health
insurance) that will reduce the incentive for precautionary savings.
5
We
do not distinguish the financial from the households sector: firms are assume to borrow directly from domes-
tic or foreign households.
12
There are two categories of firms. The first category hires labor L and capital K to produce goods
Y that are indifferently used for consumption and investment. These goods are differen- tiated and
sold to domestic and foreign customers under monopolistic competition. The firms face a price
rigidity à la Calvo (1983): at each period, each firm has a fixed probability of being unable to reset its
price. Wages are also sticky with staggered contracts.
The second category of firms specializes in accumulating capital K and renting it to the first
category of firms. Capital accumulation can be financed either by diverting dividends (internal
financing) or by borrowing Bf on the bond market (external financing). In China, firms' borrow- ings are
constrained by financial frictions. The saving behavior of capital firms is in?uenced by these
constraints: absent external financing, firms have to self-finance their investments, hence they pay low
dividends and their saving rate is high. With financial liberalization, firms can bor- row more, hence
they save less and distribute higher dividends. In China, the savings behavior
of capital firms is also in?uenced by public subsidies on capital.
6
The government receives contributions from households, positive or negative transfers from the
central bank, issues debt, provides subsidies to capital firms and pays pensions to old house- holds.
The pension level is determined by the government budget constraint, with exogenous debt target,
tax rate and, in the case of China, capital subsidies.
Finally, the central bank sets the interest rate according to a modified Taylor rule. In the United
States, the interest rate simply reacts to in?ation deviations from target.
7
In China, the rule is
augmented with an impact of official reserve accumulation. To (partially) sterilize its interven- tions,
the central bank issues sterilization bonds Bc. Depending on the amount of sterilization and on
return differentials between foreign-exchange reserves (which yield the foreign policy rate) and
sterilization bonds (which yield the Chinese policy rate), the central bank generates a positive or
negative profit that is immediately transferred to the pension system, hence indirectly transferred to old
households.
In the benchmark model, there are capital controls in China: foreign investors cannot buy as
many yuan-denominated assets as they would like, and Chinese investors cannot buy as many
foreign-denominated assets as they would like: the uncovered interest parity does not hold. This
grants the Chinese central bank some independence in the conduct of monetary policy even
when the exchange rate is fixed. Alternatively, capital controls can be relaxed, in which case UIP
does hold, impeding China from monitoring domestic in?ation unless the renminbi is allowed to ?oat.
The key equations are presented in the next sub-sections, the model being detailed in Appendix.
6
In our model, investment is restricted to the corporate sector. Alternatively, capital firms could be viewed as
intermediaries investing in real estate and renting their capital to the households. Aggregating households and capital
firms would then reproduce the impact of financial frictions on households' saving, in relation with housing.
7
Including the output-gap in the monetary rule does not change the results qualitatively, although it complicates the
resolution since the model then needs to be simulated at first with nominal rigidities in order to derive potential output.
13
Figure 2 - Diagram of the model
Bh *
B ond
market
FR
Central bank
(r,FR,Bc)
div
Pens. syst.
(tax,pens)
Bc tax pe ns
Bg
Households
Bh *
B ond
market
Bh
Households
(C,w)
I
Bk
Capital firms
(I,Bk)
K
div
Prod. Firms
L
Product
market
X,M Product
market
Y (p)
C
Foreign Home
3.2. Households
The representative household of the cohort born at time a and still living at time t consumes C
a,t
,
has a time-endowment for work of L
t
÷
a
hours, and effectively works L
a,t
.
8
At a given period, ¯
each previously-born household has a constant probabilityu
g
to survive, and new households
appear at rate (1 ÷u
g
)N , where N is the population, which thus is constant. Let 0<|< 1
¯ ¯
denote the subjective discount factor. The inter-temporal utility function of household a at time
t, U
a,t
, is increasing in consumption and in leisure:
·
U
a,t
=
s=0
(u
g
|)
s
(1÷k) log(C
a,t
+
s
) +k log(L
t
+s÷
a
÷ L
a,t
+
s
) , ¯ (1)
with 0<k< 1.
In each country, the portfolio A
t
of households consists in (i) domestic equities (ii) a complete
set of domestic contingent claims and (iii) foreign assets in the form of short-term bonds.
9
Life-
insurance companies à la Yaari (1965) pay a premium to alive households against the promise of
inheriting their assets when they are dead. We assume that insurance companies always accept
contingent claims and bonds (whether domestic or foreign), but are reluctant to insure
equities. The sharev
E
of equities eligible to life insurance is the first indicator of financial-
market development. It leads households to discount future dividends using a discount factor
8
The retirement schedule and associated pension are assumed inverse, geometric functions of the remaining life-
time labor endowment of the household.
9
The complete set of contingent claims are traded only among domestic households so that, when aggregating
households, the net holding is a domestic bond.
14
lowered by the factoru
E
=v
E
+ (1÷v
E
)u
g
< 1. Non-insured equities of dead households are assumed to
be distributed to the newborns. The budget constraint, the first-order conditions at
the cohort level, aggregate consumption and labor supply are detailled in Appendix A.1 (page 41).
The international portfolio of households consists in foreign bonds only: equities and contingent
claims are assumed not to be internationally tradable.
10
Imperfect capital mobility is taken into
account by assuming that domestic households of country j hold an amount Bh
jj
of foreign
bonds (of country j ) such that their share in the portfolio is proportional to the excess expected
return compared to the uncovered interest-rate parity:
E
jj
Bh
j
tj t
A
jt
Share in domestic portfolio
¸
jj
= 1 ÷¸
jj
E
t
jj
u
t
t+1
,j
R
j
t E
t
jj
E
t
+1
UIP deviation
÷1
(2)
where u
t
t+1
,j
is the stochastic discount factor of country j, R
j
t is the nominal interest rate on
foreign bonds, E
jj
is the nominal exchange rate between j and j , and 0<¸
jj
< 1 is a
proxy for capital mobility between the two countries. If¸ ÷ 0, the model reproduces capital controls:
even when the excess expected return is high, the ability of foreign households to buy domestic
assets (or of domestic households to sell foreign assets) is limited. On the contrary, ¸ ÷ 1 denotes
an absence of capital control as a tiny excess return motivates large capital ?ows between the two
countries. In the following, we assume that¸ may take one of the two following values:¸ = 0.5 (low
capital mobility) or¸ = 0.9 (high capital mobility).
3.3. Production, investment and nominal rigidities
Wage rigidities. Unions are used as a way to introduce monopolistic competition and wage
stickiness in the labor market. Specifically, each union is assumed to hire hours from households on a
competitive market. In turn, it rents these hours to the firms on a monopolistic competition market
where labor demand depends on the relative wage charged by the union with an elasticity
of substitution denoted
w
. Wages are sticky à la Calvo, with 1 ÷u
w
denoting the probability
for the union to be able to reset the wage level. The first-order condition on wage setting and
the derivation of aggregate-wage in?ation are derived in Appendix A.2 (page 42).
Production and price rigidities. The representative production firm i hires labor N
t
(i) from the
unions and capital K
t
(i) from capital firms to produce a differentiated good along a Cobb- Douglas
production function Y
t
(i) = A
t
K
t
(i)
o
N
t
(i)
1
÷
o
, where 0<o< 1 and A
t
denotes the
10
This assumption avoids perfect international risk-sharing. Since the stochastic discount factor will then be region- specific,
it is convenient to also assume that equities are not internationally traded in order to properly define the program of the
firms.
15
exogenous level of total factor productivity. The good is sold under monopolistic competition. The
elasticity of substitution of demand is denoted
p
. Calvo-type price stickiness is introduced
so that, at each period, the representative firm is able to reset its price with a probability 1 ÷
u
p
. Factor demand and optimal price setting and aggregate in?ation dynamics are derived in
Appendix A.3.
Investment. Capital accumulation is carried out by specialized firms that rent it to producing firms
on a competitive market. The level of capital K
t
(k) made available by a capital firm k at
the end of period t for the next production period depends on the capital stock at the end of pe-
riod t÷1 and on I
t
(k), the gross investment during period t: K
t
(k) = K
t
÷
1
(k)u
K
Itt÷(k()k
)
÷o , 1
whereo is the depreciation rate andu is a non-decreasing concave function withu(0) =
u (0) = 1 encompassing real rigidities on capital accumulation. Although the market of physi- cal capital
is perfectly competitive, capital firms can make profit due to entry barrier formed by their initial level
of capital.
Due to out-of-model financial frictions, the level of borrowing Bf
t
(k) (negative when actually
the firm borrows) is constrained by the expected value of the firm at the date the loan is paid
back (t+1). This leads to:
ç
÷Bf
t
(k) s
borrowing at t
1֍
E
t
u
t
t
+1
Vk
t
+1
(k)
expected value of the firm at repayment time
, (3)
where Vk
t
(k) is the market value of the capital firm and 0<ç< 1.
11
First-order conditions
on capital accumulation and the financial structure of the firm (debt or equities) are derived in
appendix A.4 (page 44).
3.4. Government
The role of the government is to run the PAYG pension system, distribute capital subsidies
(China only), and to conduct the monetary/exchange-rate policy.
Monetary and exchange-rate policies. In the United States (and, in the extension of the
model, in the Euro area), monetary policy is modeled through a simple interest-rate rule:
R
t
R
t
÷1
log =µ
R
log + (1 ÷µ
R
)o
t
log(t
t
), (4)
R
-
R
-
where R
t
is the nominal interest rate, R
-
its steady-state value,t
t
the production-price in?ation,
o
t
> 1 the long run reaction of interest rate to in?ation andµ
R
= 0.8 a smoothing parameter.
12
11
At
the steady-state,ç will be equal to the share of capital financed by external funds and so it is a second indicator
for financial development.
12
Interest rates and in?ation are expressed as one plus the rate itself.
16
Note that, consistent with the Taylor rule (Taylor (1993)), the real interest rate rises whenever
in?ation increases.
13
Since the model is not used to study cost-push shocks, we can safely drop
the output-gap from the Taylor rule, since the output-gap will be perfectly correlated with
in?ation.
14
This simple feedback rule is augmented in the case of China to account for the fixed exchange-
rate regime.
15
Specifically, the Chinese peg is maintained through foreign-exchange reserve
accumulation in the form of dollar-denominated bonds Bc
u
(and, in the three-country exten- t
sion of the model, of euro-denominated bonds Bc
e
t). These interventions are partially sterilized
through the issuance of sterilization bonds. Denoting by FR
t
> 0 the total amount of official
reserves at the end of period t and by Bc
c
t< 0 the stock of central bank liabilities in terms of
sterilization bonds, we have: Bc
c
t = ÷v
steril
FR
t
, where 0<v
steril
< 1 is the degree of steril-
ization of foreign-exchange interventions. According to Greenwood (2008), the sterilization of
official interventions in China during the 2000s was also channeled through a rise of reserve
requirements of commercial banks. Here the banking sector is not modeled, hence steriliza- tion
is performed only through the issuance of sterilization bonds that are directly (rather than indirectly)
purchased by households. Following Greenwood's evidence of a high degree of ster-
ilization, we setv
steril
= 0.9.
16
However, if China removes capital controls while maintaining its
peg, the amount of interventions becomes very large and it is unlikely to be widely sterilized.
In this case, we setv
steril
= 0.1.
To the extent that official interventions are not fully sterilized, they feed money creation, which
in turn exerts downward pressure on the interest rate. This effect is accounted for here through
adding a second term in the interest-rate rule for China:
R
t
R
t
÷1
log =µ
R
log + (1 ÷µ
R
) [o
t
log (t
t
) ÷ (1 ÷v
steril
)o
FR
(FR
t
÷ FR
-
)] , (5)
R
-
R
-
with FR
-
the target level for official reserves, ando
FR
> 0.
17
In order for the central bank's balance sheet to stabilize in the long run, we assume the cost
(or benefit) from official interventions to be indirectly transferred to households through the
13
For
14
We
the sake of simplicity, it is assumed here that the in?ation target is zero.
have checked the robustness of our results when introducing the output gap in the Taylor rule
15
The ability of central banks to peg exchange rate with partial capital mobility has been discussed by Mundell (1963)
and Obstfeld (1980).
16
This figure also corresponds to Ouyang et al. (2010) who find that almost 90% of reserves accumulation is sterilized
in China.
17
Note that this second term in the interest-rate rule also allows China's net foreign asset position to stabilize in the long run:
the downward pressure on the interest rate when the central bank accumulates foreign-exchange reserves reduces the
incentives of households to save, hence reduces their accumulation of foreign assets. Here, we set
o
FR
= 0.1.
17
pension system. Denoting by DCBt the amount of this positive or negative transfer, we have:
DCBt = E
u
(Bc
u
÷
1
R
u
÷ Bc
u
) + (Bc
c
t÷
1
R
c
t ÷ Bc
c
t÷
1
)
t t t t
(6)
where E
u
denotes the exchange rate of the renminbi against the dollar, R
u
the US interest rate
t t
and R
ct
the Chinese one.
18
Taxes, subsidies and the PAYG pension system. In each country, the government sets the tax
rate on labor endowments (t
t
)
19
and issues domestic-currency denominated risk-free bonds,
the stock of which is denoted Bg
t
at the end of period t, with a negative sign whenever the
government is a debtor. The debt level Bg
t
is set so that the debt ratio adjusts progressively to
a target level Bg
target
:
Bg
t
=µ
Bg
Bg
t
÷
1
+ (1 ÷µ
Bg
)Bg
target
P
t
Y
t
(7)
where P
t
Y
t
represents nominal GDP, andµ
Bg
is set to 0.95. In China under a fixed exchange-rate
regime, the pension system also receives positive or negative transfers from the central bank,
DCBt, and subsidizes capital accumulation (see Appendix A.4). Finally, the level of pensions,
H
t
, is set so as to abide with the budget constraint:
H
t
=t
t
W
t
L
t
+ Bg
t
+ DCBt ÷ R
t
÷
1
Bg
t
÷
1
÷ (t
k
÷ 1)rk
t
K
t
÷1 (8)
3.5. Trade
Domestic demand is the sum of households' consumption and capital firms' investment. This
demand in country j is distributed across domestic goods (D
jt
) and imported goods from country
j (M
t
jj ) according to a Dixit-Stiglitz CES. Imports and domestic demand write:
M
t
jj
=
q
jj
E
jj
P
t
j t
÷|
(I
j
t
+
C
j
) t
and
D
jt
=
1÷
q
jj
P
t
j
÷|
(I
j
+ C
j
),
Pc
jt j =j
Pc
jt
t t
where I
j
, C
j
and Pc
jt
denote investment, consumption and the consumption price index of
t t
country j, respectively, P
t
j
is the producer price index of country j, P
t
j
that of country j ,
18
In
the three-country extension of the model, a third term is introduced on the right-hand side of the equation to
account for earnings on euro-denominated reserves.
19
Taxing labor endowment rather than hours worked allows us to neglect the efficiency cost of taxation. Here, the tax has
no impact on the labor market.
18
|> 0 andq
jj
> 0 represents trade openness of j vis-ï¿
1
-vis j .
20
The trade balance writes, in 2
domestic currency:
TB
jt
=
j =j
P
t
j
M
t
j
j
÷ E
jj
P
t
j
M
t
jj t
We now turn to two series of simulations of the model. In Section 4, we study the impact of
structural policies in China leading to a rise in the national savings rate, under different
monetary regimes. In Section 5, we assume that China stays on a high-saving path and study the
impact of a monetary reform.
4. S T RU C T U R A L P O L I C I E S I N C H I N A
In this section, we study different structural reforms presented in Section 2, all of which are
deemed to impact negatively on China's aggregate savings rate. We contrast the impact of these reforms
depending on China's monetary regime. Specifically, we study three monetary regimes,
successively:
StatQuo The exchange rate is fixed, with capital controls and almost complete sterilization of
interventions;
CapMob Capital controls are relaxed, the nominal exchange rate is kept constant but there is
almost no sterilization;
Flex The renminbi is allowed to ?oat freely with no capital controls, no interventions, hence
no sterilization.
The corresponding key parameters are reported in Table 1.
Table 1 - Monetary and financial parameters under the three regimes
Regime Exchange-rate regime Capital mobility¸ Sterilizationv
steril
StatQuo Fix 0.5 0 .9
CapMob Fix 0.9 0 .1
Flex Flex 0.9 0
4.1. A pension reform
We first consider an increase in the generosity of the Chinese PAYG pension system (a 6.5 pp
increase in the replacement rate) financed through an increase in the tax rate on labor en-
dowment. In our overlapping-generation model, the reform induces a net transfers from future
generations (that will bear a share of higher future pensions through labor income taxation) to
20
We
present here the most general specification of the model that applies whatever the number of countries.
19
alive generations (that benefit from higher future pensions). Hence, aggregate consumption is
boosted in the short run.
21
In autarky, the fall in aggregate saving in the long run would induce a
higher real interest rate, hence lower capital, GDP, income, and consumption. To the extent that
capital controls are not complete, this effect is mitigated by international capital in?ows and the
Chinese net foreign asset position (NFA hereafter) falls in the long run. Higher capital mobility
allows the capital stock to fall less in the long run, thanks to a larger decline in the
NFA reduction, as depicted in Table 2.
22
Dynamic reaction of the Chinese economy. Consumption in China surges following the
shock, but the extent of this surge depends crucially on the monetary and financial regime (see Figure
3). In the status quo regime (currency-peg, low capital mobility), the Chinese interest rate is largely
independent from the US one in the short run: it rises sharply as a reaction to higher in?ation. This
increase in the interest rate mitigates the impact of the reform on consumption (+1% of GDP) and
drives investment downwards (-0.3% of GDP). GDP and employment in- crease in the short run
but the hike is short-lived: higher consumption induces households to reduce their labor, which
leads to a wage increase that feeds in?ation (+0.3 pp); hence the real exchange rate appreciates and
the trade balance deteriorates (-0.3% of GDP). The rise in China's interest rate also attracts foreign
capital in?ows (and reduces incentives for China's residents to invest abroad). Despite low capital
mobility, this change in net capital ?ows exceeds the de- terioration of China's trade balance, so the
central bank slightly increases its foreign-exchange reserves (+1% of GDP) to keep the nominal
exchange rate constant, which in turn mitigates the rise in the interest rate, despite the sterilization
policy. Due to higher in?ation, the real exchange rate appreciates during the first ten quarters, before
falling back to its baseline level.
When capital controls are removed while the peg is maintained (CapMob regime), large capital
in?ows limit the reaction of the interest rate to the shock in the short run. This feeds higher
in?ation (+0.6 pp), a limited fall in investment (-0.06% of GDP) and a large increase in con-
sumption (+2.2% of GDP). The trade balance deteriorates more (-0.6% of GDP). Net capital
in?ows are now much larger. In the short term, this is compensated by larger reserve accumu-
lation. After ten quarters however, the NFA falls much more rapidly than under the StatQuo
regime.
In a ?exible exchange rate regime (with no capital controls), the central bank no longer accu-
mulates foreign-exchange reserves whenever there is a surge in capital in?ows (or a drop in
capital out?ows). Rather, it allows the nominal exchange rate to appreciate. Hence, following
21
The
reform also induces transfers across alive households since elder households benefit immediately without
bearing its cost, whereas younger households bear the bulk of the tax. However, in our overlapping-generation model,
this redistribution drives no change in the aggregate saving rate as the propensity to consume wealth is independent
from age. The transitory increase in aggregate consumption then comes from the fact that part of the cost of the reform
will be borne by future generations.
22
Equity stocks are not reported in the table since equities are not supposed to be traded internationally here.
20
Table 2 - Long-run impact of the pension reform on China's balance sheet
Bonds supply Bonds demand
StatQuo CapMob Flex StatQuo CapMob Flex
Firms -3.0 -2.6 -2.5 Households -4.7 -9.4 -10.3
Government -0.2 -0.1 -0.1
Central bank -0.6 0.1 0 Central bank -0.7 0.8 0
Nation -3.8 -2.6 -2.6 Nation -5.4 -8.6 -10.3
NFA -1.6 -6.0 -7.7
Source: Author calculations.
a rise in the generosity of the Chinese pension system (and the subsequent decrease in house-
holds' savings in China), there is both a rise in the domestic interest rate and an immediate
appreciation of the renminbi.
23
Such policy is successful in curbing in?ation, at the expense of
consumption that grows less than under the CapMob regime (but as much as under the StatQuo one).
Because net exports are also affected by the short-run appreciation of the renminbi, GDP no longer
increases in the short run, neither does employment.In all simulations, employment varies like
GDP, hence this graph is omitted in the Figures. Interestingly, the fall in Chinese exports is
deeper in the short run than under a peg, but after six quarters the level of exports is the same as
under the CapMob regime, and the dynamics of the NFA position is similar in the two regimes. This
is consistent with the one-off appreciation of the renminbi, contrasting with progressive
appreciation through cumulated in?ation differentials under a fixed peg.
If Chinese authorities' objective function is close to that of the households, hence if it weighs
higher consumption as a positive outcome, then the best regime is the CapMob one since this is the
regime yielding the highest increase in consumption. Conversely, if Chinese authorities are mainly
concerned with in?ation, then the pension reform should be carried out under the Flex regime.
Finally, if Chinese authorities consider employment as the main objective, as it has been the case
since the mid 1990s, then our simulations suggests that the peg should be maintained while the
pension reform is being implemented.
Impact on the US economy. The transmission of the shock to the United States comes from
increased imports by Chinese households (which boosts US GDP) and from higher returns on
Chinese assets (which offers an incentive for US households to invest in China and reduces the
incentive of Chinese residents to invest in the US). Since both channels are the two sides of the
same coin, the transmission is magnified when capital is more mobile. Importantly, the rise in US
GDP comes along with a rebalancing of the US economy since both consumption and
investment fall in this country due to a higher interest rate. But only with high capital
23
The
renminbi subsequently depreciates, consistent with the interest-rate differential together with rational expec-
tations.
21
Figure 3 - Impact of a pension reform in China
Impact on China
2.5
2
1.5
1
0.5
0
÷0.5
pibchiC
0
÷0.02
÷0.04
÷0.06
÷0.08
÷0.1
÷0.12
÷0.14
÷0.16
÷0.18
÷0.2
pibchiI
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
÷0.1
perchiinfp
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo
CapMob
Flex
StatQuo
CapMob
Flex
StatQuo
CapMob
Flex
0
÷0.5
÷1
÷1.5
÷2
÷2.5
÷3
÷3.5
÷4
÷4.5
÷5
Consumption
pibchiNFA
1.6
1.4
1.2
1
0.8
0.6
0.4
0.2
0
÷0.2
÷0.4
Investment
perchiRER
0.25
0.2
0.15
0.1
0.05
0
In?ation perchiR
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo CapMob Flex StatQuo CapMob Flex StatQuo CapMob Flex
Net foreign assets perchiY Real exchange rate pibchiBC Nominal interest rate pibchibask
2 0.2 1.5
0.1
11.5 0
÷0.10.5
1 ÷0.2
÷0.3 0
0.5 ÷0.4
÷0.5÷0.5
0 ÷0.6÷1
÷0.7 per
c
hi
C
÷0.5 ÷0.8 ÷1.5
5 10 15 20 25 30 352.540 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo CapMob Flex StatQuo CapMob Flex StatQuo CapMob Flex
0.25
0.2
0.15
0.1
0.05
0
÷0.05
÷0.1
Real GDP
perusaY
2
1.5
1
0.5
0
0.2
0.1
0
÷0.1
÷0.2
÷0.3
÷0.4
÷0.5
÷0.6
÷0.7
Trade balance
Impact on the US
pibusaC
0.045
0.04
0.035
0.03
0.025
0.02
0.015
0.01
0.005
0
Official reserves
perusaR
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo
CapMob
Flex
StatQuo
CapMob
Flex
StatQuo
CapMob
Flex
GDP
÷0.5
5
10 Consumption
25
15 20 30 35
Nominal interest rate 40
StatQuo CapMob Flex
22
i
n
p
e
r
c
e
n
t
t
o
G
D
P
t
o
G
D
P
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
t
o
G
D
P
i
n
p
e
r
c
e
n
t
t
o
G
D
P
t
o
G
D
P
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
t
o
G
D
P
mobility is such rebalancing significant. Under a ?exible regime, the immediate appreciation of the
renminbi stimulates US exports and GDP in the very short run. Simultaneously, the fall in
consumption is larger in the very short run than under a fixed peg, because the interest rate itself
increases more rapidly. After five quarters, however, US exports, consumption and GDP are similar
under the Flex and the CapMob regimes. Hence, except in the short run, a ?exible exchange rate in
China does not make much difference for the United States to the extent that capital controls are
removed.
In the longer run, GDP and consumption fall below the baseline due to the permanent fall in the
investment rate (thus in productive capital). Hence, the positive impact of the Chinese pension
reform is only short lived for the US economy, although it has a permanent effect on the NFA
position.
In Appendix C, robustness checks are performed by comparing the impact of the pension reform
with only one monetary regime (StatQuo) but different values of the key parameters of the
model.
4.2. Other structural reforms
We now simulate the impact of the two other structural reforms described in Section 2, which both
have the direct consequence of reducing aggregate savings in China. The impact on the Chinese
and US economies is depicted in Figures B.1 and G.2 in Appendix B, for the three monetary
regimes.
A financial reform. Chinese firms are allowed to increase their borrowings by 45 percent of GDP,
while loosing a capital subsidy equivalent to 8% percent. The two measures have opposite effects on
investment: on the one hand, it becomes easier to finance fixed capital formation; on the other one,
the expected return is lowered due to the removal of capital subsidies. On the whole, the
investment rate increases by only 0.15 percent of GDP in the short run. The remain- ing borrowings
are channeled back to households through dividends. This one-shot transfer is triggered by a
sudden increase in the value of capital firms. It boosts consumption in the short run (+0.5 to +1% of
GDP, depending on the monetary regime).
24
The interest rate rises to op- pose in?ationary
pressures, which mitigates the rise in consumption and especially investment. The NFA position
declines due to net capital in?ows and consistent with the fall in the trade balance.
Interestingly, the positive impact of the reform on investment is permanent, which allows GDP
and consumption to increase in the long run (contrasting with the pension reform). As detailed in
Table 3, the stock of corporate bonds increases by 37 percent of GDP in the long run. Most
24
In
our model, the propensity to consume is the same for labor as for capital income. Hence, we do not consider
horizontal redistributive effects amongst households.
23
Table 3 - Long-run impact of the financial reform on China's balance sheet
Bonds supply Bonds demand
StatQuo CapMob Flex StatQuo CapMob Flex
Firms 37.1 37.2 37.2 Households 37.1 36.0 36.0
Government 0.2 0.2 0.2
Central bank 0 0 0 Central bank -0.1 0.2 0
Nation 37.3 37.4 37.4 Nation 37 36.2 36
NFA -0.3 -1.2 -1.4
Source: Author calculations.
of it is held by Chinese households, but the NFA position is reduced by 0.3 to 1.4 percent of GDP,
depending on the monetary regime.
Like for the pension reform, there is more rebalancing when capital controls are removed, while
exchange-rate ?exibility allows for less in?ation and a faster rebalancing effect. In fact, the main
advantage of exchange-rate ?exibility is its stabilizing impact on China's consumption and in?ation.
The effects of the shock on the US economy are similar to those obtained with a pension reform
in China: higher exports and GDP in the short run, lower consumption and investment due to a
higher interest rate, hence a rebalancing of the economy which very much depends on the extent
of capital mobility. Again, GDP falls in the long run. Like for the pension reform, the US
economy enjoys higher foreign demand more quickly under a ?exible regime than under a fixed peg
in China. The evolution of the NFA position of the US is similar under the CapMob and the Flex
regimes, except in the short run where exchange-rate ?exibility allows for quicker rebalancing. This is
paid in the US by a sharper short-term drop in consumption under the Flex regime.
In terms of current-account rebalancing, this reform is much less effective than the pension
reform, as can be seen by comparing Tables 3 and 2. In order for the financial reform to have the
same long-run impact on the NFA position than with the 6.5 pp pension reform described above,
capital firms would need to almost triple the debt-to-GDP ratio. However, since this reform has
positive impact on GDP and consumption in the long run, contrasting with the pension reform
that depresses income and consumption in the long run, it could make sense for China to combine the
two.
Government spending reform. We finally study the impact of a government spending re- form.
Specifically, the target debt-to-GDP ratio of China is assumed to be increased from zero to 15
percent of the steady-state GDP. Since in our model the government budget constraint determines
the level of pensions, the latter increase, so this reform is close to an unfunded pen-
24
Table 4 - Long-run impact of the government spending reform on China's balance sheet
Bonds supply Bonds demand
StatQuo CapMob Flex StatQuo CapMob Flex
Firms -0.6 -0.5 -0.5 Households 11.1 10.0 10.0
Government 12 12 12
Central bank -0.1 0 0 Central bank -0.1 0 0
Nation 11.3 11.5 11.5 Nation 11.0 10.0 10.0
NFA -0.3 -1.5 -1.5
Source: Author calculations.
sion reform. Indeed, contrasting with the pension reform studied above, the tax rate on labor
endowment does not increase. In the longer run, the stabilization of the debt ratio is obtained
through a downward adjustment of pensions (see Figure B.2 in Appendix B).
Not surprisingly, the impact of the shock for China is very similar to that of a pension reform:
rise in consumption, fall in investment, surge in in?ation, rebalancing. The main difference is that
the adjustment spreads over a longer period. Hence, differences across regimes are more marked
and long-lived. In the long run (see Table 4), the government debt increases by 12 percent of
GDP
25
, most of this new debt being held by Chinese households. The NFA position declines by 0.3
to 1.5 percent of GDP, depending on the monetary regime (again, there is more rebalancing when
capital ?ows are freed).
The ?exible exchange rate regime in China benefits immediately to US GDP, and the gain
compared to the MobCap regime lasts seven quarters (instead of five in the case of the pension
reform). The interest rate rises with some delay compared with the pension reform case.
On the whole, this reform is also less powerful than a funded reform of the PAYG pension
system to achieve current-account adjustment. This is because the rise in pensions is not per-
manent, hence young household cannot anticipate higher pensions for their old age, although they
will receive interests on their government bonds holdings. For the government spending reform to
have the same impact as the 6.5 pp pension reform on the long-run NFA position, the target debt ratio
would need to be increased by over 56 percent of GDP.
4.3. Structural reforms in China under a ZIRP in the US
In the above simulation, one impact of Chinese reforms is channeled to the US through the
federal Reserve's policy rate, the increase of which lowers aggregate demand. However, the
Federal Reserve may oppose such interest-rate increase if it is constrained by the zero bound:
25
Remember
that the shock is given as a percentage of the steady-state GDP: due to the price increase, the rise in
the debt-to-GDP ratio is only 12% ex post.
25
if the initial policy rate is above its Taylor level due to the zero bound, then an increase in US
in?ation does not immediately lead to monetary tightening. To account for this possibility, we
assume that, for the next eight quarters following the shock, US monetary policy follows a
modified rule where the interest rate smoothing parametero
R
is set to 0.98 instead of 0.8 in the
benchmark model.
The results are reported in Figure 4 for the same, three monetary regimes in China as previ-
ously: a fixed exchange rate with limited capital mobility (StatQuo), a fixed exchange rate with more
capital mobility (CapMob), and a ?exible exchange rate regime with high capital mobility (Flex).
Since the ZIRP only binds in the short and medium run, the long-run impact of the pen- sion reform
is the same as before, but the dynamics differ. Comparing Figure 4 with Figure 3, it appears clearly
that the US policy rate increases much less with the ZIRP than with a Taylor rule during the first eight
quarters of the simulation. Consistently, US consumption increases instead of declining during this
period. Interestingly then, the amount of rebalancing now is much moredependent on China's
monetary regime, as evidenced by the evolution of China's trade balance: under the StatQuo regime, the
pension reform in China produces almost no rebalancing in the short run; and the amount of
rebalancing is now doubled if the renminbi is not allowed to ?oat, compared to the CapMob and Flex
regime.
Turning to China, the main difference between Figures 4 (Taylor rule in the US) and 3 (ZIRP in
the US) occurs under the CapMob regime, i.e. when capital controls are removed but the nom- inal
exchange rate is kept constant. In this case, Chinese monetary policy is dependent on the US one.
Other things equal, the increase in the Chinese interest rate is muted, which produces more in?ation,
a larger fall in the real interest rate, hence a higher increase in consumption and a slower decrease in
investment. Consequently, the increase in real GDP is doubled compared with the Taylor-rule case:
the policy trade-off between employment and in?ation is twisted in favor of the former. Under the
other two regimes (StatQuo and Flex), the results for China are close whether the Fed follows a
Taylor rule or a ZIRP because China's monetary policy is isolated through capital controls (StatQuo)
or a ?exible exchange rate (Flex).
On the whole, our simulation under a ZIRP in the United States tends to mitigate our previous
conclusion that the regime for capital ?ows is more important than the exchange rate regime as far as
global rebalancing is concerned. We find that, under a ZIRP, the exchange rate regime matters
much more than under a standard, Taylor-rule type monetary policy.
5. THE IMPACT OF A MONETARY REFORM IN CHINA
One limitation of the simulations presented in Section 4 is that they analyze the impact of Chi- nese
reforms as deviations from a baseline which is a balanced steady-state. This feature hardly matches
the debate on global imbalances that blames accumulated imbalances and discusses how these
imbalances could unfold. Here we tackle this issue by studying the impact of a monetary reform
in China when the economy is not at the steady state but rather displays large current-account
surpluses.
26
Figure 4 - Impact of a pension reform in China with ZIRP in the United-States
Impact on China
3.5
3
2.5
2
1.5
1
0.5
0
÷0.5
pibchiC
0
÷0.02
÷0.04
÷0.06
÷0.08
÷0.1
÷0.12
÷0.14
÷0.16
÷0.18
÷0.2
pibchiI
1.2
1
0.8
0.6
0.4
0.2
0
÷0.2
perchiinfp
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo
CapMob
Flex
StatQuo
CapMob
Flex
StatQuo
CapMob
Flex
0
÷1
÷2
÷3
÷4
÷5
÷6
Consumption
pibchiNFA
3.5
3
2.5
2
1.5
1
0.5
0
÷0.5
Investment
perchiRER
0.25
0.2
0.15
0.1
0.05
0
In?ation perchiR
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo CapMob Flex StatQuo CapMob Flex StatQuo CapMob Flex
3
2.5
2
1.5
1
0.5
0
÷0.5
Net foreign assets
perchiY
0.4
0.2
0
÷0.2
÷0.4
÷0.6
÷0.8
÷1
÷1.2
Real exchange rate
pibchiBC
per
c
hi
C
Nominal interest rate
pibchibask
2
1.5
1
0.5
0
÷0.5
÷1
÷1.5
5 10 15 20 25 30 35 440 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo CapMob Flex StatQuo CapMob Flex StatQuo CapMob Flex
3.5
3
2.5
2
1.5
1
0.5
0
÷0.5
Real GDP
perusaY
3.5
3
2.5
2
1.5
1
0.5
0
2
1.5
1
0.5
0
÷0.5
Trade balance
Impact on the US
pibusaC
0.05
0.045
0.04
0.035
0.03
0.025
0.02
0.015
0.01
0.005
0
Official reserves
perusaR
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo
CapMob
Flex
StatQuo
CapMob
Flex
StatQuo
CapMob
Flex
GDP
÷0.5
5
10 Consumption
25
15 20 30 35
Nominal interest rate 40
StatQuo CapMob Flex
27
i
n
p
e
r
c
e
n
t
t
o
G
D
P
t
o
G
D
P
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
t
o
G
D
P
i
n
p
e
r
c
e
n
t
t
o
G
D
P
t
o
G
D
P
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
t
o
G
D
P
We proceed in two steps. First, we simulate an "imbalance" shock in the form of (i) a structural
shock whereby China's savings rate is permanently increased, or (ii) a shock on the target level of
China's official reserves. Second, we assume that, after 15 quarters, China decides to allow its
exchange rate to ?oat (and it subsequently keeps the level of official reserves constant). The choice to
undertake the monetary reform after 15 quarters results from the necessity to have the reform take
place once the economy has already accumulated significant imbalances. It does not change the
results qualitatively to select a shorter or a longer period.
5.1. A Pension "reverse" shock
First, we consider a reduction in the generosity of China's pension system, consistent with the
1997 pension reform. Symmetrically to the above pension reform, we assume that the
replacement rate is reduced by 6.5 pp. The results are reported in Figure 5. The grey line
corresponds to the simulation with a fixed exchange rate (StatQuo); the dotted line assumes a
?exible exchange rate from the start of the simulation (Flex); finally, the plain, black line shows the
path of the economy when switching from StatQuo to Flex after 15 quarters.
Not surprisingly, the restrictive pension shock leads to a fall in Chinese consumption in the short
run. The interest rate decreases due to de?ationary pressures. This has two consequences: (i) a rise in
investment, and (ii) a net out?ow of capital. Consistently, the trade balance surges and the NFA
position rises until it stabilizes at a higher level (+2.3% of GDP in the StatQuo regime, with low
capital mobility, and +9.7% of GDP in the Flex regime, with high capital mobility). The real
exchange rate depreciates during several quarters before appreciating towards a long run
equilibrium that lies above its baseline level (+0.2% in the StatQuo regime and +2% in the Flex
regime). Under a fixed peg, the central bank of China initially reduces its official reserves to counter-
balance the (limited) out?ow of capital. After two quarters, it starts accumulating reserves to
ensure the equilibrium of the balance of payment with higher trade surplus and lower private capital
out?ows due to the increase in China's interest rate towards its initial level. Undera ?exible regime,
official reserves stay constant and the nominal exchange rate depreciates in the short run before
appreciating steadily along the adjustment path of the economy. There is less de?ation in this
regime but the increase in the trade balance is doubled compared to the status quo regime, due to
capital mobility.
The switch from a fixed to a ?exible regime freezes official reserves at an in?ated level com-
pared to the steady state.
26
The switch takes place while the nominal, policy interest rate is still
depressed in China relative to the United States. Since the monetary change includes a
relaxation of capital controls, there is a sudden out?ow of private capital, and the renminbi
initially depreciates by 0.5% (although it appreciates in the long run by 2.5% in real terms).
27
26
Alternatively,
we could have assumed that the PBoC progressively reduces its excess reserves when switching to
a ?exible regime. However, this would have amounted to piling up two shocks: a change in the monetary regime, and a
portfolio shock. The results would have been less clear cut.
27
One may wonder why moving to exchange-rate ?exibility while the exchange rate is undervalued leads to an
28
Figure 5 - Flexibilization of China's exchange rate after a "reverse" pension reform
Impact on China
0.4
0.2
0
÷0.2
÷0.4
÷0.6
÷0.8
÷1
÷1.2
÷1.4
perchiC
1.6
1.4
1.2
1
0.8
0.6
0.4
0.2
0
perchiI
0.1
0.05
0
÷0.05
÷0.1
÷0.15
÷0.2
÷0.25
÷0.3
÷0.35
÷0.4
perchiinfp
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
Flex StatQuo Mon. Ref. Flex StatQuo Mon. Ref. Flex StatQuo Mon. Ref.
Consumption pibchiNFA Investment perchiRER In?ation perchiR
5 0.4 0
4.5 0.2
4 0 ÷0.05
3.5 ÷0.2
3 ÷0.4 ÷0.1
2.5 ÷0.6
2 ÷0.8 ÷0.15
1.5 ÷1
1 ÷1.2 ÷0.2
0.5 ÷1.4
0 ÷1.6 ÷0.25
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
Flex StatQuo Mon. Ref. Flex StatQuo Mon. Ref. Flex StatQuo Mon. Ref.
0.6
0.4
0.2
0
÷0.2
÷0.4
÷0.6
÷0.8
Net foreign assets
perchiY
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
÷0.1
÷0.2
Real exchange rate
pibchiBC
per
c
hi
C
nominal interest rate
pibchibask
1.5
1
0.5
0
÷0.5
÷1
÷1.5
÷2
5 10 15 20 25 30 350.440 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
Flex StatQuo Mon. Ref. Flex StatQuo Mon. Ref. Flex StatQuo Mon. Ref.
0.1
0.05
0
÷0.05
÷0.1
÷0.15
÷0.2
÷0.25
real GDP
perusaY
0.2
0
÷0.2
÷0.4
÷0.6
÷0.8
÷1
÷1.2
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
÷0.1
÷0.2
Trade balance
Impact on the US
perusaC
0
÷0.005
÷0.01
÷0.015
÷0.02
÷0.025
÷0.03
÷0.035
÷0.04
official reserves
perusaR
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
Flex
StatQuo
Mon. Ref.
Flex
StatQuo
Mon. Ref.
Flex
StatQuo
Mon. Ref.
GDP
÷1.4
5
10 Consumption
25
15 20 30 35
Nominal interest rate 40
Flex StatQuo Mon. Ref.
29
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
t
o
G
D
P
i
n
p
e
r
c
e
n
t
t
o
G
D
P
t
o
G
D
P
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
Due to lower purchasing power, consumption declines. In?ation is subsequently reduced. On the
whole, the regime shift triggers an increase in the trade surplus, before accompanying the economy
back to its balanced path through a strong appreciation of the real exchange rate.
When the regime shift takes place, US GDP is negatively affected (because the dollar appreci-
ates) but US consumption is boosted (due to the higher purchasing power of US households). This
mirror image of China confirms the failure of the regime shift to rebalance the US econ- omy.
The lack of rebalancing following a switch of China from a pegged regime to a ?exible one
arises from the relaxation of capital controls and from the reluctance of the PBoC to sell its
accumulated stock of reserves. It should be kept in mind that the relaxation of capital controls is a
precondition for the switch to a ?exible exchange-rate regime, since the proper functioning of the
foreign-exchange market necessitates both residents and non-residents to be able to trade the
renminbi for other currencies. What the model says is that switching to a ?exible regime (with free
capital mobility) in a situation of depressed demand would put downward rather than upward
pressure on the renminbi in the short run unless (i) structural reforms are simultaneously (or previously)
carried out to reduce aggregate savings in China and/or (ii) the PBoC decides to sell excess
reserves. Of course, this stylized result may be mitigated by the initial return differential between
China and the United States. It should be reminded here that only bonds yielding the policy rate
are traded between the two countries: in our model, relaxing capital controls does not induce US
investors to buy high-return Chinese stocks, which may put upward pressure on the RMB. Still, the
model highlights the complementarity between monetary and structural reforms to engineer a fast
rebalancing of the economy.
5.2. Reserve accumulation shock
Even though a "reverse" pension shock may have contributed to the high saving rate in China and
the growing imbalances, this shock is not sufficient to explain official reserve accumulation. As argued
by Cova et al. (2009), the rise in emerging Asia's current account surplus after 2001 may also be
related to an increase in the desired net foreign asset position. In the case of China, due to capital
controls, this suggests an increase in the target level for official reserves. Here, we assume that this
target increases progressively by 38 percent of GDP in 20 years. Accordingly, the interest rate rule of
the Chinese central bank is shifted upward in order to be consistent with the constant exchange rate
constant (see Equation (5)). Like in the previous exercise, China is assumed to shift to a ?exible
regime after 15 quarters.
The results are displayed in Figure 6.
28
To the extent that this policy is well anticipated, the
immediate depreciation rather than an appreciation. The reason is the uncovered interest parity with rational
expectations: the exchange rate initially jumps to a path where it can then appreciate at a pace that is consistent with the
interest-rate differential. Before the monetary reform takes place, the exchange-rate is undervalued compared
to its long-run, equilibrium level, not compared to its short-term, market level
28
Here, we do not report the results when China is running a Flex regime during the whole period since, by
30
Figure 6 - Flexibilization of China's exchange rate after a reserve accumulation shock
Impact on China
pibchiC pibchiI perchiinfp
1 0 0.5
0.5
÷0.02
0 0
÷0.5 ÷0.04
÷1 ÷0.5÷0.06
÷1.5
÷0.08÷2 ÷1
÷2.5 ÷0.1
÷3 ÷1.5
÷0.12
÷3.5
÷4 ÷0.14 ÷2
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
MonRef StatQuo MonRef StatQuo MonRef StatQuo
25
20
15
10
5
0
Consumption
pibchiNFA
2
1
0
÷1
÷2
÷3
÷4
÷5
÷6
Investment
perchiRER
0.6
0.4
0.2
0
÷0.2
÷0.4
÷0.6
÷0.8
In?ation perchiR
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
MonRef StatQuo MonRef StatQuo MonRef StatQuo
1
0.5
0
÷0.5
÷1
÷1.5
÷2
÷2.5
÷3
Net foreign assets
perchiY
2.5
2
1.5
1
0.5
0
÷0.5
÷1
Real exchange rate
pibchiBC
per
c
hi
C
Nominal interest rate Test
reserves
30
Target reserves
25
20
15
10
5
0
÷5
5 10 15 20 25 30 35 140 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
MonRef StatQuo MonRef StatQuo
0.6
0.4
0.2
0
÷0.2
÷0.4
÷0.6
÷0.8
Real GDP
perusaY
0
÷1
÷2
÷3
÷4
2.5
2
1.5
1
0.5
0
÷0.5
÷1
Trade balance
Impact on the US
pibusaC
0.1
0.05
0
÷0.05
÷0.1
÷0.15
÷0.2
÷0.25
Official reserves
perusaR
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
MonRef StatQuo MonRef StatQuo MonRef StatQuo
GDP
÷5
5
10 Consumption
25
Mon Ref
15
31
20 30
StatQuo
35
Nominal interest rate 40
i
n
p
e
r
c
e
n
t
t
o
G
D
P
t
o
G
D
P
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
t
o
G
D
P
i
n
p
e
r
c
e
n
t
t
o
G
D
P
t
o
G
D
P
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
t
o
G
D
P
real interest rate is expected to be higher until the level of reserves has reached its new target. As a
consequence, consumption initially drops (intertemporal substitution effect). In?ation decreases
in the short run, the real interest rate increases but the nominal interest rate starts by a large fall to
mitigate the decline in Chinese in?ation. Despite the fixed nominal exchange rate, the real exchange
rate depreciates. Due to nominal rigidities, the depreciation is spread over several quarters.
The Chinese economy progressively accumulates trade surpluses, hence assets on the US econ-
omy in the form of official reserves. Symmetrically, consumption increases in the US thanks to a
lower expected level of nominal interest rate in the medium run, whereas GDP is lowered as
domestic demand shifts to imported goods. After a few years, when a sizable amount of official
reserves has been accumulated, the real interest rate in China starts declining, increasing con-
sumption and reducing the current account surplus. The real exchange rate appreciates through a
recovery of in?ation in China (by 15% from its value in quarter 15 to its long run equilibrium value).
At this stage, a monetary reform that encompasses the relaxation of capital controls, a shift to
a ?exible exchange rate regime and the end of official reserve accumulation (that are frozen at
their in?ated, current value, +11.5% of GDP) induces a large appreciation of the renminbi, by 10%,
and simultaneously removes any in?ationary pressure in China. It also rebalances the world economy
as consumption drops in the US, the nominal interest rate increases and GDP recovers.
On the whole, these two successive simulations suggest that the key factor of rebalancing is less
the ?exibility of the exchange rate by itself than its implications in terms of official reserves, and
that a exchange-rate ?exibility can usefully complement but not be a substitute for structural reforms.
6. A THIRD COUNTRY: THE E U RO A R E A
Finally, we add the Euro area to the model which therefore becomes a three-country model. The
model of the Euro area is the same as for the United States, except for the share of their respec- tive
currencies in Chinese official reserves. Here, we assume that foreign exchange reserves in China
amount to 60% of Chinese GDP, dollar-denominated assets represent 80% of the total and euro-
denominated assets the remainder 20%.
29
We also assume that euro and dollar are not perfect
substitutes: capital mobility between the United States and the Euro area is however high, with¸ = 0.9.
This three-country model is then used to perform three series of simulations: (i) a pension
definition, there is no reserve accumulation under a Flex.
29
At the end of 2010, foreign reserves reached USD 2800 bn whereas Chinese GDP amounted to USD 5100 bn. The
currency composition of Chinese reserves is not released, but the share of the euro is generally estimated around 20%.
32
reform in China (as in Section 4.1), (ii) a diversification of China's official reserves, and (iii) a
diversification of China's official reserves under a ZIRP in both the United States and the Euro area.
6.1. Impact of a pension reform in China in a three-country setting
The impact of an increase in the generosity of China's pension system is similar for the Euro
area as for the United States (see Figure 7). The minor differences come from the higher share
of the dollar in Chinas' reserves.
30
Figure 7 - Impact of a pension reform in China, three-country model
Impact on the US and the euro area
0.3
0.25
0.2
0.15
0.1
0.05
0
÷0.05
Y
0.1
0
÷0.1
÷0.2
÷0.3
÷0.4
÷0.5
÷0.6
C
4
3
2
1
0
÷1
X
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo CapMob Flex StatQuo CapMob Flex StatQuo CapMob Flex
0.04
0.035
GDP
infp
0.035
0.03
Consumption R
0.6
0.5
Exports
pereurNER
0.03
0.025
0.02
0.015
0.01
0.005
0
5
10
3
2.5
15
2
20
25
30
35
40
0.025
per
c
hi
C
0.02
0.015
0.01
0.005
0
5
10
15
20
25
30
35
40
0.4
0.3
0.2
0.1
0
÷0.1
5
10
15
20
25
30
35
40
StatQuo CapMob Flex StatQuo CapMob Flex StatQuo CapMob Flex
0.6
0.5
0.4
0.3
0.2
0.1
0
÷0.1
In?ation BC
1.5
1
0.5
0
Nominal interest rate NFA
5
4
3
2
1
0
0.06
0.04
0.02
0
÷0.02
÷0.04
÷0.06
euro/dollar nominal
pereurRER
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo CapMob Flex StatQuo CapMob Flex StatQuo CapMob Flex
Trade Balance ÷0.5
5 10
15
Net foreign asset
20 25 30 35
40
euro/dollar real
StatQuo CapMob Flex ? Euro area
The Chinese reform is a symmetric shock for the United States and the Euro area. The rebal-
ancing occurs both between China and the United States and between China and the Euro area.
30
In
the steady state, the PBoC holds official reserves equivalent to 60% of GDP. In the simulations performed
under a Flex regime, this amount stays constant but the asymmetric distribution of existing Chinese reserves across the two
currencies produces some asymmetric valuation effects.
33
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
However, as long as the Chinese peg is maintained (StaQuo and CapMob regimes), there is more
rebalancing in the United States than on the Euro area, as the drop in Chinese demand for
reserves falls relatively more on the dollar than on the euro. As a result, the interest rate
increases more in the US, lowering investment and stimulating household saving. The differ-
ences between the US and Euro area are benign such that the EUR/USD exchange rate is not
significantly affected by the Chinese reform. The large sale of dollar compared to euro by the
Chinese central bank triggers a depreciation of the dollar against the euro that induces private
agents in the US and in the euro area, who enjoy free capital mobility, to buy more dollars. This
stabilizing effect contributes to the absence of euro-dollar reaction in the first period. We now turn to
simulating the impact of a change in PBoC's portfolio allocation.
6.2. Impact of a diversification of China's official reserves
We now simulate a diversification of China's foreign exchange reserves out of the dollar into the
euro, with the renminbi still being pegged on the dollar. Specifically, the share of the euro in China's
reserves is assumed to increase to 30% from 20%. As a first step, monetary policy is assumed to
follow a Taylor rule in both the Euro area and the United States. The results are displayed in
Figure 8, where the plain line represents the status quo regime (limited capital mobility) and the
dotted one shows the same simulation when capital controls are relaxed in China.
This diversification induces large capital in?ows into the Euro area, hence a sizable apprecia-
tion of the euro and a fall in Euro area's growth rate. However potential growth increases in the
medium run due to a lower interest rate. The impact is opposite in the US: the dollar deprecia- tion
promotes growth in the short run, but the potential is reduced in the long run due to lower capital
accumulation.
6.3. Impact of a diversification of China's official reserves under a ZIRP
Our last exercise is to study the impact of the same diversification of Chinese reserves under a zero
interest-rate policy in the Euro area and in the United States (see Figure 9). Due to the zero
bound, the ECB is unable to cut its interest rate, whereas monetary tightening in the United States
is also muted. Hence, the fall in European GDP is larger than under a standard Taylor-rule type
monetary policy, whereas the increase in GDP is magnified in the United States. The bilateral
exchange rate of the euro vis-a-vis the dollar appreciates much more than under unconstrained
monetary policy.
7. CONCLUSION
We have constructed a two-country model to study different scenarios of global rebalancing
depending on (i) structural reforms decided in China, (ii) the monetary regime of China, (iii) the
constraints on monetary policy in the United States and (iv) the initial level of cumulated
34
Figure 8 - Impact of a diversification of China's official reserves, three country model
Impact on China
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
perchiC 10x10
8
6
4
2
0
÷2
÷3
pibchiI 0.15
0.1
0.05
0
÷0.05
perchiinfp
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo CapMob StatQuo CapMob StatQuo CapMob
0.1
0.05
0
÷0.05
÷0.1
÷0.15
Consumption
pibchiNFA
0.4
0.3
0.2
0.1
0
÷0.1
Investment
perchiRER
0.04
0.03
0.02
0.01
0
÷0.01
÷0.02
In?ation perchiR
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo CapMob StatQuo CapMob StatQuo CapMob
Net foreign assets Real exchange rate Nominal interest rate
Y Impact on the US and the euro area (?1.5
C ) X
0.15 0.4
0.1 0.3 1
0.05 0.2 0.5
0 0.1 0
÷0.05 0 ÷0.5
÷0.1 ÷0.1 ÷1
÷0.15 ÷0.2 ÷1.5
÷0.2 ÷0.3 ÷2
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo CapMob StatQuo CapMob StatQuo CapMob
0.02
0.01
GDP
infp
0.02
0.015
Consumption R
1.4
1.2
Exports
pereurNER
0
÷0.01
÷0.02
÷0.03
0.35
0.3
0.01
per
c
hi
C
0.005
0
÷0.005
÷0.01
÷0.015
1
0.8
0.6
0.4
0.2
5
10
0.25
15
20 25 30 35 40 ÷0.02 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40
StatQuo CapMob StatQuo CapMob StatQuo CapMob
0.3
0.2
0.1
0
÷0.1
÷0.2
÷0.3
÷0.4
In?ation BC
0.2
0.15
0.1
0.05
Nominal interest rate NFA
3
2
1
0
÷1
÷2
1.2
1
0.8
0.6
0.4
0.2
0
÷0.2
euro/dollar nominal
pereurRER
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo CapMob StatQuo CapMob StatQuo CapMob
Trade
0
Balance
5
10
15
Net foreign asset
20 25 30 35
40
euro/dollar real
StatQuo CapMob ? Euro area
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
t
o
G
D
P
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
t
o
G
D
P
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
35
Figure 9 - Impact of a diversification with a ZIRP in the United states and the Euro Area
Impact on China
perchiC ÷3 pibchiI perchiinfp 0.4 14x10 0.35
0.35 12 0.3
0.3 10 0.25
0.25 8 0.2
0.2 6 0.15
0.15 4 0.1
0.1 2 0.05
0.05 0 0
0 ÷2 ÷0.05
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo ZIRP StatQuo ZIRP StatQuo ZIRP
0.25
0.2
0.15
0.1
0.05
0
÷0.05
÷0.1
÷0.15
Consumption
pibchiNFA
0.4
0.3
0.2
0.1
0
÷0.1
Investment
perchiRER
0.1
0.08
0.06
0.04
0.02
0
÷0.02
In?ation perchiR
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo ZIRP StatQuo ZIRP StatQuo ZIRP
Net foreign assets Real exchange rate Nominal interest rate
1
0.5
Y
Impact on the US and the euro area (?)3C
0.4
2
0.2
1
X
0
÷0.5
÷1
0
÷0.2
÷0.4
0
÷1
÷2
÷3
÷4
÷1.5 ÷0.6 ÷5
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo ZIRP StatQuo ZIRP StatQuo ZIRP
0.3
0.2
GDP
infp
0.02
0.015
Consumption R
3.5
3
Exports
pereurNER
0.1
0
÷0.1
÷0.2
÷0.3
÷0.4
÷0.5
0.4
0.35
0.3
0.01
per
c
hi
C
0.005
0
÷0.005
÷0.01
÷0.015
÷0.02
2.5
2
1.5
1
0.5
0
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo ZIRP StatQuo ZIRP StatQuo ZIRP
In?ation
0.25 BC
Nominal interest rate NFA euro/dollar nominal
pereurRER
0.4 3 2.5
0.2
0
÷0.2
÷0.4
÷0.6
0.2
0.15
0.1
0.05
2
1
0
÷1
÷2
÷3
2
1.5
1
0.5
0
÷0.5
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo ZIRP StatQuo ZIRP StatQuo ZIRP
Trade
0
Balance
5
10
15
Net foreign asset
StatQuo
20 25
36
30
ZIRP
35
40
euro/dollar real
? Euro area
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
t
o
G
D
P
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
t
o
G
D
P
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
imbalances. Our simulations suggest that, if monetary policy follow a standard, Taylor rule in the
United States, then structural reforms in China that reduce aggregate savings are a powerful driver of
global rebalancing, provided there is some relaxation of capital controls. A ?exibiliza- tion of the
renminbi can accelerate the rebalancing but has only a minor effect on its extent. For China,
switching to a ?exible exchange-rate regime during the reform raises a clear trade-off between
in?ation (which would be stabilized with a ?exible regime) and employment (which would suffer).
The possibility that China moves close to the "Lewis turning point" may change the terms of this
trade-off and could favor a ?exible exchange rate.
Now, if the US monetary policy is committed to a zero interest rate, then a move to a ?exible
exchange-rate regime in China has more impact on the amount of the rebalancing. In this
case, the Federal Reserve refrains from hiking its interest rate when global savings are reduced.
Hence the rebalancing can only come from a depreciation of the dollar against the renminbi,
hence from a more ?exible exchange-rate regime in China. The US ZIRP also modifies the policy
trade-off in China since removing capital controls while keeping a fixed peg on the dollar would
amount to adopting the US ZIRP, hence let Chinese in?ation would develop without control.
We then perform the reverse exercise where an initial reduction in the generosity of China's
pension system (as with the 1997 reform) raises the saving rate of households, triggering net
foreign asset accumulation under a fixed peg, before the renminbi is eventually allowed to ?oat.
To the extent that the switch from a peg to a ?oat is accompanied by a relaxation of capital
controls but not by structural reforms reducing the saving rate, and assuming that China's official
reserves are frozen at their observed level when switching to the ?oating regime (not sold out), then the
regime shift involves downward pressure on the renminbi, which delays rather than accelerating
the rebalancing. The reason is the large out?ow of private capital involved by the conjunction of a
low policy rate in China (related to excess savings) and the relaxation of capital controls.
Alternatively, if Chinese current-account surpluses are the result not of a depressed consump- tion
but of a rise in the official reserve target, then the move to a ?exible regime (with frozen reserves)
does produce the desired rebalancing features.
Finally, we introduce the Euro area as a third country in the model. We find here that despite the peg of
the renminbi on the dollar, a structural reform in China produces very similar effects on the Euro
area as on the US economy. Conversely, a diversification of China's official reserves depresses the
economy of the Euro area through an appreciation of the euro against both the dollar and the
renminbi. Interestingly, this detrimental effect is magnified by having a ZIRP in both the United
States and the Euro area.
We conclude that the contribution of China in global rebalancing should primarily rely on
structural policies that reduce aggregate savings in China and on simultaneous opening up of the
Chinese economy. The role of the exchange-rate regime would be minor under standard
37
monetary policies, although more important if monetary policies in advanced countries are con-
strained. Finally, relying only on a change in China's monetary regime could end up in delaying rather
than accelerating the rebalancing, depending on China's policy regarding accumulated reserves.
38
BIBLIOGRAPHY
Baker, M. & Orsmond, D. (2010). Household consumption trends in china. RBA Bulletin, (pp.
13-17).
Bayoumi, T., Tong, H., & Wei, S.-J. (2010). The Chinese Corporate Savings Puzzle: A Firm-
level Cross-country Perspective. NBER Working Papers 16432, National Bureau of Eco-
nomic Research, Inc.
Bernanke, B. S. (2005). The global saving glut and the u.s. current account deficit. Sandridge
Lecture, Virginia Association of Economics, Richmond, Virginia.
Blanchard, O. & Giavazzi, F. (2006). Rebalancing growth in china: A three-handed approach.
China & World Economy, 14(4), 1-20.
Blanchard, O., Giavazzi, F., & Sa, F. (2005). International investors, the u.s. current account,
and the dollar. Brookings Papers on Economic Activity, 36(2005-1), 1-66.
Blanchard, O. & Milesi-Ferretti, G. M. (2010). Global Imbalances: In Midstream? CEPR
Discussion Papers 7693, C.E.P.R. Discussion Papers.
Blanchard, O. J. (1985). Debt, deficits, and finite horizons. Journal of Political Economy, 93(2),
223-47.
Caballero, R. J., Farhi, E., & Gourinchas, P.-O. (2008). An equilibrium model of "global im-
balances" and low interest rates. American Economic Review, 98(1), 358-93.
Calvo, G. A. (1983). Staggered prices in a utility-maximizing framework. Journal of Monetary
Economics, 12(3), 383-398.
Chamon, M., Liu, K., & Prasad, E. (2010). Income uncertainty and household savings in China.
Imf working papers, International Monetary Fund.
Chamon, M. D. & Prasad, E. S. (2010). Why are saving rates of urban households in china
rising? American Economic Journal: Macroeconomics, 2(1), 93-130.
Cova, P., Pisani, M., & Rebucci, A. (2009). Global imbalances: The role of emerging asia.
Review of International Economics, 17(4), 716-733.
Cui, L. & Aziz, J. (2007). Explaining China's Low Consumption: The Neglected Role of House-
hold Income. IMF Working Papers 07/181, International Monetary Fund.
Dooley, M. P., Folkerts-Landau, D., & Garber, P. (2003). An Essay on the Revived Bretton
Woods System. NBER Working Papers 9971, National Bureau of Economic Research, Inc.
Faruqee, H., Laxton, D., Muir, D., & Pesenti, P. A. (2007). Smooth landing or crash? model-
based scenarios of global current account rebalancing. In G7 Current Account Imbalances:
Sustainability and Adjustment, NBER Chapters (pp. 377-456). National Bureau of Economic
Research, Inc.
Feng, J., He, L., & Sato, H. (2009). Public pension and household saving: Evidence from China.
BOFIT Discussion Papers 2/2009, Bank of Finland, Institute for Economies in Transition.
Guo, K. & N'Diaye, P. M. P. (2010). Determinants of China's Private Consumption: An Inter-
39
national Perspective. IMF Working Papers 10/93, International Monetary Fund.
Horioka, C. Y. & Wan, J. (2007). The determinants of household saving in china: A dynamic
panel analysis of provincial data. Journal of Money, Credit and Banking, 39(8), 2077-2096.
Huang, Y. & Wang, B. (2010). Cost distortions and structural imbalances in china. China &
World Economy, 18(s1), 1-17.
Jha, S., Prasad, E., & Terada-Hagiwara, A. (2009). Saving in Asia and Issues for Rebalancing
Growth. Working Papers 162, Asian Development Bank.
Ma, G. & Yi, W. (2010). China's High Saving Rate: Myth and Reality. Working Paper 312,
Bank for International Settlements.
Mendoza, E. G., Quadrini, V., & Rios-Rull, J.-V. (2007). Financial Integration, Financial Deep-
ness and Global Imbalances. NBER Working Papers 12909, National Bureau of Economic
Research, Inc.
Mundell, R. A. (1963). Capital mobility and stabilization policy under fixed and ?exible ex-
change rates. The Canadian Journal of Economics and Political Science / Revue canadienne
d'Economique et de Science politique, 29(4), pp. 475-485.
Obstfeld, M. (1980). Imperfect asset substitutability and monetary policy under fixed exchange
rates. Journal of International Economics, 10(2), 177-200.
Obstfeld, M. & Rogoff, K. S. (2005). Global current account imbalances and exchange rate
adjustments. Brookings Papers on Economic Activity, 36(2005-1), 67-146.
Ouyang, A. Y., Rajan, R. S., & Willett, T. D. (2010). China as a reserve sink: The evidence
from offset and sterilization coefficients. Journal of International Money and Finance, 29(5), 951-
972.
Prasad, E. S. (2009). Rebalancing Growth in Asia. NBER Working Papers 15169, National
Bureau of Economic Research, Inc.
Taylor, J. B. (1993). Discretion versus policy rules in practice. Carnegie-Rochester Conference
Series on Public Policy, 39(1), 195-214.
Yaari, M. E. (1965). Uncertain lifetime, life insurance, and the theory of the consumer. The
Review of Economic Studies, 32(2), 137-150.
40
APPENDIX
A. Details of the model
A.1. Households' program
Households at the cohort level In region j, the inter-temporal budget constraint of a household
born at time a and still living at time t writes (we skip the j superscript for clarity):
Pc
t
C
a,t
+ A
a,t
= W
t
L
a,t
÷t
t
W
t
L
t
÷
a
+ O
a,t
+ H
a,t
, ¯
(9)
where Pc
t
denotes the consumption price index, A
t,a
is the end-of-period asset holding, W
t
L
t,a
÷
t
t
W
t
L
t
÷
a
is the after-tax wage income,
31
O
a,t
represents beginning-of-period financial wealth, and H
a,t
¯
pensions received from the government. Asset holding consist in equities on domestic firms (Z
a,t
(f ),
where f is the firm index), a complete set of contingent claims in domestic currency (Bh
j
a,t
+1
) and ,j
risk-free nominal bonds in foreign currency (Bh
j
a,t ). The value of the portfolio A at the end of period t ,j
writes:
A
ja,t
= E
t
u
t
+1
,j
Bh
j
a,t
+1
+ ,j
t
E
j,j
Bh
j
a,t + Q
jt
(f )Z
j
(f )df ,j
j =j
t f a,t
(10)
Where u
t
+1
,j
is the stochastic discount factor and Q
jt
(f ) is the nominal equity price. t
Table A.1 - Households variables at the cohort level in country j
Name
C
a,t
Description
consumption in volume
Name
H
a,t
Description
pension received from the government
L
a,t
O
a,t
effective labor time
beginning of period financial wealth
L
t
÷a ¯
A
a,t
time endowment
end-of-period financial wealth
Bh
j
a,t
,j
end-of-period foreign bonds holding
Bh
j
a,t+1 ,j
end-of-period holding of contingent
(on date t+1 shocks realization) claims
Given the asset accumulation at period t ÷ 1, and conditionally on the survival of the household at t, the financial
wealth is augmented thanks to the life-insurance contract where bonds are fully eligible
whereas equities are partially eligible (a sharev
E
). Letu
E
=v
E
+ (1 ÷v
E
)u
g
.
O
ja,t
=
1
u
g
Bh
j
a,t + ,j
j =j
E
j,j
R
jt
÷
1
Bf
j
a,t÷
1
+ ,j
t
u
E
u
g
j
f
(Q
t
(f )
+ D
j
(f ))Z
j
÷
1
(f )df
(1 ÷
v
E
)Vk
j
/N
j
¯
t
t a,t
if a < t
if a = t
(11)
Where Vk
j
is the capital firm value and D
j
(f ) is the dividend received by households. Each household
t t
decides his level of consumption, labor supply and asset holdings so as to maximize his intertemporal
utility (1) under the budget constraint (9). The optimality conditions implied by the household program
are the following:
31
In
order to disentangle the redistributive from the allocation effect of labor taxation, we assume that the govern-
ment is able to tax time endowment rather than labor supply.
41
Euler equation
u
t
+1
=
t
|
+1
CC
a,t
t
c
t
a,t+1
Labor supply
L
a,t
= L ¯
a
÷
t
÷
1
÷k MRS
t
k C
a,t
Firm's stock price:
Q
t
= E
t
u
t
+1
(D
t+1
+ Q
t+1
) t
E
jj
Bh
jj
Foreign bonds holding
t a,t
=
¸
jj
t E
jj
t
÷1
A
j
a,t
1÷¸
jj
E
t
u
t
+1
,j
R
j
t
E
jj
t+1
It can be shown that, at the cohort level, consumption is a linear function of total wealth, i.e. financial
wealth (O) and human wealth HW (the future stream of labor an pension income) which is defined by:
·
HW
a,t
= u
t
+
s
u
s
(1÷t
t
+
s
)W
t
+
s
L
t
+s÷
a
+ H
a,t
+s ¯
s=0
t g
where H
a,t
+
s
represents the pension received by generation a at time t + s.
Table A.2 - Notations for regions and agents
symbol description possible values
j, j region u (the United States), c (China), e (euro area)
agent h (households), g (government), b (central bank), f (firms)
Aggregation The aggregate household behavior is described by two main relations: (i) aggregate labor
supply and (ii) aggregate consumption. The former is easily derived through direct aggregation of cohort-
specific labor supplies and is similar to the representative-agent case. The latter is proportional to total wealth,
i.e. human wealth, HW , (discounted sum of future net of tax labor income and pensions of already alive
households), plus financial wealth,O, (the market value of equities and bonds).
k C
t
L
t
= L ÷ 1 ÷k MRS , t
(12) Pc
t
C
t
= O
t
(HW
t
+ O
t
) (13)
To evaluate the nominal stochastic discount factor, u
t
+1
, we introduce a new variable, the relative t
consumption of the young rcy
t
, which is specific to overlapping-generation models. One then has:
u
t
+1
= 1 ÷ (1u÷|u )rcy PcPc
1
C
t
÷
1
,
where
t g g t t÷
t
C
t
(14)
rcy
t
= HW
t,t
+ O
t,t
(15)
HW
t
+ O
t
A.2. Unions and wage setting
The representative union allowed to resets its wage at period t chooses W
t
-
j
so as to maximize the
following objective function:
E
t
·
s=0
(u
g
u
w
)
s
u
t
+
s
(W
t
- t
÷ MRS
t
+
s
Pc
t
+
s
)
42
W
t
-j
W
t
j+s
÷
w
L
jt
+s
(16)
where M RS
t
+
s
denotes the marginal rate of substitution between consumption and leisure in period t+s.
The union receives W
t
-
j
from firms and pays MRS
t
Pc
t
to households. The first order condition of
the program defined by Equation (16) writes
W
t
-
= 1 ÷w
±
t
w W ,
where
w
Z
w
t
t
±
t
w
= E
t
·
s=0
(u
g
u
w
)
s
u
t
+s t
W
t
W
t
+s
÷1÷
w
Pc
t
+
s
MRS
t
+
s
L
t
+s
and
Z
w
= E
t
t
·
s=0
(u
g
u
w
)
s
u
t
+s t
W
t
W
t
+s
÷
w
L
t
+s
.
A.3. Production-firms and price setting
At time t, the production firm i chooses labor and capital inputs so as to maximize profit, given the
demand function Y
t
(i). Letting W
t
and rk
t
denote the nominal wage and the rent cost of capital, respec-
tively, the marginal cost MC
t
writes
1
1
o
1
1÷o
MC
t
= A t o 1÷o rk
o
W
t
1÷
o
. t
(17)
A firm that resets its price at period t will choose P
t
-
that maximizes its market value Vp
t
(i):
Vp
t
(i) = E
t
·
s=0
(u
p
u
E
)
s
u
t
+
s
[P
t
-
÷ MC
t
+
s
] t
P
t
-
P
t
+s
÷
p
Y
t
+
s
,
(18)
where Y
t
+
s
is aggregate demand at time t + s and P
t
+
s
the aggregate production-price level at the same
time.
The representative firm maximizes the objective function given by Equation (18) which leads to the
following first order condition
p
P
t
-
=
±
t
p P ,
where
1 ÷
p
Z
p
t t
±
t
p
= E
t
·
s=0
(u
g
u
p
)
s
u
t
+s t
P
t
P
t
+s
÷1÷
p
MC
t
+
s
Y
t
+s
and
Z
p
= E
t
t
·
s=0
(u
g
u
p
)
s
u
t
+s t
P
t
P
t
+s
43
÷
p
L
t
+s
.
A.4. Capital-firms and capital accumulation
Each capital firm k chooses investment I and external funding Bf in order to maximize its market value,
equal to the expected discounted sum of future dividends, taking as given the rental price of capital rk ,
the subsidies to capital accumulationt
k
, the price of investment Pi and the nominal interest rate on
loans R.
32
The capital-firm so chooses the level of capital K, the level of investment I and the amount
of external financing (Bf ) in order to maximize its market value:
·
(1 +t
k
)rk
t
+
s
K
t
+s÷
1
(k) ÷ Pi
t
+
s
I
t
+
s
(k)
Vk
t
(k) = E
t u
t
+
s
u
s
s=0
t E
+R
t
+s÷
1
Bf
t
+s÷
1
(k) ÷ Bf
t
+
s
(k)
. (19)
under the financing constraint (Equation (3), p.16) and the capital accumulation:
K
t
+
s
(k) = K
t
+s÷
1
(k)u
I
t
+
s
(k) ÷o .
K
t
+s÷
1
(k)
Because dividend ?ows are more discounted than bonds, the borrowing constraint (3) is always binding.
Thanks to this binding condition, the value of the capital firm can then be rewritten in a simpler way that can be
interpreted as the "book value" (see Box).
·
Vk
t
(k) = E
t
s=0
u
t
+
s
(ç + (1÷ç)u
E
)
s
[(1 +t
k
)rk
t
+
s
K
t
+s÷
1
(k) ÷ Pi
t
+
s
I
t
+
s
(k)] +R
t
÷
1
Bf
t
÷
1
(k). t
(20)
This last expression shows the positive effect of financial development on the value of capital firms, hence
on capital accumulation. At the steady-state, the marginal productivity of capital has to cover the
depreciation rateo and the internal financing cost, that equals R/u
E
in absence of external borrowing
(ç = 0). With developed financial markets, the internal financing cost drops to R/(ç + (1÷ç)u
E
) and
the level of capital rises.
Box: The value of the firm with borrowing constraints
Let Vk
1
(k) = Vk
0
(k) ÷ R
t
÷
1
Bf
t
÷
1
(k). Vk
1
(k) verifies the following recursive relation
t t t
Vk
1
(k) = (1 +t
k
)rk
t
K
t
÷
1
(k) ÷ Pi
t
I
t
(k) ÷ (1 ÷u
E
)Bf
t
(k) + E
t
u
t
+1
u
E
Vk
1
÷
1
(k)
t t t
Asu
E
< 1, the borrowing constraint (3) is binding which leads to
ç
÷Bf
t
(k) = 1 ÷
ç
E
t
u
t
+1
Vk
0
+1(k) ,
t t
i.e.
÷ Bf
t
(k) =çE
t
u
t
+1
Vk
1
+1(k) ,
Vk
1
(k) is rewritten as follows: t
t t
Vk
1
(k) = (1 +t
k
)rk
t
K
t
÷
1
(k) ÷ Pi
t
I
t
(k) [(1 ÷u
E
)ç +u
E
] E
t
u
t
+1
Vk
1
+1(k) .
t t t
One can then derive Equation (20)
32
At
date t, dividends are the residual of the ?ow of funds: previous borrowing enters negatively as it has to be
reimbursed whereas fresh borrowing enters positively as it can increases investment or... dividends. Capital firms as also
assumed not to be able to sell their entire capital stock. This assumption does not modify their optimization program.
44
We find two first order conditions
µ
t
=
Pi
t
,
u (
K
Itt÷
1
)
I
t+1
(k)
µ
t
= E
t
u
t
+1
((1 ÷u
E
)ç +u
E
) (1 +t
k
)rk
t+1
+ µ
t+1
ut
where µ
t
is the investment's Lagrange-multiplier.
We can write the cost of capital at the steady-state
K
t
(k)
÷ IK ((kk))
ut+1
t
I
t+1
(k)
K
t
(k)
.
rk = 1 Pit (1 ÷u Rç +u ÷ (1 ÷o) .
+
k
E
) E
A.5. Market equilibria
Good markets equilibrium
Balance of payments
jj
¬j,
Y
t
j
= D
j
+ t
j =j
M
t
j j
(21)
¬j,
j =j
E
t
Bh
j
t
j
+ Bc
j
tj
j holding of j' assets
÷ Bh
j
t
j
+ Bc
j
t
j
=
j' holding of j assets
E
jj
t
Bh
j
tj
1
+ Bc
j
tj
1
R
j
t÷
1
÷ Bh
j
t÷j
1
+ Bc
j
t÷j
1
R
jt
÷
1
+ TB
jt
j
÷ ÷
(22)
Bond markets equilibrium For the record, Bh
i,j
and Bg
i,j
denote the level of bonds denominated t
in the currency of region j and owned by the household sector and the government sector of region i
respectively. Bf
jt
is the (opposite of) the total borrowing of capital-firms in region j. The following
equilibrium holds:
¬j, Bh
j
t
j
+ Bg
j
t
j
+ Bc
j
j
+ Bf
jt
= 0 (23)
j
A.6. Calibration of the model
See Table A.3 page 46.
B. Simulations of alternative structural reforms in China
See Figure B.1 and B.2 pages 47 and 48.
45
Table A.3 - Main parameters
Name Description Value
u
g
Probability to survive 0.966
| Discount factor 0.99
k Leisure preference 0.3
w
p
v
E
o
ç
o
t
µ
R
o
F
R
v
steril
µ
Bg
Bg
target
q
|
u
w
u
p
t
k
¸
jj
o
Elasticity of substitution on the labor market
Elasticity of substitution on the goods market
Share of equities eligible to life-insurance
Share of capital in the production
Share of capital financed by external funds (at the steady-state)
In?ation parameter in the Taylor rule
smoothing parameter in the Taylor rule
Official reserves parameter in the Chinese augmented Taylor rule
Degree of sterilization
Smoothing parameter for government debt level
Government debt target
Share of import in consumption
Elasticity of substitution in trade
Wages rigidity parameter
Prices rigidity parameter
Rate of subsidy on corporate investment
Proxy of capital mobility between country j and j
Depreciation rate of the capital
77
0.9
0.3
0.6
a
2
0.8
0.1
0.1 or 0.9
0.95
0.6
b
0.1
c
2
0.75
0.67
0.08
d
0.5 or 0.9
e
0.02
a
ç
b
0
= 0.3 in China before the financial liberalization shock. in China
before the government spending reform.
c
In the three countries model, imports are shared equally between the two other regions.
d
In China before the financial liberalization shock,t = 0 otherwise. k
e
0.5 for low capital mobility and 0.9 for high capital mobility.
46
Figure B.1 - Financial liberalization
Impact on China
0.45
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
pibchiC
0.25
0.2
0.15
0.1
0.05
0
pibchiI
0.16
0.14
0.12
0.1
0.08
0.06
0.04
0.02
0
÷0.02
perchiinfp
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo
CapMob
Flex
StatQuo
CapMob
Flex
StatQuo
CapMob
Flex
0
÷0.2
÷0.4
÷0.6
÷0.8
÷1
÷1.2
÷1.4
Consumption
pibchiNFA
0.4
0.3
0.2
0.1
0
÷0.1
÷0.2
÷0.3
Investment
perchiRER
0.07
0.06
0.05
0.04
0.03
0.02
0.01
0
In?ation perchiR
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo CapMob Flex StatQuo CapMob Flex StatQuo CapMob Flex
0.5
0.45
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
Net foreign assets
perchiY
0.05
0
÷0.05
÷0.1
÷0.15
÷0.2
Real exchange rate
pibchiBC
per
c
hi
C
Nominal interest rate
pibchibask
0.4
0.3
0.2
0.1
0
÷0.1
÷0.2
÷0.3
÷0.4
÷0.5
5 10 15 20 25 30 0.4540 35 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo CapMob Flex StatQuo CapMob Flex StatQuo CapMob Flex
0.06
0.05
0.04
0.03
0.02
0.01
0
÷0.01
÷0.02
Real GDP
perusaY
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0.1
0.05
0
÷0.05
÷0.1
÷0.15
Trade balance
Impact on the US
pibusaC
0.012
0.01
0.008
0.006
0.004
0.002
0
Official reserves
perusaR
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo
CapMob
Flex
StatQuo
CapMob
Flex
StatQuo
CapMob
Flex
GDP
0
5
10 Consumption
25
15 20 30 35
Nominal interest rate 40
StatQuo CapMob Flex
47
i
n
p
e
r
c
e
n
t
t
o
G
D
P
t
o
G
D
P
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
t
o
G
D
P
i
n
p
e
r
c
e
n
t
t
o
G
D
P
t
o
G
D
P
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
t
o
G
D
P
Figure B.2 - Government spending reform
Impact on China
pibchiC pibchiI perchiinfp
0.3 0 0.09
0.08
0.25 ÷0.005
0.07
0.2 ÷0.01 0.06
0.050.15
÷0.015
0.04
0.1 ÷0.020.03
0.05 ÷0.025 0.02
0.01
0 ÷0.03
0
÷0.05 ÷0.035 ÷0.01
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo
CapMob
Flex
StatQuo
CapMob
Flex
StatQuo
CapMob
Flex
0
÷0.1
÷0.2
÷0.3
÷0.4
÷0.5
÷0.6
÷0.7
÷0.8
÷0.9
Consumption
pibchiNFA
0.3
0.25
0.2
0.15
0.1
0.05
0
÷0.05
Investment
perchiRER
0.04
0.035
0.03
0.025
0.02
0.015
0.01
0.005
0
In?ation perchiR
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo CapMob Flex StatQuo CapMob Flex StatQuo CapMob Flex
0.2
0.15
0.1
0.05
0
÷0.05
Net foreign assets
perchiY
0.02
0
÷0.02
÷0.04
÷0.06
÷0.08
÷0.1
Real exchange rate
pibchiBC
per
c
hi
C
Nominal interest rate
pibchibask
0.2
0.15
0.1
0.05
0
÷0.05
÷0.1
÷0.15
÷0.2
5 10 15 20 25 30 350.340 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo CapMob Flex StatQuo CapMob Flex StatQuo CapMob Flex
Real GDP
0.25
Trade balance
Impact on the US
Official reserves
0.03
0.025
0.02
0.015
0.01
0.005
0
÷0.005
÷0.01
perusaY 0.2
0.15
0.1
0.05
0
0.02
0
÷0.02
÷0.04
÷0.06
÷0.08
÷0.1
pibusaC
6x10
5
4
3
2
1
0
÷1
÷3
perusaR
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo
CapMob
Flex
StatQuo
CapMob
Flex
StatQuo
CapMob
Flex
GDP
÷0.05
5
10 Consumption
25
15 20 30 35
Nominal interest rate 40
StatQuo CapMob Flex
48
i
n
p
e
r
c
e
n
t
t
o
G
D
P
t
o
G
D
P
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
t
o
G
D
P
i
n
p
e
r
c
e
n
t
t
o
G
D
P
t
o
G
D
P
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
t
o
G
D
P
C. Robustness of the model
We now test the robustness of our model by changing some parameters. To do so, we simulate the pen- sion
reform shock with three different sets of parameters, successively, and compare these simulations with the
"StatQuo" simulation of Section 4.
Elasticity of labor supply. First, we changek (the share of leisure in utility, see Equation (1)) from 0.3 to 0.2.
According to the labor supply (Equation (12)) the elasticity of labor supply to wage is given
by:
k C
L
t
,W
t
= 1 ÷k MRSt L tt
Thus, an increase ink involve a decrease in the elasticity of labor supply. As we can see in Figure C.3, the
increase in the real wage involve a less important increase of the labor supply in the case wherek is greater,
but the differences remain limited.
3.5
3
chi
C
Figure C.3 - Pension reform with different elasticities of labor supply
Impact on China
1.2
1
0.8
0.6
0.4
0.2
0
÷0.2
perchiL 2.5
2
1.5
1
0.5
0.25
0.2
0.15
0.1
0.05
0
÷0.05
÷0.1
÷0.15
÷0.2
÷0.25
perchiw
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
÷0.05
perchiinfp
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo (k=0.3) k=0.2 StatQuo (k=0.3) k=0.2 StatQuo (k=0.3) k=0.2
Labor supply
0
5 10 Real wage
15 20 25 30 35 40 In?ation
StatQuo (k = 0.3) k = 0.2
Elasticity of substitution between goods. The calibrated elasticity of substitution between goods
produced in China and goods produced in the US (|) is set to 2. Figure C.4 shows the impact of changing this
parameter to 5.
A rise in this parameter (both in China and USA) involves more competition between goods of the two
countries. Thus, the pension reform shock involves a larger decrease in China's trade balance and a more limited
real appreciation of the renminbi.
49
l
e
v
e
l
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
chi
C
3.5
Figure C.4 - Pension
3
eform with different elasticities of substitution of goods r
Impact on China
0.2
0.1
0
÷0.1
÷0.2
÷0.3
÷0.4
÷0.5
÷0.6
2.5pibchiBC
2
1.5
1
0.5
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
÷0.1
÷0.2
perchiRER
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo (o=2) o=5 StatQuo (o=2) o=5
Trade Balance
10
0
5
15 20 Real exchange rate
25 30 35 40
StatQuo (o = 2) o=5
Share of imports in consumption The calibrated trade openness of the US and China is set to 10%.
Figure C.5 shows the impact of changing this parameter to 20%. The pension reform has now
more limited impact on the real exchange rate and
chi
on the export level, but the results are qualitatively C
unchanged. 3.5
Figure C.5 - Pension
3
reform with different shares of imports in consumption
Impact on China
0.5
0
÷0.5
÷1
÷1.5
÷2
÷2.5
2.5
2
1.5
1
0.5
perchiX
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
÷0.1
÷0.2
perchiRER
5 10 15 20 25 30 35 40 5 10 15 20 25 30 35 40
StatQuo (,C=0.1) ,C=0.2 StatQuo (,C=0.1) ,C=0.2
Exports
0
5
10 15 20
Real exchange rate
25 30 35 40
StatQuo (,
C
= 0.1) ,
C
= 0.2
50
l
e
v
e
l
i
n
p
e
r
c
e
n
t
t
o
G
D
P
l
e
v
e
l
i
n
p
e
r
c
e
n
t
i
n
p
e
r
c
e
n
t
LIST OF WORKING PAPERS RELEASED BY CEPII
51
doc_283837943.docx