Options and Impact of China’s Pension Reform:
A Computable General Equilibrium Analysis1
Yan Wang
The World Bank, Washington, D.C. 20433
Dianqing Xu
University of Western Ontario, London, Ontario, Canada
Zhi Wang
U.S. Department of Agriculture, Washington, D.C. 20036
Fan Zhai
Ministry of Finance, Beijing, 100820, China,
Revised December 20, 2002
Proposed running head: Impact of China’s Pension Reform
Proof should be sent to
Dr. Yan Wang
J4-123,
The World Bank
1818 H Street, NW
Washington, D.C. 20433
United States
[email protected]
Wang, Yan, Xu, Dianquing, Wang, Zhi, and Zhai, Fan— Options and Impact of
China’s Pension Reform: A Computable General Equilibrium Analysis
1
We are grateful to Estelle James and Mark C. Dorfman for advice. We extend thanks to Deepak
Bhattasali, Louise Fox, Robert Holzmann, E.C. Hwa, Shantong Li, Jun Ma, Yvonne Sin, Larry
Thompson, Xiaoqing Yu, Kangbin Zheng, the editor of this journal, and two anonymous referees
for comments and suggestions. The views and results expressed here are entirely our own and
should not be attributed to the institutions with which we are affiliated.
A serious obstacle to China’s economic reform is the lack of a sustainable pension
system. Using a newly designed computable general equilibrium (CGE) model that
differentiates 7 productive activities and 22 age and gender groups, this study compares
various options for financing the implicit pension debt, and estimates the effects of pension
reform on the sustainability of the system and on economic growth. Simulation results
show that the current pay-as-you-go system is not financially sustainable and the implicit
pension debt is estimated at around 46 to 63 percent of GDP in 2000. The paper proposes
to use value added tax revenue to finance the transition cost, which would make the new
multipillar system financially sustainable.
JEL Classification Code: H55; D58; P52
2
1. INTRODUCTION
China’s population has been aging rapidly, and the burdens of supporting the
elderly are distributed unevenly across regions and sectors. A serious obstacle to China’s
economic reform is the lack of an effective and sustainable national pension system. Two
major problems with China’s current pension system—the short-run problem of the
pension burdens of state-owned enterprises and the longer-term problem arising from the
rapid aging of the population—have deepened over the past few years. Many state
enterprises have not been able to afford to pay payroll taxes and thus pension funds in
many municipalities are in deficit. These deficits could threaten the fiscal stability of the
central government as well as local governments. 1 Reforming the current pension system
is now a matter of ur gency.
Building on previous studies, this article addresses the most urgent issue in
China’s pension reform, namely, how to finance the unfunded pension liabilities. Put
another way, this study investigates ways to recapitalize the pension system, which is
financially nonviable. 2 Using a newly designed computable general equilibrium (CGE)
model that differentiates 7 productive activities and 22 age and gender groups, this study
compares various options for financing the implicit pension debt, estimating the effects of
China’s pension reform on the sustainability of the system and on economic growth.
The issues addressed have significant implications for China’s fiscal stabilization
and for the alleviation of poverty and inequality. First, unfunded pensio n liabilities
represent a significant part of the direct and implicit (partly current and partly future)
liabilities of local and central governments. 3 If not monitored and checked carefully,
these liabilities could threaten the central government’s fiscal sustainability. Second,
pension reform is closely linked to restructuring of the state sector and financial sectors.
It is desirable to find some synergy between the transition problems of the pension
system and the state sector. Third, all reform options involving taxation and other forms
of financing have benefits and costs. This study seeks to inform the decisionmaking
process by comparing various reform options.
2. PENSION REFORM IN CHINA
Because of rapid increases in life expectancy and declining fertility rates,
populations are aging more rapidly in developing countries than they did in industrial
countries (World Bank, 1994a). In China the proportion of population older than 60 will
rise from 9% in 1990 to 22% in 2030 (World Bank, 1997). Based on new base year data
and our estimates, China’s old age dependency ratio (the ratio of the population 65 and
older to the population 15–64) will rise from 11% in 2000 to 25% in 2030 and 39% in
2050 (Figure 1).
FIGURE 1 ABOUT HERE
Population aging has put severe pressure on pension systems around the world.
Many countries have undertaken major or minor reforms of their pension systems. It is
now widely recognized that pay-as-you- go (PAYG) systems generate many problems.
These include rising payroll tax rates; evasion; early retirement due to incentive
problems; misallocation of public resources; lost opportunity to generate long-term
3
savings; unintended intergeneration transfers (often to higher income groups, which live
longer); and the growth of a large implicit pension debt and financing gap in the face of
an aging population. 4 The World Bank (1994a) has been recommending, and many
countries (including China) have been moving toward, multi-pillar systems that contain a
mandatory, tax- financed, publicly managed pillar for redistributive and coinsurance
objectives; a mandatory, fully funded, privately managed defined contribution pillar
(individual accounts) for savings; and a voluntary, fully funded pillar funded through
personal savings or commercial insurance.
China used to have an urban- and enterprise-based pension system covering only
the state sector and some large collective enterprises. No national system existed, as there
was no formal pooling arrangement across enterprises. Pension reform started in 1986,
when State Council Document 77 encouraged pension pooling across state enterprises on
a limited basis at the municipal level; individual contributions were implemented for
contractual workers only. State Council Document 33 of 1991 called for the
establishment of three tiers in the pension system: a basic benefit, a supplementary
benefit to be provided by enterprises in sound financial condition, and a benefit based on
individual savings.
In 1995 State Council Document 6 proposed two plans for the basic pension tier.
Municipal and prefecture governments were given the right to select a reform design and
provincial governments the right to approve the choice. This led to a highly fragmented
system, in which provincial and local governments and line ministries selected various
combinations of the two plans. A World Bank study (1997) pointed out many problems
and recommended a multi-pillar system. The study was optimistic, however, in its
assumptions about expanding coverage. Moreover, it could not investigate the effects of
using different taxes to finance the transition cost, because of limitations of the actuarial
model.
State Council Document 26 of July 1997 defined more clearly the direction of
pension reform. Under this document, which is largely consistent with the World Bank’s
recommendations, a multi-tier pension system combining social pooling with individual
account was to be instituted by 2000. Funds were to be pooled at the provincial level,
with contributions from enterprises and individual workers. The basic pension system
consists of two components, a PAYG component and a mandatory individual account,
both publicly managed by designated government entities (see section 6 for details on
contributions and benefits). There is also a voluntary and supplementary pension scheme,
which is not the focus of this article. This document provides the starting point of our
policy simulation.
State Council Document 42 of 2000 defined the guidelines for further pension
reform. A pilot program has been implemented in Liaoning Province based on the
guidelines set forth in this document. The 20% contribution by enterprises will go
entirely to the social pooling fund; individual accounts will be financed by the 8%
contributed by individuals. This document requires the segregation of the management of
individual accounts from the administration of the social pooling funds.
4
China still has a fragmented, municipality-based PAYG system (plus notional
individual accounts) that is publicly managed. This hybrid system is in the process of
transition. The current system has the following problems.
The system is still fragmented. By 2000 only 5 of the 31 provincial- level entities
(Shaanxi Province and Beijing, Chongqing, Shanghai, and Tianjin) had achieved full
pooling at the province level. In other provinces a small adjustment fund has been
established at the provincial level, but contribution rates continue to vary by municipality
within provinces. The fragmentation has led to inequality across regions and enterprises
and the lack of portability of pensions.
Social pooling is incomplete. In most localities, contributions and pension
expenditure are not administered separately by different agencies, and enterprises submit
a net amount to the municipal pool. Contributio n rates vary significantly across provinces
and cities and in some cases even across different enterprises.
Coverage is expanding, but the benefit level remains too high. By the end of 2000
the basic pension system covered almost 100% of workers in state-owned enterprises and
urban collective enterprises, but coverage in foreign- funded enterprises and private
enterprises was still low. Some industrial sectors with relatively young work forces have
resisted the requirement of pension participation or unification. Benefit levels remain
high, with replacement rates at 70%–90% (Table 1).
TABLE 1 ABOUT HERE
Individual accounts are largely notional. Reserves accumulated in the individual
accounts have been borrowed to pay current retirees, as the government has not made a
clear decision on how to finance the transition cost. Neither Document 26 of 1997 nor
Document 42 of 2000 addressed the issue of how to finance the transition cost.
The funding problem is acute. Many local pension pools are bankrupt and require
government subsidies. The total deficit of pension funds increased from a little more than
RMB10 billion in 1998 to almost RMB40 billion in 2000 (MOLSS, 2001). In 1999 the
Ministry of Finance transferred RMB17 billion to 25 provinces to cover pension
shortfalls at the local level. In 2000 budgetary subsidies for pension funds exceeded
RMB20 billion, according to Ma and Zhai 2001. Other sources report budget subsidies of
RMB34 billion, or 17% of pension spending (James 2001). The central and local
governments are in effect bailing out local pension pools. If unchecked this trend will
threaten the fiscal sustainability of the central government.
3. IMPLICIT PENSION DEBT AND TRANSITION ISSUES
Implicit pension debt refers to the benefit promises a pension scheme makes to
workers and pensioners. It is measured by adding the present value of benefits that have
to be paid to current pensioners and the present value of pension rights that current
workers have already earned and would have to be paid if the system were terminated
today. Implicit pension debt is usually calculated under the hypothesis that the unfunded
system is to be terminated immediately and that all pensioners and workers must be
compensated for their future pensions and accrued rights. Implicit pension debt, a stock
5
concept, represents the direct and implicit liabilities that a government can predict with
certainty and must include in its fiscal plans.
Transition cost, a flow concept, arises from the financing gap created when
expenditures to pensioners and future retirees must continue even though part of the
contributions has been diverted to funded individual accounts. Transition cost stems from
the need to pay off, over some years, the debt of the old system. This financing gap stems
from the implicit pension debt but it is not equal to the implicit pension debt, since some
of the expenditures are for new obligations that accrue each day and some of the current
obligations are covered by ongoing contributions.
The size of the implicit pens ion debt depends on many economic and
demographic factors, such as the age structure of covered workers and pensioners, the
coverage of the pension system, the level of pension benefits, the retirement age,
replacement rates, the indexation mechanism, and discount rates. The World Bank (1997)
estimates China’s implicit pension debt at 46%–69% of 1994 GDP, based on the
hypothesis that the system would be terminated in 1994. A recent estimate puts the
implicit pension debt at 94% of 1998 GDP (Dorfman and Sin, 2000).
Reluctant to recognize the pension debt explicitly, the Chinese government tried
to use a combination of two methods to finance the transition. It has kept a small PAYG
pillar, with a 17% contribution and a 20% replacement rate (State Council Document 26).
About seven percentage points of the contribution was designed to finance the transition.
At the same time, the government maintained pension benefits, which remain high,
hoping that expanding pension coverage and borrowing from the individual accounts of
younger workers would provide extra funding. It is now evident that these methods have
not provided sufficient funding to finance the transition cost.
International experience suggests several options for financing the transition.
First, the implicit pension debt could be reduced by downsizing the existing system and
reducing benefits. Second, asset-debt swaps could be used during the corporatization or
privatization of state-owned enterprises. Third, general tax revenue could be used, or the
government could issue debt. A recent World Bank study (Dorfman and Sin 2000)
suggested increasing contribution rates or using general revenues or dedicated Social
Security taxes, debt issued by the central government or local governments, or the
proceeds of selling state-owned enterprise or other assets to finance the transition. 5
We focus on the option of using tax revenues to finance the transition cost. But it
is also feasible to use proceeds from corporatization or privatization. Chile used proceeds
from privatization to finance its transition cost from pension reform. Bolivia uses
proceeds from the privatization of its six largest state enterprises to establish a flat
minimum pension for everyone. Although the benefit is low, the program reaches the
poorest and most vulnerable group in the countries—the elderly poor, who cannot save
for retirement. This topic is beyond the scope of this article. 6
4. THE ADVANTAGE OF COMPUTABLE GENERAL EQUILIBRIUM MODELS
There has been a great deal of academic interest in analyzing and evaluating
pension reform policies quantitatively. Feldstein (1974) pioneered the use of
mathematical tools for describing aggregate capital accumulation in a social security
6
system. Sargent and Wallace (1995) used a rational expectations model to examine the
impact of pension fund system. Seidman (1986), Arrau (1990), Kotlikoff (1997, 1998),
and many others have advanced the study of pension systems in recent years. The World
Bank has developed a user- friendly computer program to calculate the effects of pension
fund collection based on an actuarial model (the Pension Reform Options Simulation
Toolkit, or PROST). However, it is a partial equilibrium model, in which prices and
wages are assumed fixed.
Using a CGE model to analyze the impact of pension reform has several
advantages. First, it has theoretical consistency. The CGE model can be considered as
incorporating particular specifications of production and demand functions in the wellknown Arrow-Debreu general equilibrium framework. Prices and quantities are
determined endogenously in simulating the results of an external shock or a policy
change. An economywide consistency check can be performed by means of Walras’ Law.
Second, CGE models impose accounting consistency. A CGE model usually
builds on a closed accounting system, such as a social accounting matrix, which details
all the basic identities for the modeled economy. Expenditure and income have to match in
that firms, households, and governments cannot spend more than they earn and every agent
must balance its budget. This implies that the cost of any subsidy or transfer to a pension
system has to be financed, and revenue from any tax has to be allocated. Therefore, the
consequences of any changes in pension revenue or expendit ure can all be quantitatively
measured.
Third, CGE models can be used to analyze changes in economic structure, which
are very important to changes in the welfare of various groups of households. Since
pension policy change has welfare consequences, properly measuring changes in welfare
is important for policy evaluation.
Finally, CGE models are able to analyze large, discrete policy changes that differ
greatly from the baseline. Econometric models make questionable inferences when
shocks are outside the range of historic variation. CGE models are calibrated to actual
input-output data for all sectors in the economy, ensuring that the relative size and
importance of markets are taken into account when tracing the impacts of pension reform
through the national economy. This helps policymakers identify those parts of the
economy where important adjustments may take place when the new pension reform
measures are implemented.
5. STRUCTURE OF THE MODEL
Building on a long tradition of multi-sector CGE models used in analyzing trade
and public policies (Beghin and others, 1994; Garbaccio, 1994; Wang and Zhai, 1998),
the model is recursive dynamic. 7 It divides labor inputs into 22 age and gender groups
and has a built- in module on population dynamics and labor supply. The model includes
seven production sectors and two representative households. It is specified and solved in
levels. 8
5.1 Firms’ Ownership Structure and Production
The model distinguishes between goods and activities. There are three composite
(Armington) goods—agricultural goods, nonagricultural goods, and public service—but
and seven activities (firms). For agriculture and public service, there is one representative
7
firm in each sector, which produces only one product. The firms in the nonagricultural
sector are disaggregated by five ownership types: namely, township and village
enterprises (TVEs), rural informal sector, state and collective enterprises, other urban
formal sector (private enterprises, joint stock companies, and foreign joint venture
companies) and urban informal sector. 9 The model assumes that the goods produced by
different firms in the nonagricultural sector are imperfect substitutes. A CES aggregation
function with relative high elasticities of substitution is used to combine the products into
a single commodity.
Production uses primary factors and other products (both domestic and imported) as
variable inputs in a cost- minimizing way. It is characterized by a multi- level nesting of
constant elasticity of substitution functions. At the first level, firms are assumed to use a
composite of primary factors (value-added and an aggregate intermediate input according
to a constant elasticity of substitution cost function). Technology in all sectors is assumed
to exhibit constant retur ns to scale. At the second level, the division of other intermediate
demand is assumed to follow a Leontief specification; therefore, there is no substitution
among other intermediate inputs. At the same level, the value-added bundle is divided
between land-capital and aggregate labor bundles, which are further split into three age
groups: young, middle-age, and old workers. 10 Agricultural land and physical capital
inputs are also split at the third level. In the fourth and fifth nests, each labor group is
divided into detailed age and gender groups. To maximize profits, the firm allocates its
production output between domestic sales and exports to the world market.
Agricultural land as a sector-specific factor is used only in agricultural production;
physical capital is mobile across sectors. The labor force is distinguished by its location
(urban or rural). It is region specific, with rural sectors using only rural labor and urban
sectors using only urban labor. Within urban and rural areas, the labor force is fully
mobile across sectors. The transfer of labor from rural to urban is implemented through
exogenous population migration.
5.2 Domestic and Import Demand
Agents are assumed to consider products from domestic supply and imports as
imperfect substitutes (the Armington assumption). The two representative households
(one in an urban area, the other in a rural area) are assumed to maximize a Stone-Geary
utility function over the three composite (Armington) goods, subject to their budget
constraints, which leads to the extended linear expenditure system. Household savings
are treated as a demand for future consumption goods with zero subsistence quantity
(Howe, 1975). A household-specific, aggregate consumer price index is specified as the
price of savings. It represents the opportunity cost of giving up current consumption in
exchange for future consumption (Wang and Kinsey, 1994). Other final demands,
including public spending and investment demand, are based on constant share functions
that fix their structure in real terms. The intermediate inputs for the firms, household
consumption, and other final demands constitute the total demand for the same
Armington composite of domestic products and imported goods from the rest of the
world.
8
5.3 Income and Government Policy Instruments
Production generates income, which is distributed to four major institutions:
enterprises (corporations), households, the government, and extra-budget public sectors.
Corporate earnings equal a share of gross operating surplus (i.e., the sum of capital
remuneration across all sectors minus corporate income taxes). A part of net company
income is allocated to households as distributed profits based on fixed shares, which are
the assumed shares of capital ownership by households. Another part of net company
income is allocated to extra-budget public sectors as a fee. Retained earnings (i.e.,
corporate savings for new investment and capital depreciation replacement) equal aftertax company income minus the distributed profits and fee.
Household income consists of labor earnings and the returns from land and capital
the household owns. Additionally, households pay taxes and receive pension benefits,
transfers from the government, and remittances from the rest of the world. Household
disposable income equals the sum of household income from different sources less
various taxes.
The government derives revenues from direct corporate and household income
taxes, import tariffs, and various types of indirect taxes. Subsidies and export tax rebates
enter as negative receipts. Two types of indirect taxes are included in the model. The
value-added tax, the most important indirect tax in China after the 1994 tax reform, is
treated as a tax levied on production factors; its revenues equal total sector value-added
multiplied by a tax rate. The other indirect tax is treated as a production tax levied on
sectoral outputs.
All tax rates are taken as parameters in the model. However, they can be
endogenized to meet government fiscal targets, in which case an adjustment parameter
associated with each type of taxes become endogenous. This adjustment parameter shifts
in or out to achieve government budget balance. Otherwise, the tax schedules are
exogenous and the adjustment factors remain at their initial value of one. An adjustment
factor on government transfers, similar to the adjustment factors on other taxes, provides
another fiscal instrument to achieve a specified budget target.
The extra-budget public sector collects fees from enterprises and households. This
income is allocated to consumption and saving. Spending by extra-budget public sectors
and the government constitutes public consumption, one type of final demand.
5.4 Intra-Period Equilibrium and Macroeconomic Closure
Equilibrium is defined as a set of prices and quantities at which demand equals
supply for all goods and factors, each industry earns zero profit, and gross investment
equals aggregate savings, which is the sum of domestic savings plus foreign capital
inflows.
There are three ma jor macroeconomic balances in the model: the government
budget, aggregate savings and investment, and the balance of payments. In the
benchmark equilibrium all three accounting balances hold. The behavioral aspect of
macroeconomic closure in a CGE model involves choosing a mechanism through which
macroeconomic balances are brought back into equilibrium when exogenous shocks
disrupt the benchmark during an experiment. A macroeconomic scenario is thus imposed
9
on the CGE model, and the sectoral implications of the assumed macroeconomic
behavior can be traced out (Devarajan, Lewis, and Robinson, 1990).
Given China’s small share of world trade, a small-country assumption is used for
imports (i.e., we assume that the local consumption of imports does not affect the border
price of imports). Exports are demanded by the rest of the world according to constantelasticity demand curves, the price elasticities of which are high but finite (Pomfret,
1997). An exchange rate is specified to convert world prices into domestic prices. Either
this exchange rate or the balance of payment can be fixed while the other is allowed to
adjust, providing alternative closure rules.
Since the purpose of this article is to estimate the implicit pension debt and
outline government options for financing the transition cost, we keep the domestic
savings and investment gap constant in all the simulations. This is achieved by keeping
the balance of payments and real government spending, except pension expenditure,
fixed as a share of real GDP, thereby imposing macroeconomic rigidity on the model. If
government pension expenditure changes from the baseline because of an increase in
government pension payment, an increase in government tax earnings must be
implemented or a government deficit will occur. By a macroeconomic identity, this
closure implies that a constant sum of domestic savings and taxes in real terms is needed
to finance both real investment and real government expenditures. Thus any changes in
real GDP are induced by changes in real absorption (i.e., household consumption plus
other final demand), making it easy to compare the efficiency impacts of different
simulations.
5.5 Inter-Period Linkage and Recursive Dynamics
The inter-period equilibrium is solved recursively from 2000 to 2050. Between
each two periods, dynamic growth in the model originates from three sources: the rate
and demographic structure of labor force growth, the accumulation of physical capital
stocks, and improvement in total factor productivity (TFP). Capital stock in each period
equals the previous period’s capital stock plus gross investment minus depreciation. All
net investments in the previous period are assumed to become new production capital in
the next period. Accumulation patterns for capital stock depend on the depreciation rate
and gross investment rate, which is set exogenously as a percentage of GDP. However,
household and corporate savings, the government surplus (or deficit), and foreign capital
inflow (foreign savings) are assumed to be perfect substitutes and collectively constitute
the sources of gross investment.
5.6 Population Projection and Labor Force Dynamics
The sustainability of a pension system is closely related to the demographic
structure of the labor force, which depends largely on the dynamics of population growth.
Correctly modeling the structure of population and labor force growth is a necessary
condition for the successful design and analysis of pension reform.
Our population projection model is based on the deterministic difference equation
model developed by Leslie (1945), which has been used widely in mathematical
demography and population dynamics studies. In such a model China’s population is
taken as a whole, and all factors that influence population growth, such as fertility and
10
mortality rates, are defined in terms of statistical means. We assume that changes in time,
birth, and death are the three major determinants of population growth. The influence of
other social and economic factors is included implicitly in these three determinants.
Economic growth is assumed to affect the fertility and mortality rate through an
exogenously specified total fertility rate and through changes in life expectancy.
Migration from rural to urban areas is also exogenous to reflect the fact of China’s rapid
urbanization.
The newborn population each year equals the number of women between 15 and
49 multiplied by the corresponding fertility rate. The population in each age group equals
the previous year’s population at the next youngest age group multiplied by one minus
the corresponding mortality rate. Nobody lives longer than 100 years. Fertility and
mortality rates, the proportion of each gender in the newborn population, and the base
year population demographic distribution are obtained from China’s 1990 population
survey.
Labor supply at each age (15–70) for a given year equals the population in that
age group for that year multiplied by a labor participation rate, which is gender and age
specific. These age-specific labor supplies are then aggregated into the 22 groups of labor
inputs specified in the cost and labor demand functions. We assume that workers start to
retire at age 40. The number of new retirees in a given year is the sum of all surviving
workers 40 and older multiplied by a retirement rate that is gender and age specific. The
number of retirees in a year is the sum of the surviving retirees from all previous years
plus the number of new retirees. Retirees are differentiated by sector, enabling the model
to reflect the fact that pension coverage for workers differs across firm ownership type in
China.
5.7 Pension System and Transitional Cost
The current pension system in our baseline model consists of two different
components, which are provided only to the urban formal sector. Informal and rural
sectors are not covered by any pension system. Based on State Council Document 26
(1997), the basic pension pillar consists of two components, one of which is a PAYG
pillar, or the basic social pooling fund. This component receives contributions from
enterprises (17% of a worker’s wages) and provides a replacement rate of 20% of final
wages. The second component is the mandatory notional individual account, financed by
an 11% contribution from employees and employers. At retirement this component
provides a monthly benefit equivalent to 1/120th of the account’s notional accumulation.
This component is a defined contribution in name, but it is actually a defined benefit
(PAYG), since the fund of individual accounts is managed together with the basic social
pooling fund. Therefore, in our baseline, the first and second components are treated as
one defined-benefit PAYG system. Pension contribution in the basic pension pillar is
implemented through a payroll tax. Payroll tax for emplo yees is considered a wage tax
and is levied on employees’ labor income; payroll tax for employers is levied on the
capital income of enterprises. All pension benefits in the pension schemes are indexed by
the average of an economywide consumer price index and average urban wage index.
Pension fund surpluses or deficits are the differences between pension revenue
and pension expenditure. The model traces the accumulation of pension fund surpluses
11
and deficits, which constitutes pension fund reserves and debts. The interest income and
payments of a pension fund reserve or debts are a part of revenues or expenditure for the
pension fund. An initial interest rate of 7% is assumed for the base year. The change in
the interest rate is determined by the change in the economywide return of capital.
Therefore, financial flows over the next 50 years can be traced, and the financial
resources needed over time to pay current pension and accrued pension rights of current
workers can be estimated by the model.
The equilibrium data set of the model at base year is constructed around a Chinese
social accounting matrix for pension reform estimated for 2000. The matrix includes
seven productive activities and two representative households. 11 It provides a consistent
framework for organizing the relevant flow-of-value statistics for China’s economy to
satisfy the requirements of a benchmark data set for CGE modeling, as outlined in
Whalley (1985).
6. BASELINE CALIBRATION AND SIMULATION DESIGN
In this section we describe our baseline calibration and the rationales for the
design of various simulations.
6.1 Major Assumptions in Baseline Calibration
To calibrate a baseline of the model, we introduce economywide Hicks- neutral
TFP variables into the model. These are solved endogenously in calibration to match a
prespecified path of real GDP growth. The growth rate of real GDP is set to decline
linearly, from 7.5% in 2000 to 3% in 2050.
Since life expectancy will increase with rising living standards as the economy
grows, gradually declining mortality rates are calculated each year in the baseline
according to exogenously specified life expectancy estimates from the China Population
Information and Research Center. The total fertility rate for women during childbearing
years is assumed to increase slightly between 2000 and 2050 because of China’s
relaxation of its one child policy in the coming years. Rural–urban migration rates are
from the same source.
The individual accounts in China’s current pension system are largely notional,
and funds are managed together with pillar 1. Thus individual accounts are essentially a
part of the PAYG system. We therefore treat pillars 1 and pillar 2 as a large PAYG pillar
in our baseline calibration and assume 60% replacement and 24% contribution rates.
Detailed assumptions on macroeconomic closure, demographic parameters, and pension
system characteristics used to calibrate the baseline are summarized in Table 2.
Assumptions on demographic parameters are shown in Table 3.
TABLES 2 AND 3 ABOUT HERE
6.2 Simulation Design
We conducted three sets of policy simulations and nine experiments, the results of
which are summarized in Table 4. Simulation Set 1 describes some limited changes in
the current PAYG system. The first set of policy simulations examines the effects of
expanding program coverage and increasing the retirement age.
12
TABLE 4 ABOUT HERE
Is a new multi-pillar system with fully funded individual accounts going to help
solve the above problem? Two closely related questions arise when considering this
option. First, what is the transition cost to this multi-pillar system, and can it be financed?
Second, is the proposed new system financially superior to the current system?
Simulation Sets 2 and 3 are designed to provide insights into these two questions.
Simulation Set 2 focuses on the transition to various multi-pillar systems the
government is implementing or experimenting with. Experiment 2.1 describes a transition
to the multi-pillar system defined in State Council Document 26 (1997). Experiment 2.2
describes the transition to Document 26 and the expansion of coverage to public
institutions. Experiment 2.3 describes the transition to a system as defined in State
Council Document 42 (2000), or the Liaoning experiment, which raises the enterprise
contribution to pillar 1 and privatizes the individual account. Experiment 2.4 describes
the transition to a new multi-pillar system proposed here, in which the contribution rate
for pillar 1 is lower and the government finances the total transition cost with valueadded tax (VAT) revenue. In each case, we estimate the transition cost and outline
options for financing it.
Simulation Set 3 examines the financial sustainability of the new multi-pillar
system based on Experiment 2.4, after the transition issue has been solved. This set of
simulations assumes that the transition cost is financed using VAT revenue and that the
new pillar 2 becomes a fully funded individual account. Our aim is to investigate whether,
under different assumptions about coverage and retirement ages, the public pillar and the
individual account pillar can both become financially sustainable after injecting fiscal
resources to finance the transition cost.
The government is considering a new social security tax covering not only old age
pensions but also other benefits. It would offer wide coverage and be unified across the
country. The new multi-pillar system, defined in Experiment 2.4 and examined in
Simulation Set 3, could be viewed as a basis for the design of the proposed social security
tax (the old age pension part only).
7. SIMULATION RESULTS
Results from the baseline calibration are presented in Table 5, which provides a
general picture of China’s economic and demographic trends over the next 50 years.
Under the assumptions specified in Tables 2 and 3, this calibrated benchmark serves as a
basis of comparison for the counterfactual simulation analysis presented later. All the
endogenously solved macroeconomic and demographic variables seem to fall into a
reasonable range. TFP—as a residual and an adjusting mechanism in the model to match
the prespecified real GDP growth rate and the gross investment rate—declines gradually,
from 2.35% a year in 2010 to 1.01% in 2050. These figures are consistent with recent
estimations of TFP growth in China.
TABLE 5 ABOUT HERE
13
The results from baseline calibration also show that with the population growing
at a decreasing rate, the Chinese labor force stops growing between 2015 and 2020 and
declines thereafter. The age group 65 and older continues to grow, however. Thus the old
age dependency ratio (the ratio of people older than 65 to people 15–64) rises steadily,
from 11% in 2005 to 25% in 2030 and 39% in 2050. The system dependency ratio—
which is 35% today, implying 2.8 workers supporting one retiree—rises to 69% in 2050,
or 1.4 workers supporting one retiree. The annual deficit of the PAYG system, which was
RMB40 billion in 2000, grows to more than RMB1.4 trillion in 2050. The accumulated
reserve becomes negative during the next few years, reaching–RMB41.2 trillion by 2050
(Table 5).
If we increase payroll tax rates to maintain the financial viability of the pension
system, the balanced contribution rate would have to rise from 28% in 2000 to 45% in
2030 and 59% in 2050. There is no doub t that the high contribution rates for the pension
system would seriously depress investment and reduce economic growth. The results of
the baseline calibration confirm that the current PAYG system is not financially
sustainable and that the high annual deficit threatens China’s fiscal stability.
In Table 6 we estimate a range of implicit pension debt, based on the termination
hypothesis and assumptions about the baseline. Assuming the discount rate equals the
interest rate, the implicit pension debt would be as high as RMB4.3 trillion (in 2000
yuan), or 47.8% of GDP, in 2000. If the government and public institutions are included,
the implicit pension debt would be RMB5.7 trillion, or 63.1% of GDP. If the discount
rate were higher, the implicit pension debt would be lower, and vice versa.
TABLE 6 ABOUT HERE.
7.1 Limited Reform under Current PAYG System
If the current pension system in China is not sustainable, the question arises:
Could we solve the problem by raising the retirement age and expanding program
coverage to all urban nonstate sectors, without implementing any fundamental reform of
the current PAYG system?
The pension fund annual balances in the baseline and in Experiments 1.1–1.3 are
shown in Figure 2. In Experiment 1.2 after expanding the sys tem coverage, the pension
fund shows a surplus during the first few years and the financial situation improves
compared with the baseline. However, the financial situation worsens in the long run,
because of a rapidly aging population. In Experiment 1.3 the retirement age for women is
raised one year every year for five years between 2011 and 2015; between 2021 and 2025
it is raised for all workers by another five years, one year at a time. Raising the retirement
age decreases the system dependency ratio and helps improve the financial situation in
the medium run. But the pension fund falls into deficit after 2035, with an annual deficit
of RMB321.6 billion in 2050 (Figure 2).
FIGURE 2 ABOUT HERE
The feasibility of raising the retirement age is questionable (it took the U.S. Social
Security system 12 years to raise the retirement age just 2 years). A critical issue is
whether additional labor supply can be absorbed into the economy. In our model all
workers are assumed to be fully employed. But raising the retirement age dramatically
14
may lead to unemployment and reduce the marginal productivity of labor. Therefore,
based on Experiments 1.1 and 1.2, it is clear that the current PAYG system is not
financially sustainable in the long run. A more fundamental reform of the current pension
system is necessary.
7.2 Transition t o Various Multi -Pillar Systems
Simulation Set 2 focuses on transitions from the baseline to various multi-pillar
systems. Experiments 2.1–2.3 are options the government is implementing (State Council
Document 26 in 1997) or experimenting with (Document 42 in 2000). Experiment 2.4 is
a new—and better—option proposed here. In each case we estimate the financing gaps
and financing options, the present value of the transition cost, and the present value of
fiscal resources needed, as a percentage of GDP.
The transition cost. Since the pillar 1 contribution is more than enough to cover
the pillar 1 benefit (with only 20% replacement rate), the transition cost is the financing
gap left by insufficient funds in the individual accounts for current retirees (the benefit
above the 20% from pillar 1) and middle-age workers as they retire. Thus the financing
gap is calculated as revenue minus expenditure in pillar 2 for current retirees and middleage men (Table 7). Obviously, the financing gap depends on the contribution level, the
benefit level, and many other parameters in each scenario. Figures 3 and 4 show the
transition cost each year and its ratio to GDP. In Experiment 2.1 the transition cost or the
financial gap is RMB73.5 billion yuan, or 0.8% of GDP, in 2000, rising to a peak of 1.2%
of GDP in 2025 and declining thereafter. If we include public institution and government
workers (Experiment 2.2), the transition cost starts from the same level in 2000, rises
more rapidly, peaks at 1.5% of GDP in 2027, and declines thereafter. Considering that
total government revenue is about 15% of GDP and continuing to rise, this transition cost
is high but manageable.
FIGURES 3 AND 4 ABOUT HERE
Financing options. From Experiments 2.1–2.4 we assume that each year the
government provides subsidies to all legitimate “transitional retirees” (contributors to the
pre-1997 PAYG system, including current retirees and middle-age workers as they
retire), so that their pension benefit level (above the 20% from pillar 1) can reach the
60% replacement rate, even though there is little or no fund in their individual accounts.
This subsidy will continue until these legitimate “transitional retirees” (“old men and
middle men,” in Chinese terminology) die out. After individuals exhaust their own
individual accounts, the government uses funds from two sources to finance this gap: the
pillar 1 surplus (about 7% of the pillar 1 contribution was designed to be used for
transition) and general tax revenue (in Experiments 2.1–2.3) or VAT revenue. A priori
relying on the pillar 1 surplus would result in high payroll tax rates, discouraging
participation in the pension system and generating more labor market distortions. Using
general tax revenues or VAT is better, as the tax base is larger and the distortions smaller.
The present values of cumulative transition cost in Experiment 2.1–2.4 are high,
ranging from RMB5.6 to RMB7.6 trillion yuan, or 62%–84% of GDP in 2000 (Table 7).
The present value of the transition cost is the highest and the fiscal subsidies the lowest in
Experiment 2.3 (the Liaoning experiment). This is because the pillar 1 contribution rate
(20 percent) is set too high—high enough to cover most of the transition cost. The
15
Liaoning experiment represents an overtaxation of enterprises’ payrolls, which
discourages enterprises from participating in the pension system and creates large labor
market distortions. This conclusion is consistent with that reached by Ma and Zhai (2001).
TABLE 7 ABOUT HERE
In Experiment 2.4 a “fire wall” is established between pillars 1 and 2, and the
government finances the transition cost entirely from VAT revenues. Individual accounts
become fully funded and individually owned, and no one can borrow from the individual
accounts of young workers for transition purposes. Experiment 2.4 is superior to the other
options because the tax burden on enterprises is reasonable, with a 13% contribution rate,
and the transition cost is financed by VAT revenue, which allows pillar 1 to become selfsustainable and individual accounts to become fully funded. As a result of less labor
market distortion and other factors, this option yields higher levels of GDP levels. The
fiscal resources needed would be higher than in the other options, but the levels would be
reasonable, ranging from RMB228 billion, or 1.7% of GDP, in 2005 to RMB773 billion,
or 0.8% of GDP, in 2045 and declining thereafter.
Required fiscal resources. In order to compare the fiscal cost of various options,
we estimate the required changes in tax rates needed for Experiments 2.1–2.4 (Table 8).
The figures in Table 8 imply changes in tax rates relative to the baseline. For example, in
Experiment 2.1 the required changes in tax rates are within 3%–6% of the baseline tax
rate. In the most dramatic case, Experiment 2.4, although the statutory VAT rate was
17% for 2000, we use the actual collection rate (8.6%) in our baseline, to reflect
extensive tax evasion and exemption. Taking evasion and exemption into account, the
required tax hike would be on the order of 0.8–1.0 percentage points. Considering that
the tax base has been expanding rather rapidly, the government could afford the required
tax hike.
TABLE 8 ABOUT HERE
7.3 Sustainability of the New Multi -Pillar System
Following the proposed reforms, a true multi-pillar pension system would be
established, with a small pillar 1 (a PAYG pillar) and a defined contribution pillar (fully
funded individual accounts). A “fire wall” should be established between pillar 1 and
pillar 2, so that no one could borrow freely from the individual accounts, which would be
managed independently from the social pooling fund in pillar 1. 12
Experiment Set 3 includes extensions based on Experiment 2.4. In Experiment 3.1
we assume that pillar 1 coverage is gradually expanded to all other urban formal sectors
and half of all urban informal sectors by 2005. In Experiment 3.2 the retirement age for
female workers is increased one year each year for five years between 2011 and 2015,
after which the retirement age for all workers is increased one year each year for five
years between 2021 and 2025. All other assumptions remain the same as in Experiment
2.4.
Results from the simulations show that the sustainability of the new pension
system is guaranteed following the proposed reforms. Table 9 shows the pension fund
annual balance for all experiments. The first section covers the annual balance for the
baseline and nonreform options, where large deficits appear at different stages. The
16
middle section covers the transition scenarios (Experiments 2.1–2.3). Pillar 1 would be in
surplus as a result of the increase in taxation of enterprises (from 17% to 20%) and a low
replacement rate (20%). The bottom section of the table shows the pillar 1 annual balance
of the new multi-pillar system (Experiments 2.4, 3.1, and 3.2). After the transition cost is
financed by VAT revenue, pillar 1 remains in surplus for more than 35 years. These
surpluses will be cumulated and invested for future use. The small deficit that appears
after 2040 could be covered by the pension fund reserves cumulated over the previous
three decades.
TABLE 9 ABOUT HERE
Experiment 2.4 shows that even without expanding coverage and raising
retirement ages, the reformed multi-pillar system is financially sustainable, with reserves
in both pillars 1 and 2 rising gradually (Table 10). Reserves in pillars 1 and 2 increase
more rapidly in Experiment 3.1, when coverage expands, and in Experiment 3.2, when
the retirement age is raised.
TABLE 10 ABOUT HERE
The proposed reforms will positively affect aggregate output as measured by GDP,
albeit only slightly (Figure 5). In Experiments 2.4 and 3.1 the new system increases GDP
in the short to medium run, largely as a result of the reduced burden (and distortion) of
payroll tax, and the improved incentives for workers. The level of GDP rises gradually,
reaching a peak in 2015 that is 2 percentage points higher than the baseline. After 2015
GDP declines. In Experiment 2.4 GDP in 2015 is about 500 billion RMB higher than
under the baseline (in 2000 constant prices)—a small effect given the size of the Chinese
economy.
GDP increases due mainly to three reasons. First, in the reform scenario, the
government needs to announce that it will utilize general tax revenue or VAT to finance
the transition gap, this will reduce the heavy burden of payroll tax. Payroll tax is know to
have the distortionary effect of discouraging labor market participation and inducing
early retirement, and its reduction will improve efficiency. Second, in the reform
scenario, individual accounts will become fully funded, privatized and portable. This will
improve incentives for workers to work more and contribute more to their individual
accounts, and reduce evasion. Third, increased savings will be invested efficiently,
because of the assumption of a perfect capital market in our model. In reality, the
efficiency depends on how pension funds are managed and regulated. Household will
benefit from the proposed reform because extra resources are used to pay the transition
gap, and pension payments to pensioners and older workers are ensured. There is a secure
funding source for the “unfunded pension liabilities” of the government, and uncertainty
is reduced.
If the retirement age is raised as in Experiment 3.2, the level of GDP would be 3.5
percentage point higher than in the baseline in 2030. However, as mentioned earlier, the
feasibility of raising the retirement age is questionable.
FIGURE 5 ABOUT HERE
17
8. CONCLUSIONS
This article examines the impacts of various design options for pension system
reform on the sustainability of the system and on economic growth in China. We
combine the CGE model and a population growth model into a recursive dynamic
framework, which provides a flexible tool for simulating various pension reform plans in
a general equilibrium setting. We then conduct three sets of simulations: limited changes
within the current PAYG system (no structural reform); transition scenarios linked to
programs the government is implementing or experimenting with or a new option
(Experiment 2.4), which proposes to finance the transition cost by VAT revenue; and two
extensions of the new multi-pillar system.
Our simulation results provide useful insights into the design of China’s pension
system reform and illustrate how CGE models can be valuable tools for evaluating
different pension systems and transition options. The results show that the current PAYG
system is not financially sustainable: in the baseline scenario the system will be in deficit
beginning in 2000. The implicit pension debt in 2000 is estimated at about 46%–63% of
GDP in 2000 yuan, depending on different assumptions on coverage and discount rates.
Expanding coverage under the current system would improve the financial situation in
the short run, but it would lead to a weakening of the financial situation in the long run.
The present value of the cumulated transition cost is estimated at 60%–80% of
GDP. If VAT revenue is used to finance the transition cost, as proposed in Experiment
2.4, both pillars 1 and 2 become financially sustainable, with significant accumulation of
reserves. The reform would have a positive impact on the level of GDP relative to the
baseline, but the impact is small in terms of growth rate.
Several important limitations need to be mentioned. First, we assume that agents
adjust their behavior according to information received in the last period. Thus the
existence of a future pension benefit does not affect agents’ saving behavior today.
Second, the model does not explicitly specify the different behavior rules of the various
types of firms in China but assumes that all operate to maximize profit. Third, the model
may overestimate the impact of various pension reform policies on macroeconomic
variables, because it does not include an explicitly specified financial market. Gross
national savings, including pension reserves, are assumed to become gross investment,
implying a perfect capital market, which is far from the reality in China. It is assumed
that pension funds can earn interest that is equal to the average return on capital. Fourth,
the simulation results may under- or overestimate the real effects of tax financing policy,
because the model does not take tax collection costs into account. Finally, the model
assumes full employment. Increases in the labor supply resulting from raising the
retirement age are assumed to be absorbed by labor demand, which is not realistic. Given
these limitations, the simulation results must be interpreted with caution and may be best
understood as indicative rather than conclusive.
18
References
Aaron, Henry J., Demographic Effects on the Equity of Social Security Benefits. Washington, DC:
Brookings Institution, 1977.
Ahmad, Ehtisham, and Hussain, Athar, “Social Security in China: A Historical Perspective.” In
Ehtisham Ahmad, Jean Drèze, John Hills, and Amartya Sen, Eds., Social Security in
Developing Countries, Oxford: Oxford University Press, 1991.
Arrau, Patricio, Social Security Reform: The Capital Accumulation and Intergenerational
Distribution Effect. Washington, DC: World Bank, 1990.
Atkinson, A.B., “Income Maintenance and Social Insurance.” In Alan J. Auerbach and Martin
Feldstein, Eds., Handbook of Public Economics, Vol.2, Amsterdam: North Holland,
1987.
Auerbach, Alan J., and Kotlikoff, Laurence J., “Social Security and the Economics of the
Demographic Transition.” In Henry Aaron and Gary Burtless, Eds., Retirement and
Economic Behavior, Studies in Social Economics Series, Washington, DC: Brookings
Institution, 1984.
Beghin, John, Dessus, Sebastien, Roland-Holst, David, and van der Mensbrugghe, Dominique,
“Prototype CGE Model for the Trade and the Environment Program-Technical
Specification.” Working Paper. Paris: Organisation for Economic Co-operation and
Development, 1994.
Brooke, Anthony, Kendrick, David, and Meeraus Alexander, GAMS: A User’s Guide. New York:
Scientific Press, 1998.
China RGSSS (Research Group for the Social Security System), Government of China, “Project
for Primary Pension Insurance for Urban Staff and Workers.” Beijing, China 1995.
Dahl, Henrik, Devarajan, Shantayanan, and van Wijnbergen, S., “Revenue-Neutral Tariff
Reform: Theory and Application to Cameroon.” CPD Discussion Paper no. 1986–25.
Washington, DC: World Bank, 1986.
Devarajan, Shantayanan, Lewis, Jeffrey D., and Robinson, Sherman, “Policy Lesson from TradeFocused, Two-Sector Models.” J. of Policy Modeling 12, 4:625–657, 1990.
Dorfman, Mark C., Lorch, Dlaus, and Zhang, Chunlin, “Linking Pension and State Enterprise
Reform in China.” Paper presented at the China Development Forum Conference on
China’s Social Security System, Beijing, August 2000.
Dorfman, Mark C., and Sin, Yvonne, “China: Social Security Reform.” Technical Analysis of
Strategic Options, Human Development Network. Washington, DC: World Bank, 2000.
Feldstein, Martin, “Social Security, Induced Retirement and Aggregate Capital Accumulation.” J.
Polit. Econ. 82, 5:905–26, 1974.
Feldstein, Martin, and Samwick, Andrew, “Social Security and Marginal Tax Rates.” Natl. Tax J.
45, 1:1–22, 1992.
Friedman, Barry, James, Estelle, Kane, Cheikh, and Queisser, Monika, “How Will China Care for
Its Aging Population?” PRD Discussion Paper. Washington, DC: World Bank, 1996.
Garbaccio, Richard F., “Reform and Structural Change in the Chinese Economy: A CGE
Analysis.” Ph.D. dissertation, University of California at Berkeley, Department of
Agricultural and Resource Economics, 1994.
Hang, Feng, “The Situation of Development of China’s Basic Pension Insurance Undertaking.”
Paper presented at World Bank seminar on pension reform, Beijing, June 2000.
Howe, Howard, “Development of the Extended Linear Expenditure System from Simple Saving
Assumptions.” European Econ. Rev. 6, 3:305–310, July 1975.
Holzmann, Robert, “Reforming Public Pensions.” Paris: Organisation for Economic Cooperation
and Development, 1988.
Hussain, Athar, “Reform of the Chinese Social Security System.” In Ehtisham Ahmad and Athar
Hussain, Eds., Social Security in the People’s Republic of China, Oxford: Oxford
University Press, 1993.
19
James, Estelle, “Income Security for Old Age: Conceptual Background and Major Issues,” Policy
Research Department, Poverty and Human Resources Division. Washington, DC: World
Bank, 1992.
James, Estelle, “New Systems for Old Age Security,” Policy Research Working Paper no. 1766.
Washington, DC: World Bank, 1997.
James, Estelle, “New Models for Old Age Security—and How Can They Be Applied in China?”
Paper presented at World Bank workshop, Beijing, June 1–5, 1999.
James, Estelle, “How Can China Solve its Old Age Security Problem? The Interaction between
Pension, SOE and Financial Market Reform.” Paper prepared for conference on financial
sector reform in China, Harvard University, 2001.
Kotlikoff, Laurence., “Privatizing Social Security at Home and Abroad.” Discussion Paper,
Boston University and NBER, 1996.
Kotlikoff, Laurence J., “Generational Accounting in Europe-An Economic, Social and Moral
Imperative,” Working Paper, Boston University. Cambridge, MA: NBER, 1997.
Kotlikoff, Laurence J., “Simulating the Privatization of Social Security in General Equilibrium.”
In Martin Feldstein, Ed., Privatizing Social Security . NBER Project Report, pp. 65–306.
Chicago: University of Chicago Press, 1998.
Kumar, Anjali and Wang, Yan, “Pension Fund Management and China’s Financial Sector.” In
Jason Z. Yin, Shuanglin Lin, and David F. Gates, Eds., Social Security Reform: Options
for China, pp.153–180. River Edge, NJ: World Scientific Publishing Co., 2000.
Leslie, P.H., “On the Use of Matrix in Certain Population Mathematics.” Biomatrika 33:183–
212, 1945.
Ma, June and Zhai, Fan. “Financing China’s Pension Reform.” Deutsche Bank and Ministry of
Finance of China. Paper prepared for conference on financial sector reform in China,
Harvard University, 2001.
MOLSS (Ministry of Labor and Social Security), YangLao BaoXian. Beijing, 2001.
Pereira, Alfredo, and Shoven, John, “Survey of Dynamic Computational General Equilibrium
Models for Tax Policy Evaluation.” J. of Policy Modeling 10, 3:401–436, 1988.
Pomfret, Richard, Is China a “Large Country”? China’s Influence on World Markets. Paris:
Organisation for Economic Cooperation and Development Development Centre, 1997.
Research Group for the Social Security System, (RGSSS) Government of China, “Project for
Primary Pension Insurance for Urban Staff and Workers.” Beijing, 1995.
Sargent, Thomas J., and Wallace, Neil, 1995. “Rational Expectations and the Theory of
Economic Policy.” In Saul Estrin and Alan Marin, Eds. Essential Readings in
Economics, pp. 366–82. New York: St. Martin’s Press, 1995.
Seidman, Laurence S., “A Phase-Down of Social Security: The Transition in a Life Cycle Growth
Model.” National Tax J. 39, 1:97-107, March 1986
Wang, Zhi, and Kinsey, Jean, “Consumption and Saving Behavior under Strict and Partial
Rationing.” China Econ. Rev. 5, 1:83–100, 1994.
Wang, Zhi, and Zhai, Fan, “Tariff Reduction, Tax Replacement, and Its Implication for Income
Distribution in China.” J. of Comp. Econ. 26, 2:358–87, 1998.
Whalley, John, Trade Liberalization among Major World Trading Areas. Cambridge, MA: MIT
Press, 1985.
World Bank, Averting the Old Age Crisis: Politics to Protect the Old and Promote Growth.
Washington, DC: Oxford University Press, 1994a.
World Bank, World Population Projections, 1994–95. Baltimore: John Hopkins University
Press, 1994b.
World Bank, Old Age Security: Pension Reform in China. China 2020 Series. Washington, DC,
1997.
Zhou, Xiaochuan, “Social Security: Economic Analysis and Reform Suggestions.” Reform 3,
1994.
20
TABLE 1
Characteristics of China’s Public Pension System, 2000
(percent)
Sector
State and collective
enterprises
Other urban formal sector
Government and public
institutions
Employee
coverage
100
62.5
25.4
Retiree
Wage/labor
coverage
income
100
81.2
100
17.6
81.2
79.8
a. Ratio of pension benefits to average wage.
Source. China Labor Statistics Yearbooks (2001) and authors’ estimation.
21
Contribution rate
(employee plus
Replacement
employer)
rate a
22.5
71.5
23.3
23.3
70.0
94.5
TABLE 2
Assumptions on Baseline Calibration
Category
Macroeconomic
variables
•
•
•
•
Assumptions
Real GDP is exogenous in the baseline.
TFP growth rate is endogenous.
Gross investment rate declines from 36.9% of GDP in 2000 to 30% of GDP in 2050.
Trade balance gradually declines to zero in 2020.
Demographic
parameters
•
See Table 3.
Pension system
characteristics
•
•
PAYG system.
Pension system coverage, contribution rates, and replacement rates are all fixed at the
base year level (see Table 1).
Share of wage in employee’s labor income increases from 80% in 2000 to 100% in
2010.
Compliance rate is 90%.
Pension benefit is indexed to the average growth rate of the CPI and the average
wage.
Pension expenditure is financed by payroll tax and general tax revenue.
•
•
•
•
Government
fiscal closure
•
•
•
Government consumption, transfers, and saving (excluding saving/deficits for
pension) are exogenous.
Corporate tax rate is endogenous to achieve government budget balance.
All other tax rates are fixed at base year level.
Source. Authors.
22
TABLE 3
Assumptions on Demographic Parameters, 2000–2050
Urban
Year
2000
2010
2020
2030
2040
2050
Rural
Life
Life
Total
Life
Life
expectancy, expectancy, fertility
expectancy, expectancy,
male
female
rate
male
female
70.98
76.3
1.39
68.79
73.33
72.98
78.63
1.60
71.42
76.44
74.23
79.52
1.58
72.85
78.32
75.13
80.38
1.58
74.25
79.78
76
81.22
1.57
75.42
80.65
76.86
82.05
1.57
76.29
81.49
Source. China Population Information and Research Center.
23
Total
fertility
rate
2.10
2.24
2.22
2.21
2.20
2.20
Rural–urban
migration
(millions of
people)
7.08
7.67
10.04
12.25
7.57
6.24
Simulation
Set 1
Experiment 1.1
TABLE 4
Summary of Policy Simulations
Description
Limited reform in the current PAYG system
Raising contribution rates and lowering pension benefits to be consistent
with rates defined in State Council Document 26 (1997):
• Contribution rate: 8% of wage by employee and 20% of wage by
employer.
• Replacement rate: 60%.
• Status quo is maintained regarding transition (funds in individual
accounts are used to finance transition cost). System is thus still a
one-pillar PAYG system, and the defined-contribution pillar does
not exist.
Experiment 1.2
Experiment 1.1 combined with expansion of PAYG system to urban
formal sector and half of urban informal sector in 2005.
Experiment 1.3
Experiment 1.1 combined with higher retirement age. Between 2011 and
2015 retirement age for female workers rises one year annually. Between
2021 and 2025 retirement age for all workers rises one year annually.
Set 2
Experiment 2.1
Transition to a multi-pillar system
Transition to the multi-pillar system defined by State Council Document
26 (1997):
•
Pillar 1 (PAYG): 17% contribution by enterprises, 20%
replacement rate.
•
Pillar 2 (individual accounts): 3% contribution by enterprises plus
8% contribution by individuals, providing an annual benefit of
10% of reserves in individual accounts.
•
Government subsidizes the pensions of old and middle -age men
and for those whose benefits cannot reach the stipulated 60%
replacement rate.
•
Subsidies are financed by surplus of pillar 1 and general tax
revenue.
Experiment 2.2
Experiment 2.1 combined with expanded coverage to other urban formal
sector and public organizations and government workers in 2005.
Experiment 2.3
Transition to the multi-pillar system defined by State Council Document
42 (2000)
• Pillar 1 (PAYG): 20% contribution from enterprises, 20%
replacement rate.
• Pillar 2 (individual accounts): 8% contribution from individuals,
providing an annual benefit at 10% of reserves in the individual
account.
• Other assumptions are the same as Experiment 2.1.
24
Experiment 2.4
(proposed new
system)
Transition to a new multi-pillar system; government finances all transition
costs through VAT revenue:
• Pillar 1 (PAYG): 13% contribution from enterprises, 20%
replacement rate.
• Pillar 2 (individual accounts): 3% contribution by enterprises plus
8% contribution from individuals, providing an annual benefit of
10% of reserves in the individual account.
• Surplus of pillar 1 would be accumulated as the reserve of pillar 1
and would not be used to subsidize the transition cost.
• Other assumptions are same as Experiment 2.1.
Set 3
Experiment 3.1
Sustainability of the new multi-pillar system, based on Experiment 2.4
Experiment 2.4 combined with expanded coverage to all urban formal
sector and half of urban informal sector in 2005.
Experiment 3.2
Experiment 2.4 combined with higher retirement age. Between 2011 and
2015 retirement age for female workers rises one year annually. Between
2021 and 2025 retirement age for all workers rises one year annually.
Source. Authors.
25
TABLE 5
Baseline Calibration, 2000–2050
Year
2000 2005 2010 2015 2020 2025 2030 2035 2040 2045 2050
Annual growth rate (percent)
GDP
n.a. 7.18
6.73 6.29
5.86 5.43
4.98
4.54
4.09
3.64
3.18
Labor
n.a. 1.45
0.87 0.39
0.01 -0.23 -0.25 -0.34 -0.53 -0.59 -0.51
Capital
n.a. 9.23
8.06 7.24
6.51 5.83
5.19
4.62
4.09
3.60
3.13
TFP
n.a. 2.23
2.35 2.22
2.03 1.83
1.61
1.46
1.34
1.19
1.01
Participation rate (percent)
Old age dependency ratio
(percent) 65+ /15-64
Old age dependency ratio 2
(percent) 60+/ 15-59
System dependency ratio
(percent)
Replacement rate (%)
State and collective
enterprises (%)
Other urban formal
enterprises(%)
Public sector(%)
79.9
80.1
79.6
79.1
78.8
77.0
76.5
77.6
78.0
76.7
75.9
10.6
11.4
12.4
14.5
18.2
21.0
25.3
31.5
36.6
38.1
39.2
16.1
17.0
19.4
24.0
27.7
33.3
41.0
47.7
50.4
52.7
56.9
35.4
72.6
35.1
73.0
36.2
73.3
38.6
73.6
43.0
73.7
48.7
73.8
53.0
73.7
56.0
73.7
59.9
73.7
65.1
73.7
69.2
73.7
71.6
71.6
71.6
71.6
71.6
71.6
71.6
71.6
71.6
71.6
71.6
70.2
94.5
70.2
94.5
70.2
94.5
70.2
94.5
70.2
94.5
70.2
94.5
70.2
94.5
70.2
94.5
70.2
94.5
70.2
94.5
70.2
94.5
Pension (RMB10 billion, in 2000 yuan)
Revenue
17.2 25.8
Expenditure
21.2 30.6
Annual balance
-4.0 -4.8
Accumulated reserve
9.7 -12.3
Balanced contribution rate
(percent)
28.0 27.0
36.3 43.5 50.3 56.9 63.9 71.3 78.5 84.3 90.1
42.9 58.7 78.3 102.5 126.9 149.4 174.0 203.1 235.3
-6.6 -15.1 -28.0 -45.6 -63.0 -78.1 -95.5 -118.8 -145.2
-45.9 -111.7 -245.9 -484.2 -860.1 -1384.0 -2070.2 -2966.1 -4126.3
27.0
30.7
35.5
n.a. Not applicable.
Source. Authors’ calculations based on assumptions in Tables 2 and 3.
26
41.1
45.3
47.8
50.5
54.9
59.5
TABLE 6
Estimated Implicit Pension Debt under Various Assumptions
Description
Excluding public institution and government workers,
discount rate = interest rate
Including public institution and government workers,
discount rate = interest rate
Excluding public institution and government workers,
discount rate = 5 %
Including public institution and government workers,
discount rate = 5 %
Excluding public institution and government workers,
discount rate = 4.5 %
Including public institution and government workers,
discount rate = 4.5 %
Implicit pension debt
RMB10 billion in Percentage of
2000 yuan)
GDP
431.6
47.8
570.4
63.1
415.8
46.0
549.4
60.8
436.4
48.3
576.6
63.8
Source. Authors’ calculations based on assumption that the system terminates in 2000 and Table 2
assumptions about baseline.
27
TABLE 7
Transition Cost and Its Financing under Various Options
Experiment
Experiment 2.1
Present value of accumulated transition cost
Pillar 1 surplus + reserve
General tax revenue
RMB10 billion yuan
Percentage of 2000 GDP
56,3.96
28,2.09
28,1.87
62
31
31
Experiment 2.2
Present value of accumulated transition cost
Pillar 1 surplus + reserve
General tax revenue
70,2.81
38,9.51
31,3.30
78
43
35
Experiment 2.3
Present value of accumulated transition cost
Pillar 1 surplus + reserve
General tax revenue
76,2.97
54,8.42
21,4.55
84
61
24
Experiment 2.4
Present value of accumulated transition cost
Pillar 1 surplus + reserve (not an option)
VAT
68,9.36
n.a.
68,9.36
76
n.a.
76
n.a. Not applicable.
Source. Authors’ estimates.
28
TABLE 8
Tax Burden for Financing Transition Costs, 2000–2050
(Percentage Change Relative to Baseline)
2000
2005
2010
2015
2020
2025
Financing transition cost by transfer from pillar 1 and general tax
Experiment 2.1
0
3.7
3.3
4.2
4.9
5.1
Experiment 2.2
0
2.0
2.2
4.1
5.5
6.2
Experiment 2.3
0
0
0
0
3.8
5.2
Financing by VAT only
Experiment 2.4
0
28.6
30.0
30.1
28.2
24.7
Source. Authors’ estimates.
29
2030
2035
2040
2045
2050
4.5
5.6
5.3
3.3
4.3
4.2
2.3
3.0
3.0
5.8
7.5
5.9
4.5
5.8
4.3
19.8
14.6
10.2
15.7
10.8
TABLE 9
Annual Projected Balance of Pension Fund under Various Options, 2000–2050
(RMB10 billion)
2000
2005
2010
2015
2020
2025
Current PAYG system (pillar 1 and notional account together)
Base
-3.95
-4.81
-6.63 -15.11 -28.02 -45.63
Experiment 1.1
-3.95
3.00
6.04
2.42
-4.83 -15.92
Experiment 1.2
-3.95
14.97
17.63
11.36
-0.20 -17.08
Experiment 1.3
-3.95
3.00
6.04
6.61
4.22
10.47
Transition options, pillar 1 only
Experiment 2.1
3.39
12.56
17.87
19.34
19.02
17.22
Experiment 2.2
3.39
19.47
26.11
27.54
26.53
23.53
Experiment 2.3
3.39
24.59
33.32
36.28
36.38
34.56
New multi-pillar system, after transition cost financed by taxation, pillar 1 only
Experiment 2.4
3.39
12.45
16.27
15.87
13.26
8.77
Experiment 3.1
3.39
15.97
19.85
19.20
16.04
10.79
Experiment 3.2
3.39
12.45
16.27
17.89
17.73
21.60
Source. Authors’ estimates.
30
2030
2035
2040
2045
2050
–63.02
–26.65
–33.26
10.39
–78.10
–35.44
–46.38
2.84
–95.46 –118.84 –145.16
–46.32 –62.75 –81.56
–62.08 –85.51 –112.48
–7.74 –17.80 –32.16
15.73
20.87
33.17
15.22
19.56
33.25
13.86
17.43
32.52
9.38
11.66
27.89
4.35
5.45
22.85
4.50
5.90
22.82
1.39
2.48
20.34
–2.54
–1.85
16.56
–9.54
–9.86
13.13
–17.35
–18.85
7.22
TABLE 10
Sustainability of the Proposed New Multi-Pillar System, 2000–2050
(10 billion yuan)
2000 2005
2010
2015
2020
2025
2030
2035
2040
2045
2050
New multi-pillar system, after transition cost financed by taxation
Experiment 2.4
GDP
903.5 1,289.5 1,800.5 2,450.8 3,257.5 4,234.5 5,387.8 6,711.5 8,185.1 9,768.1 11,418.5
Pillar 1
Annual balance
3.4
12.4
16.3
15.9
13.3
8.8
4.5
1.4
–2.5 –9.5 –17.3
Reserve
9.7
48.4
141.2
267.8 409.9
554.1 691.4
824.6 957.6 1079.7 1176.2
Pillar 2
Reserve
0.0
63.1
194.6
377.0 587.1
801.6 995.7 1,154.2 1,265.1 1,377.3 1,560.5
Experiment 3.1
GDP
903.5 1,291.1
Pillar 1
Annual balance
3.4
16.0
Reserve
9.7
55.3
Pillar 2
Reserve
0.0
69.2
Experiment 3.2
GDP
903.5 1,289.5
Pillar 1
Annual balance
3.4
12.4
Reserve
9.7
48.4
Pillar 2
Reserve
0.0
63.1
1,804.2 2,455.8 3,262.7 4,238.8 5,390.7
6,713.0 8,185.0 9,765.9 11,415.2
19.9
170.1
19.2
323.2
16.0
494.3
10.8
668.1
5.9
834.5
2.5
–1.8 –9.9 –18.9
997.9 1163.5 1319.1 1447.3
221.5
431.5
672.9
919.8 1,144.0
1,329.3 1,462.4 1,605.9 1,828.1
1,800.5 2,460.4 3,284.5 4,301.9 5,497.2
6,850.1 8,350.2 9,966.7 11,630.4
16.3
141.2
17.9
271.1
17.7
429.9
21.6
616.1
22.8
846.6
20.3
16.6 13.1
7.2
1,104.0 1,377.6 1,669.8 1,977.5
194.6
379.5
601.6
851.2 1,128.6
1,394.5 1,616.6 1,784.1 1,885.3
Source. Authors’ estimates.
31
Population
(Mn person)
1200
45%
40%
1000
35%
800
30%
15-64
65 and above
Old dependence ratio
600
25%
20%
400
15%
10%
200
5%
0%
20
00
20
03
20
06
20
09
20
12
20
15
20
18
20
21
20
24
20
27
20
30
20
36
20
36
20
39
20
42
20
45
20
48
0
FIG. 1. China’s aging population, 2000–2050.
Source. Authors.
10 bn RMB yuan
40
20
0
2000
2005
2010
2015
2020
2025
2030
2035
2040
2045
-20
-40
-60
-80
-100
-120
Base Case
Exp 1.1
Exp 1.2
Exp 1.3
-140
-160
FIG. 2. Pension fund annual balance without major reform. 2000–2050.
Source. Authors’ estimates.
32
2050
100.0
1.6
90.0
1.4
80.0
60.0
1.0
50.0
0.8
40.0
0.6
30.0
0.4
2050
2048
0.2
2046
2044
2042
2040
2038
2036
2034
2032
2030
2024
2022
2020
2018
2016
2014
2012
2010
2008
2006
2004
2002
0.0
2000
10.0
2028
Deficits in pillar 2 for middleaged workers
Ratio to GDP
20.0
0.0
FIG 3. Transition cost defined as deficits in pillar 2 for middle-age workers, excluding
public institutions and government workers, 2000–2050.
Source. Authors’ estimates.
33
(%)
1.2
70.0
2026
(10 Bn Yuan, 2000 price)
Transitional Costs - Excluding Public Institutions
simulation results from Exp. 2.1
100.0
1.6
90.0
1.4
80.0
60.0
1.0
50.0
0.8
40.0
Deficits in pillar 2 for
middle-aged workers
Ratio to GDP
30.0
20.0
0.6
0.4
20
48
20
45
20
42
20
36
20
39
20
33
20
30
0.0
20
24
20
27
0.0
20
21
0.2
20
12
20
15
20
18
10.0
20
09
(%)
1.2
70.0
20
00
20
03
20
06
(10 Bn Yuan, 2000 price)
Transitional Costs - Including Public Institutions
simulation results from Exp. 2.2
FIG 4. Transition cost defined as deficits in pillar 2 for middle-age workers, including
public institutions and government workers, 2000–2050.
Source. Authors’ estimates.
Percentage change in GDP relative to base case
4.0
3.5
3.0
2.5
2.0
1.5
Exp 2.4
1.0
Exp 3.1
Exp 3.2
0.5
0.0
2000
2005
2010
2015
2020
2025
2030
2035
2040
2045
FIG 5. Percentage changes in the level of GDP relative to the baseline, 2000–2050.
34
2050
Source. Authors’ estimates.
WB21569
C:\Documents and Settings\WB21569\My Documents\research\China pension reform edited 12-20.doc
December 21, 2002 15:19
Endnotes
1. Statistics from the Ministry of Labor and Social Security indicate that the total deficit of pension funds
increased from a little more than RMB10 billion in 1998 to almost RMB40 billion in 2000 (MOLSS,
2001). In 1999 the Ministry of Finance transferred RMB17 billion (US$2 billion) to cover pension
shortfalls at the local level. This transfer has been increasing in recent years.
2. On China’s pension reform, see Ahmad and Hussain (1991); RGSSS (1995); Dorfman and Sin (2000);
Friedman and others (1996); Hussain (1993); James (1997, 1999, 2001); World Bank (1997); and Zhou
(1994).
3. Government liabilities can be classified into four categories: explicit and direct (formal debt, budgeted
expenditure); explicit and contingent (government guarantees and deposit insurance schemes); implicit and
current (social insurance expenditure); and implicit and contingent (default of state-owned enterprises,
banks, and social insurance funds).
4
These problems were examined in many seminal works on pension systems. See, for example, Aaron
(1977); Arrau (1990); Atkinson (1987); Auerbach and Kotlikoff (1984); Feldstein (1974); Feldstein and
Samwick (1992); Holzmann (1988); James (1992, 1997); Kotlikoff (1996, 1997, 1998); and Seidman
(1986). For a literature review, see World Bank (1994a).
5
This method was implemented briefly when the State Council issued a document on “Divesting the State
Shares in State-Owned Enterprises to Finance the Social Security Fund” (People’s Daily June 16, 2001, p.
5). The document stipulated that beginning June 13, when new shares are issued as part of a state-owned
enterprise’s initial public offering (IPO) or additional share offerings (domestic or foreign) are offered, the
state will sell a portion of its shares in the state-owned enterprise equivalent to 10% of the IPO proceeds.
These privatization proceeds will be transferred to the newly established National Social Security Fund.
This promising reform measure was later suspended because of collapsing stock prices. This approach is
expected to resume once the methodological issues are resolved.
6
See Dorfman, Lorch, and Zhang (2000) on China.
7
A survey of the application of the CGE model in taxation and public policy appears in Pereira and Shoven
(1988).
8
The model is implemented using the General Algebraic Modeling System (GAMS) (Brooke, Kendrick,
and Meeraus, 1988). Due to space limitations, only the major characteristics of the model are presented. A
detailed algebraic specification is available from the authors upon request.
9
The Chinese economy is a mixed economy under transition in which the state sector plays a declining role
in production and employment and the nonstate sector is expanding rapidly. The multi-ownership structure
35
of the economy is expected to persist for some time. China’s pre-1995 pension system covered only stateowned enterprises. Thus an analysis of pension system reform in such an economy requires a model that
differentiates production and employment by ownership types.
10
Workers 15–34 are classified as young, workers 35–49 are classified as middle-age, and workers age 50
and older are classified as old.
11
The base year data set and key parameters used in the model are available from the authors upon request.
12
Pension fund management is beyond the scope of this article. For details, see the proposals in Chapter 4
of World Bank (1997) and Kumar and Wang (1999).
36