Docstoc

multi_page

Document Sample
multi_page Powered By Docstoc
					Public Disclosure Authorized




                                                                                  ReportNo.:   17348 CH.A
Public Disclosure Authorized




                                                                   CHINA

                                             SOCIAL SECTOR EXPENDITURE REVIEW
Public Disclosure Authorized




                                                               FEBRUARY1996
Public Disclosure Authorized




                               Rural and Social Development Operations Division
                               China and Mongolia Department
                               East Asia and Pacific Regional Office
                               The World Bank                                     FIL      WPY
                                 TECHNICALANNEXD



1.                       RegressionEstimatesof the Determinants
                                   of InfantMortality
                                 (29 Provinces1982-92)
                          Dependentvariable = Infantmortality



       Right hand                  Fixed-effect
     side variables                  Model                T-statistics


     Per capitaIncome               -0.448                  -6.05
     Sanitation                     -0.270                  -3.64
     DPT 3                          -0.165                  -1.91
     Per capitaHealth spending      -0.564                  -7.94
     InteractionincomeX
       health spending              -0.190                  -7.94

     AdjustedR2                      0.970




2.             The followingreports containthe results on the impactof public expenditure
                             on
and other policy interventions severalhealth outcomeswith different econometric
specifications.


Malaysia: "Fiscal Reforms for StableGrowth" April 1992(No. 10120-MA),The World
Bank.

The Philippines:An Openingfor SustainedGrowth,April 1993 (No. 11061-PH),The World
Bank.

Viet Nam: "Impactof Public Policyon Health Outcomes"March 1994 (mimeo), The World
Bank.
   CURRENCY EQUIVALENTS

    Currency Unit = Yuan(Y)

    Average 1994 = 8.41 Yuan




         FISCAL YEAR

     January 1 - December 31


   WEIGHTS AND MEASURES

          Metric System




ABBREVIATIONSAND ACRONYMS


COE =    collectivelyowned enterprise
GEIS =   GovernmentEmployeesInsuranceSystem
MOCA =   Ministry of Civil Affairs
MOF =    Ministry of Finance
MOL =    Ministry of Labour
MOP =    Ministry of Personnel
MOPH =   Ministry of Public Health
NHS =    National Health Survey
SEdC =   State EducationCommission
SOE =    State Owned Enterprise
SPC =    State Planning Commission
SSB =    State StatisticalBureau
STS =    SecondaryVocationalSchools
                                            TABLE OF CONTENTS

                                                                                                                  Page

INTRODUCTION............................................                                                            I

I.    EDUCATION ...........................................                                                         2

            Overall Development ...................................                                                 2
            Education Cost and Financing .          ..............................                                  2
            Equity Aspects of EducationExpenditures .......                             ...............             8
            Outcomes and Efficiency . ................................                                             13

I. HEALTH            .............................................                                                 20

            Equity Aspects of Health Expenditures ........................                                         25
            Outcomes and Efficiency .................................                                              27

III. SOCIAL SECURITY .                         ......................................                              31

            Introduction ......................................                                                    31
            Overview of the System . ................................                                              32
            Social Insurance .....................................                                                 32
            Social Assistance .....................................                                                35
            Public Expenditure: Trends, Structure and Composition ....                               ..........    36

IV.    MAIN ISSUES AND STRATEGIC DIRECTIONS.40

            Education ...........................................                                                  40
            Health .........................                                                                       44
            Social Security ...........................                                                            52
            Program of Pension Reform ..........................                                                   52
            UnemploymentInurance ..........                      ...............                                   59

ANNEX TABLES .........................                                                                             61
                                      PREFACE


      This report was prepared by Shaikh Hossain (Task Manager) with contributions
from Sonia Xiaosong Zhao, and Duan An. Helene Genest and Carlina Jones also
provided assistance. The report is based on the findings of missions to China in October
1994 and February 1995. The World Bank would like to express its gratitude to
counterpart teams in China who provided valuable information for the report in various
ways.

       The counterpart team was led by Mr. Jin Li Qun, Director, World Bank
Department, Ministry of Finance and comprised the following lead officials: Ms. Zou
Jiayi, Ministry of Finance, Mr. Sun Ping Sheng, State Education Commission, Mr. Liu
Peilong, Ministry of Public Health, Mr. Bai Li, Ministry of Labour, Mr. Zhang Zhong,
Ministry of Social Welfare and Mr. Wang Dong Sheng, State Planning Commission
(Beijing); Mr. Zhang Hexi, Ms. Lee, Bureau of Finance Mr. M.A Zhenhai, Provincial
Education Commission Mr. Yuan Dong-he, Department of Public Health, and Mr. Diao
Yu Hua, Education Commission (Zheng Zhou, Henan Province); Mr. Qu Bao Yuan,
Bureau of Finance, Mr. Wang Hin Guo, Bureau of Finance and Mr. Liu Ai Hua, Bureau
of Health (Chang Sha, Hunan Province).

       A panel of reviewers include Tamar Manuelyan, Jeffrey Hammer, Anjali Kumar,
Albert Keidel, Halsey Beemer, Alan Piazza and Yuan Wang. Carl Jayarajah, William
McGreevey, Vinay Bhargava and Jagadish Upadhyay provided useful comments.

      The report was discussed with the Chinese Government in November 1995.
Joseph Goldberg is the Division Chief and Nicholas Hope is the Department Director.
                              EXECUTIVE SUMMARY

                                      Background

1.       As China consolidates the momentum of its recent macroeconomic gains and
drives more openly into the market environment, the quality of human resources will be
an important determinant of the country's competitiveness. China has made impressive
gains in human resource development in the past two decades, and continuing its strides
will help redress poverty, both indirectly by increasing the productivity and efficiency
of its labor force and directly by fostering the earning capacity of the poor. Eventually
enhanced human resources will spur economic growth and welfare. Yet, its progress has
been mixed in recent years, mainly because of two broad problems. First, the distribution
of services continues to discriminate against the poor, largely because they have not
benefitted from rapid economic growth and are being choked off from service access.
Second, some current mechanisms and arrangements for devolving government financing
for services have created over-decentralization, and the loss of central government
control has led to targetting inefficiencies. As such, the government's priorities are to
enhance financing and improve the equity, efficiency and quality of social services.

                                     A. Education

2.      Overall DeveloRment. Educational development in China has been quite
inpressive. By the early 1960s, China achieved virtually universal enrollment in primary
education, a coverage rate that still outpaces the rates found in several Asian
counterparts. But, qualitative indicators, such as average grade attainment of the current
school-age population and cohort survival rate, although showing steady progress, are
modest when compared with those of Malaysia, Korea, Thailand and Sri Lanka. For
instance, average grade attainment is close only to India but significantly lower than its
close comparators. Likewise, cohort survival rate is 68 percent, quite lower than its close
comparators and similar only to the Philippines.

3.      Educational Expenditure. At least three important features characterize the
composition of public educational expenditure over the past decade. First, recurrent
expenditures have increased in real terms; the bulk going towards such "autonomous"
categories as teachers' salaries and gradually meager proportion is being used for
instructional inputs and supplies of basic materials and equipment for schools. Second,
real capital expenditures have been volatile and even fell on average between 1985 and
1991. And, third, the share of educational expenditure in GNP and government
expenditure is low compared with neighboring Asian countries that have a similar income
per capita and economic structure.

Equity and Efficiency

4.      (a) Structural Equity. Three important characteristics have shaped a structurally
equitable educational system in China; a flatter enrollment pyramid--nearly universal
entry into basic education and lower enrollments in higher education; and, more
instructive, a pattern of government spending by educational level that conforms to the
                                             ii

priorities of the government--morespending in basic education than in higher level.
Another feature, althoughnot yet very significant,is the growingprivate enrollmentsin
higher education, suggestingthat the proportion of fees being used to finance post-
primary levels of education is increasing. A single summary indicator of structural
equity obtainedby applying the annual subsidiesper student to projected graduates by
level shows that 10 percent get 26 percent of cohort resources and 90 percent receives
the balance, reflectingan egalitariandistributionof public subsidyin education.

5.      Yet, the structural equity has not translated into equitable educationalbenefits
across provinces.Whilethe benefitsof primary educationa relativelytargettedto poor
                                                           are
provinces,accessibility post-primary
                       to              educationshowsmarkedinterprovincial    variance.
Per-pupil expendituresare disproportionately  higher in rich provinces comparedto those
in poor provinces. The richest quintile of provinces Shanghai, Beijing, Tianjin,
Guangdongand Liaoningspent almost 1.5 times more than the poorest provinces such
as Guizhou, Guangxi, or Henan.

       (a) Access to Education. The enrollment-ratedisparity between rich and poor
provinces grows as studentsmove on to secondaryand higher levels. In the absence of
compulsionin the subsequentage groups the gap in attendance rates widens steadily.
Urban children and those from high-incomeprovinces are much more likely to attend
secondaryschool(about2.2 times)comparedto children fTom  rural areaand low-income
provinces.

        (b) Social Selectivit . Disparitiesin drop-outrates at the secondaryeducation
level are quite pronounced . Male-femaledifferencesin primary education are modest
but very significantin secondaryeducation. Drop-outraitesalso widely vary according
to geographic location and level of education: high in poor provinces and secondary
education, and vice versa. In secondaryeducation, in rural areas, more females tend to
drop out of school than do males. Economic difficulties and inadequate learning
environmentin poorer provincesare chief reasons for early exit from schoolsystem.

6.      (c) Intemal Efficiency. At the aggregate level, educationaloutcomes-rates of
cohort survivaland transition--vary directlywith income. The rate of survivalto the end
of the primnary  cycle is low in poor provinces than in better-off provinces. Eclectic
statistical anaysis strengthen the conclusion that public expenditures are important
determinants of tansition from one level to another. The results indicate that keeping
budgetaryallocationto primary educationat a high enoughlevel will compensatefor the
negative effect of low income by making the likelihoodof transition to secondaryand
higher levels of schoolingmore equal throughoutthe couatry.

7.      (d) External Efficiency. Primary education exhibits the highest social rate of
return (14.4%), followedby secondaryeducation(12.9%), implyingthat the demand for
labor with these educationallevels is still high. Sinceprivate returns are also on the high
side, investmentin these levels of schooling will promote both equity and efficiency
objectives. However, return to higher education is below the returns to investment in
                                            iii
alternative formns of capital, casting doubt whether further public subsidies in this
subsector are justifiable. Part of the problem lies with very narrow access to tertiary
education. The students who exit from education system create an oversupply of labor
with vocational, technical and senior secondary schooling, thereby affecting their wages
and eventually dampening returns on investment.

                                       B. Health

8.      From 1983 to 1993, three salient features have characterized public health
financing in China. First, real spending on health grew modestly, yet as a percentage of
GNP it actually declined from 0.77 to 0.48. Moreover, health spending is low when
compared with that in neighboring Asian countries whose income levels are similar.
Second, capital expenditure in real terms fell; the bulk has gone to medical equipment
purchases in hospitals above the county level. And, third, the ratio of expenditures on
curative health care to preventive health services is very high.

9.      Sources of Financing. There has been a continuous decline in the government
budget for health, and the dramatic rise in patient contributions. Of total health
expenditures, the share of the government budget dropped from 25 percent in 1980 to
only 14 percent in in 1993, while the share of household contributions (patients' fees)
rose from 23 percent to 37 percent; the expenditure share in insurance sponsored by state
enterprises (labor insurance) increased from 30 percent to 37 percent. The introduction
of a user-fee policy and the dismantling of the commune health system have been
responsible for this shift in the sources of financing health expenditures from the
government to SOE insurance and patient fees.

10.     With the devolution of health services financing, local governments now finance
more than over 80 percent of total government expenditures in health. But as revenue
collection and expenditure allocation effectively came under the control of local
governments, they shifted health resources toward greater spending on income-generating
investment. Indeed, this practice led to a precipitous drop in their share of total health
financing, from 24 percent in 1980 to only 12.9 percent in 1993. Moreover, since
collected revenue is shared up from the local governments to the center, the central
govermmentis essentially recipient, and has thus had little leverage to increase spending
on the health sector. Consequently, the central government's share of total health
expenditures declined from its peak of 1.8 percent in 1986 to only 0.8 percent in 1993.

11.     The dramatic rise in expenditure contributions by GEIS and labor insurance
schemes is attributable to the rising number of claims from individuals age 60 and older,
a legacy of declining fertility rates and then attendant aging of the population; but it is
also due to the higher prices of drugs and medical supplies, and to purchase expensive
medical equipment in tune with technological advancements. Moreover, "moral hazard"
factors have escalated the costs of health services even further--that is the existence of
insurance schemes have created incentives for health providers to offer more care than
                                            iv

patients actually need. Also, substantial revenues from drug sales provide incentives for
hospitals to over-prescribe.

Eguity and Efficiency

12. Eguit. The average number of inpatient and outpatient visits by province,
ranked by per capita income from poor to rich show that the redistributive, or target,
impact of public health spendingwas highest for users in the poorest provinces in 1980,
but these indicators are at odds with an apparent reduction in usage among poorer
provinces in 1993. The elimination of the rural cooperative health scheme and the
introductionof user fees prompted costs of access to health services to rise among the
poor.

 13. Accordingto the 1993NationalHealthSurvey(NHS), utilizationvaries positively
with household income per capita. Province-widedata indicate greater under-use of
servicein poorerprovinces is associated  with economicreasons,and vice versa. Of total
households,58.8 percent in rural areasand 39.8 percent in urban areas do not use public
healthfacilitiesdue to economicdifficulties.Another factor is the deteriorationin quality
of public health servicesdue to the reduced public spendingsince the mid-1980s. The
emergenceof private sector and liberalizationof pharmaceuticalsales permitted users,
particularlythe rich, to move away from public sector to seek costly and perhaps better
quality private health care providers. A third factor may be the declining age-specific
morbidity rates, requiring fewer visits to health facilities. But, age and gender-specific
mortalityand morbidityrates have shiftedfrom infectiousto more endemicdiseases,and
not declined dramaticallyto warrant such marked reduclionin the utilization of health
services. This finding clearly indicatesmore interventionin preventive serviceswould
be desirable on both equity and efficiencygrounds.

14.     Outcomesand Efficiency. Public spendingon health is regressive--thatgreater
health expenditures are made in richer provinces. A more instructive fmding is that
incomeand vital health indicatorsappearto have played a considerablerole in budgeting
decisions. Poor provinces in which infant mortality rates are higher than the national
average, receivedmeagerbudgetaryresourcesand spent less. Contrary to pro-healthrisk
targeting, poor and high infant mortality provinces--Hunan,Sichuan, and Guangxi--
received considerablylow health budgetaryoutlays. Chief outliers are Tianjin, Beijing,
and Shanghai,which have high per capita income, a low infant mortality rate, and high
per capita spending. Progressivityhas been reduced in the late 1990s and their efficacy
appearsto be quite modest in poorer provinces.These variationsare associatedprimarily
with differences in per capita household expenditures as well as in government
expenditures,imbalanceddistributionof medicalpersonnel, the lack of health facilities,
or both by province.

15.    RelativeEffects of Expenditures. It is importantto assess if the improvementin
health status is due to specific government policies or, as is also possible, it is due
mainly to gains in overall income. To disentanglethe relative impacts of income and
                                             v


goverunental policies on health status, regressions are used. Overall, the results clearly
show that government interventions in immunization, sanitation and water supply have
had profound impact on reducing infant mortality and gains in life expectancy of the
population, reflecting the efficacy of public health interventions rather than health care.

                                C. Strategic Directions

16.     Although dramatic economic growth and several fiscal reforms have generated
more govermnent revenues in the past two years, repayment of public debt and financing
the budget deficit may pose hard choices for financing the development of social sector.
As such, the government's priorities are to enhance financing and improve the equity,
efficiency, and quality of educational and health services. The most important
interventions would be to increase the aggregate level of spending by local governments
that have surplus budgets and face little intersectoral competition, and improve
expenditure targetting by central and local governments that face budgetary pressures.
These two measures are crucial, because provinces compete with each other to obtain a
greater share of investment outlays; often, the location of projects rather than their
importance becomes the more critical criterion for central government allocation
decisions. Greater complementarity should be also fostered between the public and
private sectors, and assistance for high-priority needs should be sought from new or
existing donors whose commitments have been modest in the past.

Education

17.    Economic rates of return reinforce the justification for more public spending in
primary and secondary education mainly to improve its capacity to enroll and retain poor
students in economically disadvantaged areas, and interventions in higher education
mainly for expanding specific fields of study in order to foster an externally efficient
education system. To bring about growth in desired directions, the following strategies
deserve urgent attention.

18.    Pirmarveducation. The weak capacity to retain students in primary and
secondary levels in disadvantaged areas creates regressive selectivity in the sector. To
offset this trend improved targetting is necessary by increasing public spending through
per student payments to schools serving poorer children and by reducing it in schools
serving better-off areas. To reduce the private costs of education for the poorest chidren,
complementary measures should include pricing reform such as elimination of tuition and
fees, distribution of free textbooks, and allocation of money to compensate for
opportunity cost. Another step involves food for education in primary and secondary
schools for truly indigent by international standards.


19.    Secondarv Education. The dynamics of China's new labor market should
increase demand for secondary school graduates, necessiating some forms of
governmental interventions. Although the present policy of 9-year compulsory education
                                             vi

that extends the junior secondary level does enhance coverage, yet additional intervention
 via financial incentives for school attendance covering direct and foregone earnings
 particularly in less developed areas would be needed with the caveat that this incentive
 could be reversed when further justification no longer applies. The case for such
 intervention is consistent with the empirical findings of transition rates from primary to
junior secondary education, and does not nullify the case for more narrowly targetted
 subsidy and financial incentives such as scholarships for poor but meritorious students
 in senior secondary education.

20.      Higher Education. Although social rate of return from higher education is low,
yet future demand in China's expanding labor market argues for finding ways to enable
universities to expand current enrollments. Once the demand for secondary school
graduates rises, which is a likely prospect in itself, this pool will be enlarged from which
only an expanded tertiary education can absorb the intake. In particular, engineering,
business administration, agriculture, and medicine are becoming priority fields of study.
But diversifying the content, type and level of courses and attracting students at public
institutions will require improving internal inefficiency, greater but selective user fees
and student finance schemes, promoting distance education, and raising quality.

        ImDroviWg  Internal Efficiency. Two important measures could alleviate the
burden of high costs and make this subsector internally more efficient. First, internal
structure of universities should be reconfigured by eliminating duplication of programs,
streamlining administration, and increasing teachers' workload in tune with international
standard. Second, the several currently underutilized carnpuses and facilities should be
merged, creating economies of scale. Rather than establishing separate facilities and
compete for each other, all line agencies should work together to improve administrative
arrangements that would raise modest pupil-faculty ratios.

        Increasing User fees selectively. Given the high private rate of returns to higher
education, government subsidy may limit private spending that students and their parents
want to make. Indeed, there are two indications that the private sector is able and willing
to pay: the existence of excess demand for education i.e more applicants than available
places, and donations. Rising enrollments reflect that increasing user fees selectively
would not reduce the demand for higher education but instead allow government to
mobilize private contributions for expanding access and thus improving equity. The
government should further reduce subsidies for the cost of living that students would
incur whether or not they are enrolled. These non-instructional subsidies are still high
relative to many Asian comparators and should not be a substitute for private spending
that could be made anyway by many households. Such st:epcan alleviate crowding-out
problem, freeing-up government resources for expanding enrollments in other levels of
education.

        Improving Student Finance Schemes. High subsidies and low user fees in public
higher education tantamount to a blanket subsidy that can be redressed by reconfiguring
the existing mechanisms for student finance. It is clear that priority fields of study will
                                          vii

be offered mostly in public universities.But in future private (Minban)universitieswill
continueto grow not only in responseto increasingdemandbut also as a resultof limited
places in public universities.In such situation,there will be more pressureto expandthe
access to public universities. But given the high unit cost in public relative to private
universities,such expansionmay cause further inefficiencyand greater social selectivity
because the better-off studentswill get more opportunitieswho would otherwise have
gone to private institutions.

       Scholarship scheme targetted to poor and qualified students can avoid such
perverse outcomes. Although such scheme is already in place in China, public
universitiesare still guaranteeda captiveclienteleand thusinsulatedfrom the competitive
                                                                   of
pressures of having to justify the quality and cost-effectiveness their educational
servicesto users and communities.So scholarshipsshouldbe untied and studentsshould
be allowed to select institution of their choice, public or private, thus spurring
competitionand promotingefficiency.The studentloan schemesare also availablein 37
public universities. However, to avoid re-emergenceof blanket subsidy and higher
default rates in future, loans shouldbe given only to studentswho do not qualify for
scholarshipsbut who cannot fund studiesfrom their own sources. Repaymentprograms
should be made into income-contingent        payment systems where loans are repaid
according to the post-graduateearningsof the borrower.

 21.     EnhancingDistance Education, Qualityand ResourceGeneration. Further
 developmentof distance education would expand coverageand reduce unit costs. But
.- any technical colleges and post secondary institutionsproduce graduates who are
 preoccupied with the need to obtain a formal degree for formal sector jobs, with
 inadequate attentionto necessaryjob skills imparted through education and vocational
 training system. This leads to a decliningquality of higher education.As such, adopting
 high standard of admissions, training and instructions, relaxing strict regulationsfor
 private universities, and providing accreditationon a differentiatedscale of academic
 excellenceare critical. Finally, althoughamplescope exits for cost savingsand revenues
 from the sales of services to SOEs, donations,too much dependenceon these sources
 may be counterproductive.Experience shows that the advantage in these cases may
 diminishbeyond a revenuerecoveryrate of 15 percent, and educationalinstitutionsmay
 turn into profit-makingorganizations.

Health

22.     Achievementsin health sector need to be sustainedand institutionalized  through
renewed politicalcommitmentand provisionof adequatebudgetary support. The sector
has suffered set backs due to financialconstraintsover the past several years. Priorities
are to improveintrasectoralallocationof resources, redeploymedicalpersonnel,contain
costs and augment public and private sector roles in medical insurance, improve
                                            Together, thesemeasureswill providepublic
targetting, and realignuser fees selectively.
health servicesmore effectivelyto the poor and needy withoutaffecting quality.
                                            viii

 23.     Improving Intrasectoral allocation. Disproportionately greater spending on
 curative health services relative to preventive care benefits mostly the insured urban
population at the expense of a vast majority of rural population including the poor.
Although this pattern reflects epidemiological transition leading to a greater demand for
hospital care, funding for disease prevention services, township hospitals, health
promotion services are being squeezed out, leading to a much higher incidence of risk
factors. Further reduction in funding would increase the risk of premature deaths. A
major policy priority should be to raise the aggregate level of spending on preventive
health services and health promotional services that would reduce risk factors of chronic
diseases. And a case for increased levels of spending for these services is particularly
strong in provinces where vigorous intersectoral competition for resources that had
previously shifted resources away from the health sector toward income generating
projects is little, where spending levels on health are below the national mean, and where
budgets are surplus. Other policy priorities are to improve hospital networking, impose
tobacco tax, inform and motivate the population about the benefits of changing their
lifestyle, provide incentives to industries to adopt less hazardous technologies, and
develop family hygiene regimens.

24.     Redevlovine'  Medical Personnel. Rural areas in poor provinces are deprived of
 more qualified medical staff. Besides their impacts on health outcomes, the different
mixes of health personnel has important bearing on cost: efficiency. In provinces with
high morbidity and mortality rates from infectious and communicable diseases,
redeploying the personnel mix toward more preventive health services staff like nurses
and midwives can generate further cost savings. To attract and retain more qualified staff
in poorer provinces, the government should provide suchL    incentives as housing, bonus,
education for dependents, and better posting after sexving in rural areas. Another
powerful policy instrument is to upgrade training and retraining for less qualified health
personnel and then substitute them for more qualified staff where feasible. In this respect,
the current rigid controls on remuneration should be made more amenable to local needs.

25.     Cost efficiency in Health Insurance. Two important problems with the health
insurance system have emerged in recent years: rising costs and uneven coverage. The
government should pursue four options to mitigate these problems: centralize the
management of health insurance, standardize co-payment rate, pool medical costs, and
introduce health card system in rural areas. Implementing the first and second options
would facilitate implementing the third, all of which would make the system operationally
more efficient. The fourth option would promote equity. Another policy priority would
be to make hospitals liable for all overspending on the capitation payments, forcing them
to invest and create trust funds with profits and unused capitation money, thus gradually
eliminating the 50 percent government contribution to such overspending.

26.    Introduction of rural health card system, as done in Thailand since 1983, would
mitigate rural risk-coverage problems. Under this scheme households can purchase an
annual health card, generally priced at less than the annual average household
expenditure. It entitles the purchaser to a fixed number of treatments and an unlimited
                                           ix

number of visits for preventiveservicessuch as maternal and child health care. Village-
level drug revolving funds are another scheme for local health insurance coverage.
Finally, people may need supplementaryinsurance and private sector can play an
important role to cover additionalhealth services. The government should encourage
private insurancewhich would promotegreater diversity and competitionin the supply
of health services. But specific legislationshouldbe enactedto guard against abuse and
forbid exclusionof pre-existingmedicalconditions.Sincemandatory social insuranceis
already in place implementing  safeguardsover emergingprivate insuranceshouldnot be
too difficult.

27.     Better Targetingof resources. At present,the insuredconsumethe bulk of health
servicesin China, therebyshiftingservicesaway from the vast majority of uninsuredand
indigentpatients. Compounding inequityis the wide disparity in the distributionof
                                this
provincialgovernmenthealth expendituresespeciallyin the past few years. Moreover,
over the past, richer provinces have had much greater increase in government health
spendingthan in poorer ones, implyingthat even with equitablegrowth in incomeacross
provinces--anunlikelyprospect--thedistributionof provincialgovernmentexpenditures
on health will becomemore unequal.If economicgrowth in the provinces alone is less
likely to reduce the imbalancein provincialhealth expenditures,a strong case can be
made for the central governmentto target an increasingshare of central government
resourcesto the poorest provinces.

28.    Realigning UserFees. Bettertargettingof public health expenditureto improve
accessof the poor to health facilitieswould dependon two policy.instruments:reducing
user costs associatedwith treatment and drugs, and improvingthe quality of care for
those who actuallygain access.A more effectivemechanism    shouldbe devisedto protect
the poor from futureuser fee increase.The recenthouseholdexpendituredata showthat
the burden of paying for service charge and drugs is higher among the poor than the
better-off. Sincethis unevenburdenpreventsthe poor from using public healthfacilities,
then improving access for them may require selective price reductions for essential
services;the pooresthard-to-reach  group shouldbe exemptfrom paying fees, whichmay
be compensatedfor by increasesin public spending, and other fees selectively.This
needs an appropriatepricing policy that differentiatesprices by the income of users.

29.     Finally, improvingthe qualityof public healthserviceswould enhanceutilization.
One indicator of the quality gap is the variation across income groups in the amount of
householdexpenditureon public health, but another indicator, availablefrom the 1993
NationalHealthSurvey (NHS), is that householdsdo not use public health facilitiesdue
to the lack of beds and the absence of proper services and treatment. Interprovincial
variation is even stark. Another indicatoris that the poor are less likely than the better-
off to receive treatmentby qualifiedand trained doctors when they are sick. Spending
on medical equipment, supplies, drugs, and physical infrastructurein townshiphealth
centers and the redeploymentof trained personnel in needy areas would considerably
improve the qualityof servicesavailableto the poor.
                                            x

Social Security

30.     RetirementPension. As the populationcontinuesto age, as more younger people
move to private sector, and as unemploymentrises due to restructuring of SOEs, the
number of active people in public sector will decline, thus reducing contributionsand
eventuallyimposinga greater fiscal burden on governmentexpendituresfor pension. As
such, Chinese social security system particularlypension and unemploymentschemes
shouldbe reformedif better financialsolvencyand a more equitableand efficient system
are to be achieved. Given its current economic reforms, political realities and culture,
China shouldundertake a step-wiseapproachto reformingits pension system.

31.    Medium-term reforms. Prioritiesare to improve benefit formula, and unify all
pension schemes. Restructuringbenefit formula includesgradual increase of retirement
age, introduction of an alternativelinear accrual factor, actualizedlifetime income for
pension in lieu of pensions on the basis of the last basic wage, pension indexation,and
                        and
ceilingson contributions benefits.Unifiedpensionsystem-witha specializedagency--
would be most desirablein order to promoteportability, foster efficiency, improvefund
management,and prevent possible evasion.

32. Long-term reforms. Given emerging demographic pressure, the medium-term
                                                    of
program will not be able to ensurethe sustainability the system in the long run. Thus,
a bold reform, incorporatinga multi-tier(pillar) system, will be necessary to avoid any
painful transition later. The first pillar, the mandatory public one, financed by
government revenues and contributions of beneficiaries, should have flat rate minimum
pension, independent of salary level but proportional to actual length of service, and
earnings-relatedbenefits based on lifetime and linear accrual rate. As the market
matures, second pillar i.e mandatorysavingsbased on individualcapitalizationaccounts
and third pillar i.e voluntary savings, shouldbe introduced.

33.    Unemplovment Benefits. While unemploymenat       insurance in most OECD
countries provides job search incentives, the system in China is purported first to
encourage labor mobility and second to facilitate adjustments of labor supply in SOEs,
mainly by reducing surplus labor. The schemehas fostered labor mobilitybut done little
to adjust SOE labor supply. Because a significant proportion of the surplus labor
represents the old and unskilled, and the scheme does not provide any incentives for
training and retraining. So if they are laid off the prospectof being re-employedis slim,
making the schemevirtually irrelevantfor them. To remedythis, the followingmeasures
are crucial: training and retrainingtailoredspecificallyto upgradeskills of surplus labor;
subsidy to employers if they hire long-term unemployed, to compensate them for having
to upgrade their skills or to retain workers they may otherwise have displaced;a lump
sum severance payment to long-time workers in lieu of monthly unemployment benefits;
and early retirement, but this should be done selectively, so as not to place undue
pressure on the retirement pension.
                                   INTRODUCTION


1.01 Education, health, and social security are the main components of the social
service sector in China. Together, they account for more than 25 percent of total
government expenditure and more than two thirds of the total government budgets for
social services. The government has been the dominant provider of these three social
services; in fact, all of the government's development plans have explicitly recognized
the importance of these services to human resources development, and have called for
more investment in the sector. Today the social service sector is increasingly competing
with other sectors for budgetary resources; the competition will become more acute in
the future, since the demand for these services is expected to grow continuously as age-
pattern of fertility and mortality change. Against this background, it is essential that a
greater balance between the equity and efficiency of public resources use be pursued
effectively.

 1.02 Equity is primary concern because the public sector plays such prominent role in
providing social services, and it is important to determine how the benefits of public
spending on such services are distributed. Many public services are publicly provided
private goods, and their utilization is the main determinant of the benefits derived from
them. Improving access to and the consumption of social services has received urgent
attention in the government's overall strategy to redress poverty and enhance social
welfare. But there is some concern that improvements in several social indicators have
been slower than expected. Poverty constrains the activities of the poor, causing them
to forego potentially profitable human capital investment, especially in education and
health. In this respect, assessing whether increases in the income of the poor have been
matched commensurate increases in access to and the utilization of education and health
services is a policy priority.

1.03 Efficiency is also an important issue, because any savings from distributing
services to the poor more effectively can be used to improve the quality of education and
health services. A recurring question in China is whether desired outcomes in education
and health have been achieved at minimum costs, and how past outcomes and
achievements can be improved. Thus, the objective of this study, inter-alia, is to evaluate
the extent to which public financing for social services has been a promising instrument
for achieving growth in a way that also helps alleviate poverty.
                                             2



                                        I. EDUCATION


   Overall Development

   1.04 For more than three decades, educationaldevelopmentin Chiinahas been quite
   impressive. The adult literacy rate, an outcome of past cumulative investment in
   education, grew from 60 percent in 1960 to 74 percent in 1992. By the early 1960s,
   China had also achievedvirtuallyuniversalenrollmentin primary education,a coverage
   rate that still outpacesthe rates found in severalAsiancounterparts.But, China does lag
   behind other countriesregardingother dimensions educational
                                                       of           progress. For instance,
   its rate of enrollmentin secondaryand particularlyin higher educationis belowboth the
   averageamonganlAsiancountriesand the rates found in low-incomecountries,let alone
   its close comparators. Qualitativeindicators, such as average grade attainmentof the
   currentschool-age    population cohortsurvivalrate, althoughshowingsteadyprogress,
                                   and
   are modest when compared with those of Malaysia, Korea, Thailand and Sri Lanka
   (Table 1.1). For instance,averagegrade attainmentis 5.6 years in China, which is close
   only to Indiabut significantly lower than its closecomparators.Likewise,cohort survival
   rate is 68 percent, quite lower than its close comparators and similar only to the
   Philippines.

   1.05 Yet, when incomelevel is accountedfor, the level of educationalattainmentin
   China is moderatelyhigh relativeto other countries(Figure 1.1). Part of the difference
   can be explainedby how the three levelsof education--primary,secondaryand tertiary--
   are affected by budgetary allocations, private-public sector demand, and level of
   economicdevelopment China. Thus, thepace of improvements qualitativeindicators
                         in                                      in
   in China would depend on the choice of investmentsand sectoral policies, especially
   those affecting the developmentof primary and secondaryeducation in needy areas.
   There may be little scope for improving quantity but there is room for quality
   enhancement in schooling, which may infact generate more significant gains in
   educationalattainment.Evidencefrom severaldevelopingcountriessuggestthat the rate
   of return to improvingthe qualityof schoolingis high comparedto improvingquantity
   of schooling of a given quality '.


   Education Cost and Financing

   1.06 Trends, Structure and Composition.             At least three important features
   characterize the composition of public educational expenditure over the past decade


    I For empirical evidence in a number of countries see, Jere Behrman and Nancy
Birdsall,"The Quality of Schooling: Quantity Alone is Misleading" American Economic Review,
1983. Jere behrman, Donald Ross and Richard Sabot, "Improving the Quality Versus Increasing
Quantity of Schooling" 1992, Colby College (Mimeo), Waterville, MA
                                                        Figure 1.1




                                           CHINA
                  Relationship Between Grade Attendance Rate and Income:1993


                   (year)
     Gradeattendance
     12

     11

     10                      Sri Lanka
                                    Philippines                                Malaysia
      9_


                                    Indone-sia
      7 _etnamrr                                                    Thailand
              -myanmar

      6

      5
                                           Papua New Guinea
      4           ,en%tpdesh
                    Neal Pakistan

     3        -


     2            Bhutan


     0
          0        200     400      600   800    1000                   2000         3000
                                                        GNPper capita




Source: Table 1.1
                                         3



            Table 1.1: Basic Indicatorsof OverallEducationalDevelopment,1993


                        Public
                     Expenditure
                      in Education                                     Cohort
                  As a percentageof:  GrossEnrollmentRatio   Grade    Survival
                            Public          (Percent)       Attendance RateY
                  GNP      Spending Primary Secondary Higher (year) (percent)

Bangladesh         1.9       11.3       59        18        5       4.0        25
Bhutan             3.8       8.6         31        5      0.1       1.6        16
China              2.0       12.1       104       51        2       5.6        68
India              3.3       13.9        99       41       10       5.2        39
Indonesia          3.6       14.8       119       48        9      7.5         61
Korea              3.4       16.7       105       87       38      11.8        97
Lao PDR            9.6       9.5        110       27        2      4.9         41
Malaysia           7.4       18.5       102       58        7       9.5        97
Myanmar            2.1       11.0       103       24        6       6.8        NA
Nepal              1.6       10.4       86        31        5       3.7        35
Pakistan           2.8
Philippines        1.8       13.2       101       70       28       9.4        67
Papua New
 Guinea            7.0       15.3       72        14        3      4.5         67
Sri Lanka          3.2       10.7       107       71        6      10.0        87
Thailand           3.9       16.6        87       28       16      7.2         83
Viet Nam           1.3       5.6        105       40        5      7.1         60

Mean               3.6       12.6       93.1      41       10       6.4        60.2

 a. End of primary level.
 Source: World Bank Data Base, UNESCO.
                                                  4



    (Table 1.2)   2   First, recurrent expenditures have increased in real terms; the bulk going
    towards such "autonomous" categories as teachers' salaries and gradually meager
    proportion is being used for instructional inputs and supplies of basic materials and
    equipment for schools. Increase in expenditure on salaries was a part of conscious policy
    to improve retention rates among teachers in various provinces. Many teachers were
    recently being attracted into private sector employment as salaries in other sectors were
    being driven up progressively with the recent surge in economic growth. In 1993, the
    government envisioned to revamp the way the government hires, classifies job
    description, structures pay and awards bonuses. Accordingly, teachers' salaries rose
    dramatically from the current twelfth to fifth grade 3.

    1.07 Of total, expenditures on non-personnel instructional materials fell from 30
   percent in 1990 to 23 percent in 1993 (Annex 1 Table 4) . By devoting its expenditures
   largely to teachers' salaries in order to retain a pool of skilled personnel, the government
   has been forced to find alternative mechanisms for financing educational supplies and
   materials. As such it allows all primary and secondaiy schools to charge fees and
   universities to charge tuition and fees, and to engage in revenue-generating activities.
   Increasingly, revenues from students' fee, tuition in post-basic education, and
   contributions from collectives are used to finance purchase of books, supplies and
   teaching aids. This user-fee policy in education is designed to simultaneously minimize
   the crowding-out effect and maximize the equity in the allocation of government
   education spending. The additional resources mobilized from private spending thus
   helping the govermnent to provide greater coverage at all levels of education.


   1.08 Second, real capital expenditures have been volatile and even fell on average
   between 1985 and 1991. However, the percentage of cuts in capital expenditures on
   education were smaller than the average cuts in capital expenditures overall for the public
   sector. This pattern conforms to the general finding that spending on education tends not



    2 Unless otherwise noted, public expenditure includes government budgetary expenditures
excluding spending by state owned enterprises (SOE). (Technical Annex A)

    I The personnel management in the government, including SOE, is inflexible and somewhat
complex, rewarding longevity rather than performance, with a system of hiring employees at low
or modest pay, and rewarding them near retirement with higher pay and pension.
                                               5



                    Table 1.2: Public Expenditure on Education, 1983-93
                                       (billion yuan)


                        Current prices                   1987 Prices           As a percentage of
                                                                                      Public
    Year          Recurrent Capital Total      Recurrent Capital Total         GNP expenditure

   1983           10.38      2.41    12.79       8.03       1.87        9.90   2.2       9.9
   1984           11.66      3.16    14.82       9.49       2.57       12.06   2.1       9.6
   1985           14.02      4.38    18.40      12.66       3.96       16.62   2.1      10.0
   1986           22.47      4.00    26.47      21.26       3.78       25.04   2.7      11.4
   1987           24.21      3.56    27.77      24.21       3.56       27.77   2.5      11.3
   1988           24.58      3.43    28.01      21.95       3.06       25.01   2.0      10.3
   1989           24.68      3.64    28.32      20.06       2.96       23.02   1.8       9.3
   1990           27.79      3.41    31.19      21.05       2.58       23.63   1.8       9.0
   1991           30.17      3.47    33.64      21.44       2.47       23.91   1.7       8.8
   1992           36.19      4.00    40.19      24.58       2.72       27.30   1.7       9.2
   1993           55.11      4.48    59.60      38.27       3.11       41.39   2.0      12.1

   Average     25.57         3.63    29.20      20.27       2.97       23.24   2.0      10.1
   Annual average
   growth       12.4        10.6     12.2          4.7      3.0         4.5
   (%) a/

       Note: Includes central and local govermnent expenditures.
       a. OLSQ growth rate.
       Source: State Education Commission and Staff Estimates.


   to suffer disproportionately during budget crisis, due partly to the difficulty of impeding
   school building construction, equipment purchase, reducing the number of teachers
   employed, and their salary 4.

   1.09 And, third, the share of educational expenditure in GNP and government
   expenditure is low compared with neighboring Asian countries that have a similar income
   per capita and economic structure. Historically, China has committed an average of about


   4  Contrary to conventional wisdom, here, and studies elsewhere by Pinstrup-Anderson
(1990), Jarmitto and Stewart (1989) suggest that on average social sector expenditures were the
most sheltered of other important sectors in countries pursuing economic reforms.
                                                6



    2 to 2.5 percent of GNP on education, representing 13 to 14 percent of total public
    expenditure. But these shares are relatively low compared to Asian neighbors that have
    a similar income per capita and economic structure. Indeed, when Figure 1.2 plots the
    relationship between per capita income and public education spending for 16 Asian
    countries, China turns out to be a negative outlier predicted for its income level in 1993.
    This is partly attributable to comparatively low per-pupil expenditures in primary and
    secondary public education, and partly ascribed to either greater efficiency of a system
    that depends on private sector or underinvestment. But, the rates of return to educational
    investments show that the degree of underinvestment is not significant in China.5

     1.10 The distribution of public spending by level of education is shown in Table 1.3.
    Of total educational expenditures, 36 percent is spent on primary education, followed by
    34 percent in secondary education and the balance is accounted for by higher education.
    This distribution of public spending on education is thus equitable, which is noteworthy
    when compared with the patterns observed in many neighboring Asian countries, where
    higher education receives the bulk of subsidies. Again, the unit cost of public education
    in China--measured as per-pupil expenditure as a percentage of GNP per capita--is low.
   In primary education, this is close to unit cost in Nepal, but substantially lower than in
   Malaysia, Korea and Thailand. The unit cost in secondary education in China is lower
   than those in Bangladesh and Papua New Guinea. Similarly unit cost in higher education
   is lower than these two countries. The relatively low uInitcost in primary education is
   due to high pupil-teacher ratios and the low capital costs of education. The high unit cost
   in tertiary education in China, when compared with those, say, in the Philippines and
   Thailand, is attributable to very low enrollments and higher expenditures on training and
   research in public institutions, whereas a significant proportion of spending on higher
   education is borne by the private sector and open universities in the Philippines and
   Thailand, respectively.

   1.11 Sources of Financing Educational Expenditures. The government has a
   longstanding policy of using local revenue and community resources to finance primary
   and secondary education, so as to free-up more central government funds for other
   purposes particularly for higher education and research. The bulk of the central
   government appropriations go to university programs run by the State education
   Committee and to other specialized schools. Of total spending on primary and secondary




    5
    Also see for example, Shaikh I. Hossain and George Psacharopolous, "The Profitability of
School Investments In An Educationally Advanced Cointry", International Journal of
Educational Development, vol. 14, 1994.
                                                  Figure 1.2




                      Relationship Between Education Expenditure and
                        GNP per Capita in 16 Asian Countries (1993)



 Educationexpenditure of GNP)
                     (%
 10
                Lacs
  9_


                                                0 Malaysia
  7 _PNG

  6



  4                  Indn
             India          rna                                                 SouO Korea 0
  3                  >rlaS          0 Thadano

  2 -
          *     Paibstan
       Ne9aJ
  1    -..     Vet Nam                I
      o                     1,000   2.000       3.000        4,000    5,000   6,000        7,000
                                                      GNPper capita



Source: Table 1.1
                                             7



Table 1.3: Unit Recurrent Cost in Education, 1990
(as a percentage of GNP per capita)


  Country                        Primary                 Secondary                 Higher
 Bangladesh                       6 (40)                  30 (47)                 265 (12)
 China                            9 (38)                  29 (34)                 244 (28)
 India                            6 (41)                  14 (29)                 231 (26)
 Indonesia                        13 (60)                 23 (28)                  91 (12)
 Korea                            17 (54)                 23 (34)                  71 (8)
 Malaysia                        14 (41)                  21 (39)                 190 (19)
 Nepal                            9 (52)                  14 (17)                 249 (30)
 Papua New Guinea                29 (46)                  65 (19)                 1050 (34)
 Philippines                      6 (60)                   7 (37)                  52 (3)
 Sri Lanka                        6 (54)                   9 (32)                  83 (13)
 Thailand                        16 (56)                  15 (20)                  40 (23)
 Viet Nam                         4 (49)                   7 (31)                 133 (20)

  Note: Figures in parentheses are the shares in total public educational expenditure.
  Source: World Bank data files



education, more han 75 percent comes from government budgets and education levies,
and the balance is accounted for by private sources-local contributions from parents
communities, tuition and fees, school enterprises, and various school raised funds.
Educational levies comprise 1 percent of value added taxes in urban municipalities, 1
percent of sales taxes, and 1 percent from consumption taxes. The funds raised from
these levies are also used to finance teacher salaries and school building construction, but
they are not included in the government budget. School raised funds comprise tax-exempt
revenues from school factories; these revenues are normally used to improve the living
conditions of teachers. (Annex 1, Table 3)

1.12 The government subsidy in Chinese higher education limits private spending that
students want to make. One indication of private willingness to pay is the existence of
excess demand for higher education, i.e, more applicants than available places. Imposing
                                             8



selective tuition and fees would not reduce the demand for education but would instead
allow government to mobilize private contributions for expanding access and thus
improving equity. Evidence from neighboring Asian countries show that higher user fees
tend to expand the coverage of educational system (Psacharopolous 1992). As the costs
of education rise with the level of education, the absence of fees at the higher levels
invariably creates an unequal distribution of public education expenditures. By way of
illustration, Korea, the Philippines and Thailand have free primary education, but the
governments in these countries recover 20 to 30 percent of recurrent costs at the tertiary
level from user fees. The additional resources mobilized from private spending enable
these countries to provide greater coverage at all levels of education.


Equity Aspects of Education Expenditures

1.13 Overall assessments about the extent to which educational expenditures are
equitable are based on time-series data on the enrollment structure and the patterns of
public spending in education. Together, they show whether the educational system in
China is structurally equitable or not. But, the structural equity alone does not necessarily
imply that people from all socio-economic backgrounds and from all geographic regions
have equal access to schooling. Therefore, accessibility to education and educational
achievements by different income groups and across all provinces are examined on the
basis of both time-series and household-level data.

1.14 (a) Structural equity. At least three important characteristics have shaped a
structurally equitable educational system in China; a flatter enrollment pyramid--nearly
universal entry into basic education and lower enrollments in higher education; and, more
instructive, a pattern of government spending by educational level that conforms to the
priorities of the govermment. From 1982 to 1993, more than 60 percent of public
educational spending was allocated to basic education--primary and junior secondary
levels--so that the poor can get access to it. Another feature, although not yet very
significant, is the growing private enrollments in higher education, suggesting that the
proportion of fees being used to finance post-primary levels of education is increasing.
The data in Table 1.4 supports these facts. Gross enrollment rates at the primary level
had been higher overtime, but the enrollment growth compared to other levels is slow
because the coverage at this level was already extensive for a long time. Slower growth
in enrollment in primary education also stems from declines in birth rates since 1970s
and a reduction in dropout and repetation rates that reduced the proportion of students
in the primary school-age range. And, public spending by level of education clearly
shows that more resources are devoted to primary and secondary education than to higher
education.
                                             9



Table 1.4: Structureof Enrollmentand PublicExpenditureon Education

                                                       1980        1985          1993
 Gross enrollment ratio (percentage)
 Primary                                               148.0       126.0        104.0
     Secondary                                          32.0        39.0         48.0
     Higher                                              1.1         1.6          2.0
 Public expenditure on education

 As a percentage of GNP
 Primary                                              0.95        0.79         0.75
 Secondary                                            1.00        0.79         0.74
 Higher                                               0.55        0.58         0.51
 As a percentage of public expenditure
 Primary                                              3.4         3.6          4.7
 Secondary                                            3.6         3.5          4.3
 Higher                                               1.9         2.8          3.1
 As a percentage of education expenditure
 Primary                                              38          36           40
 Secondary                                            40          36           39
 Higher                                               22          27           21
 Percentage of cumulative public spending                         29           27
 received by best educated decile a/


a! Staff estimates
Source: State Education Commission and SSB.



1.15 A single summary indicator of structural equity is provided by the share of
cumulative public expenditure received by the best educated decile in a generation.
Overall, 10 percent get 26 percent of cohort resources and 90 percent receives the
balance, reflecting an egalitarian distribution of public subsidy in education. This is also
well below the regional Asian average of 39 percent. Normally, the best educated decile
in developing countries receives more than its population share of the cumulative public
expenditure on education. The cutoff at 10 percent is an arbitrary one, but it provides an
                                                      10




   Table 1.5: Gross Enrollment Rates by Quintile of Provincial Per Capita Income
   (Yuan)

                                                Age group of children
                       7-11years old                  12-14years old                  14 and above
     Quintile   Male     Female    Total      Male         Female      Total   Male     Female       Total
     I          90.2      83.2         86.7    28.5         24.0       26.2    3.8        1.9        2.8
     I          92.0      90.3         91.1    38.1         33.0       35.5    5.2        2.3        3.7
     III        93.1      91.0         92.0    39.2         37.1       76.3    6.8        3.0        4.9
     IV         95.1      94.8         94.9    66.3         71.4       68.8    7.9        3.3        5.6
     V          97.6      97.2         97.4    72.9         78.6       75.8    9.2        3.8        6.5
     Rural      92.3      87.6         90.1    34.8         31.0       32.9
     Urban      96.8      96.6         96.7    70.2         75.0       72.5

    Source: Staff estimatesfrom SSBand StateEducationCommissiondata.



                                                                        6
   indicator of the concentration of public resources on education. The relative extent of
   structural equity is captured by a Lorenze Curve and the corresponding gini concentration
   ratio (figure 1.3). The Lorenze curve for China is close to the line of equality and the
   gini coefficient is quite low (0.39). It is useful to note that the gini coefficient is smaller
   than it was in 1985 (0.43), reflecting that the distribution of public spending on education
   has become much more equitable in the past seven years.

   1.16 This trend owes much to the increasing enrollments in private universities and
   introduction of user fees in public higher education. However, the differences in the rates
   of cost recovery and unit costs within higher education are not captured here. But, it is
   very unlikely that a more sophisticated calculation would reverse the finding that Chinese
   system of education is fairly egalitarian in its global structure of access and public
   expenditure.


     6 Cumulative public spending on education refers to the total amount of public resources
appropriated to a generation of people passing through their schooling years. It captures the
distribution of cumulative benefits (education) among people with different school careers, rather
than the distribution of single period benefits at a given level of education. For details on
calculation, see Alain Mingat and Jee-Peng Tan,"On Equity in Education Again: An
International Comparison" Journal of Human Resources, no 2, 1985
                                           11



 1.17 Yet, the structural equity has not translated into equitable educational benefits
acrossprovinces. Whilethe benefits of primary educationare relativelytargeted to poor
provinces,accessibilityto post-primaryeducationshowsmarkedinterprovincialvariance.
When provinces are ranked by per capita income from poor to rich, the avowed policy
of the government to ensure an egalitarian distribution of educational expenditure
increasinglyvanishes(Figure 1.4). Per-pupilexpendituresare disproportionately   higher
in rich provinces comparedto those in poor provinces. The richest quintileof provinces
Shanghai,Beijing, Tianjin, Guangdongand Liaoningspent almost 1.5 times more than
the poorest provinces such as Guizhou,Guangxi, or Henan.

  1.18 (b) Access to education by provinces. The foregoing is substantiatedby the
 differencesin enrollmentrates for primary, secondaryand higher education by quintile
 of provinces in Table 1.5. The attendancerates at the primary level are, by and large,
 similarirrespectiveof poor or rich area, confrming the earlier observationfrom national
 data of universal entry at this level of education. But: the enrollment-rate disparity
 between rich and poor provinces grows as studentsmove on to secondaryand higher
 levels. Such disparity is particularlyacute amongthe 12-14and 14 and older age groups-
 -correspondingto post-primaryeducation--than    among the 7-11 age group. Urban-rural
                                                                to
differentials are also pronounced,and this can be attribuLtable differences in income
 and quality. Primary education that correspondsto age range 7-11 is compulsory in
China, but attendance is not universal in rural area particularly among the females.
Urban children and those from high-incomeprovinces are much more likely to attend
 secondaryschool(about2.2 times) comparedto childrenfrom rural area and low-income
provinces. In the absence of compulsion in the subsequent age groups the gap in
attendancerates widens steadily. Higher educationis mostly received by householdsin
richer provinces. What has happenedto the access to education by provinces in other
years? Similar household surveysare not available, but enrollment rates by provinces,
ranked by per capita income from poor to rich, indicatethat the importance of income
as a correlate of school enrollmenthas not changed, the richer the province the higher
the enrollmentrate in post-primaryeducation.

1.19 (c) Social selectivity. Chineseeducationalsystemsuffersfrom social selectivity.
Disparities in drop-out rates at the secondaryeducation level are quite pronounced .
Male-female differences in primary education are modest but very significant in
secondaryeducation. Drop-outrates also widely vary according to geographiclocation
and level of education: high in poor provincesand secondaryeducation, and vice versa.
In secondaryeducation, in rural areas, more females tend to drop out of school than do
males (Table 1.6). The reasonis that due to obviouseconomicdifficulties, females from
poor families leave the schoolsearlier to start workingat home or to help their families.
Inadequatelearning environment poorerprovincesis anotherreason for early exit from
                                 in
school system.
                                            12



Table 1. 6 Drop out Rates by Area and by Gender, 1992


                           Rural                               Urban
 Grade           Male     Female     Total          Male     Female      Total
 Primary
 level
   1             37.8      31.9       33.8           21.7     42.1        31.0
   2             21.1      22.7      22.2            8.7       10.5       9.5
   3              17.4     21.4      20.1            43.5      15.8       31.0
   4              13.4     13.5       13.4           17.4     21.1        19.0
   5              9.4       8.6       8.8            8.7       5.3        7.1
   6              1.0       1.9       1.6            0.0       5.3        5.3
 Secondary
 level
   7             29.5      33.4       31.8           23.5     22.5        23.1
   8              55.1     53.7       54.3           47.1     65.0        53.7
   9              15.4     12.9       13.9           29.4      12.5       23.1

Source: SSB: Situation of Children, 1992.



1.20 (c) Household educational expenditures. Table 1.7 shows the distribution of
household expenditure on education in urban areas. The most poignant feature is the
slight progressivity of education spending but not significantly-per capita household
expenditure on educational items rises gradually as individuals move on to upper income
brackets. Since fees are basically uniform in basic education, per capita household
expenditure does not vary by rich or poor people. In higher education, a smaller
proportion of poorest households spend income on tuition and fees compared with those
in the 4th and 5th quintiles. Likewise, poorer households spend less on school
stationaries and uniforms than do their richer counterparts. Higher per capita household
spending on higher education and other categories reflects the fact that wealthy
households prefer or can afford higher education where tuition is higher.
                                            13



Table 1. 7: Per Capita Household Expenditure on Education by Quintile of Income Distribution
                                           1992
                                          (Yuan)

 Item                     I        II      III       IV       V      Expenditure     T-
                                                                      elasticity    ratio
 Textbook               6.26     6.41      6.91     6.75     6.85        1.12       9.06
 Tuition and Fees
   BasicEducation       41.05    43.58    45.79    41.77    43.45        1.08       3.45
  Kindergarten          5.23     7.19     7.12      7.50     7.59        1.11       5.66
  Higher                2.60     2.54     3.77     5.40      7.68       1.23        8.23
 Stationaries           3.15     3.69     4.00     4.15      3.86       0.74        4.55
 Othera                 2.47     2.88     5.30     2.91      3.66       0.86        6.21
 Total                  60.74    66.29    72.89    68.47    72.09        1.27       10.82

  a. Allowance,other expenses.
  Source: SSB: Urban HouseholdSurvey, 1992.


1.21 Variations in the calculated expenditure elasticities of each educational item with
respect to total per capita consumption expenditure partly indicate differences in quality
between schools. Expenditures on all items, except stationaries, rises more than
proportionately with total expenditure level as the elasticiitiesare greater than unity. A
positive association between education spending and consumption is due partly to
lifecycle effects on household expenditures. Normally, older households tend to earn
more and have older children. Their children are more likely to attend private schools,
if available, and pursue higher education for which costs are higher. For rural areas,
detailed breakdown of private expenditure by educational items is not available. But the
1994 SSB figures show that the poor spend less on education in poorer provinces than
those in wealthy provinces.

Outcomes and Efficiency

1.22 This section focuses on outcomes that are internal to the educational system--
specifically, cohort survival rates and transition rates, and outcomes that are external to
the system, such as private and social rates of return to educational investments. Both
sets of outcomes are critical, since together they help determine the overall efficiency of
the educational system. Yet, few efforts are made to assess the impact of public policy
on these outcomes in China. The dearth of such assessments is due in part to the scarcity
                                                          Figure 1.3



                                                   CHINA
                             Distribution of CumulativeEducationSpending: 1993
Percentagecumulative education spending

100




 80




 60




  40


                                                                                               20 Gini 0.395


  20




        0                 10            20       30         40      5o        60       70              80        90        100
                                                      Percentage school age population
                                                                of




                                                            Figure 1.4


                                China: Education:                   Per-pupil        expenditure               by Province
                                                                      1993

1200

                                                                                                       _____               v,
1000             ___._.__



 800              -_                                         _
                                                         ____-_.-                             ----
                                                                                             --


  600 -                                                                               --- ........




  400           . . .............

       00

            I            3          s        7    9       11     13     15      17    19          21   23      25     27        29
            Poor                                                                                                           Rich
                                                         Primary         -      Secondary
                                          14



of appropriate data and in part to the fact that these outcomes were not important to
policymakersunder the command economy.

1.23 (a) Internal efficiency. Conventionally, an education system is internally
efficient if it generatesthe maximumoutput at minimum cost. Outputsare measuredby
such indicatorsas transitionrates and cohort survivalrates or the proportionof students
who start grade 1 and completeeach subsequentcycle, students' cognitiveand technical
skills, scientificknowledge,improved and the like. Ideally, a combinationof individual
survey and time-seriesdata are used for this analysis. However, in the context of the
present study, readily compiled provincial-leveltime-series data are used to provide
results that would be helpful for policy interventionsfor improving system's internal
efficiency.

 1.24 At the aggregate level, educational outcomesvary directly with income. The
transition rates from primary to secondaryand secondaryto higher level of educationare
higher in higher income provincesas compared with those in lower income provinces.
Other eclectic statisticalanalysis strengthenthe conclusionthat public expenditures are
important determinantsof transitionfrom one level to another. A regression is done in
which transition rates are explained on the basis of per capita income, per student
educationalexpenditureand agriculturalwages. Agriculturalwages are used to measure
the opportunitycost of childrens' time and to the extendpossible the expected value of
education.The results clearly suggestthat both incomeand educationalspendingtend to
increase transition rates (Table 1.8). But more importantly, the results indicate that
keepingbudgetaryallocationto primary educationat a high enough level will compensate
for the negativeeffect of low incomeby makingthe likelihoodof transitionto secondary
and higher levels of schoolingmore equal throughoutthe country.

 1.25 The transitionfrom junior to senior secondaryeducationis negativelyassociated
with the transition from primary to secondarylevels. This reflects that some of the less
qualified studentsfrom the high-incomeprovinces where transition rates are high, fail
to survive the next level or perhaps they go out to work with families or in the labor
market due to higher opportunitycosts. This is reflected by the opposite signsof income
and education expenditure variables and higher negative value of agricultural wage
variable in equation (2). Primary education is compulsoryand supply-driven,while at
post-primary levels, the householdchoice for extendinge,ducation be predicated on
                                                                    can
the opportunitycosts to parents of pupils' time. Transition rates to secondaryeducation
are low in poor and more agriculturalprovinces than in rich provinces that are less
agricultural.Thus, equity and better educationaloutcomeswelljustify public expenditure
at primary education.

1.26 It is also important to assessif low transitionrates to secondaryschooling are a
concomitantof poor schoolingconditionsin primary educationor of weak private
                                             15



                 Table 1. 8 Regression Results for Transition Rate, 1992


Dependent
Variable = Transition rate, 1992


                                                   Primary         Junior Secondary
Right hand                                            to                   to
side variables                                    Secondary        Senior Secondary


constant                                           -86.4*               -0.17
                                                    (-9.23)             (-0.77)
log Income per capita                              12.91*               -0.571*
                                                   (2.02)                (-2.49)
log Expenditure per pupil                          10.33*               -0.077
                                                   (3.05)               (-0.56)
log Agricultural income                             1.69                 -0.106
                                                   (0.95)                (-1.74)
Deviation of Grade 6 Achievement Score             0.112*                -NA-
From National Mean                                 (2.55)
   N                                                 29                   29
Adj. R2                                             0.655                0.305


Note: The T-statistics are in parenthesis.
 (*) Significant at 5% level.


demand for secondary education. Empirical evidence in several countries suggest that
schooling conditions in primary education is critical because they create learning
environment for students and motivate them to continue education at post-primary levels.
The regression results in Table 1.8 suggest that the demand for post-primary education
is not weak, as transition rate is positively associated with per capita household income--a
measure of the strength of private demand. Moreover, mean achievement test score--an
indicator of the quality of education--is strongly associated with higher transition rates.
Thus, improving the quality of primary schools--an outcome that can be influenced by
                                              16



the government budgetary decisions--would enhance continuation of education to post-
primary levels.

1.27 Another piece of evidence is the differences in cohort survival rates at the end of
primary school by provinces. Higher cohort survival rates are associated with lower per-
pupil costs in wealthy provinces relative to low-income provinces, reflecting that the rich
provinces have more internally efficient educationail systems than their poorer
counterparts. Students in high-income provinces including Beijing, Shanghai, Tianjin,
Guangdong and Liaoning survive until the end of primary cycle. While in poorer
provinces like Guizhou, Guangxi or Anhui about 20 to 25 percent of students do not
persist to the end of primary cycle. (Figure 1.5)

 1.28 (b) External Efficiency. Concerning external efficiency, the main questions are--
has educational expenditure had any impact on household welfare? Where should
resources be allocated in the sector? The approach here is to compute the economic rates
of return to investments in various levels of education. A summary of economic rates of
return to three levels of education, based on "elaborate" or "complete" method with the
data from 1992 Labor Survey by the Ministry of Labor, is presented in Table 1.9. Both
the private and social returns to each level of education were high in 1992, making
investments in education socially profitable in China.

1.29 The social rates of return is the highest for primary education (14.4%), followed
by secondary education (12.9%), implying that investmenit in these levels of schooling


                 Table 1.9: Rates of Return to Education (Percent), 1992


                     Private                     Social             Index of
                  Rate of Return             Rate of Return       Subsidization
                                                                       (a)


Primary                 19.2                       14.4                 1.33
Secondary               15.4                       12.9                 1.19
Higher                  13.0                       10.3                 1.26


(a) The ratio of private to social rates of return.

Source: Staff calculations from MOL Survey 1992.
                                                         Figure 1.5




                  Cohort          Survival                       Rate          and                    Unit         Cost   1993

   Poor
              Guizhou
              Guangxi
                 An,ui                                                                    ,
                Henan
                 Gansu                                     in,                     EE                  ,,
              Sichuan
               Jiangxi                                                                  .-,,,,,
              Yunnan                                          mum
              Shaanxi                                    -----
                Hunan                                mi-                                          ,
                  Tibet                                                   ___________
               Ningxia                               -
               Qinahai
   Inner     Monaolia
                Hebei                                      inA

                 ijian
                FX                               -               -        _    _    _         _        _    _

      Heilongjiang                                                                                    1
         Shandong
               liang
             Zhe                             -                   -    I
              Lia   , ng
           Guangdon 5                   .-
               Tianjin                                                    _             _,
                Be 'lne                                                                           ,
             Shaniha_i
                           120   100   80            60          40       20        0             5       10 15 20 25 30 35 40
   Rich                                                   _          CSR           M              Unit      Cost

  Note: Provinces are ranked by percapita                                                         income




CSR       = Cohort Survival Rate at the end of primary cycle.
Unit Cost = per-pupil government expenditurein primary education as a
            percentage of provincial GNP per capita.
                                              17



    will promote both equity and efficiencyobjectives. These findings are consistent with
   those elsewherethat additionalyears of primary educationwill increase productivityand
    incomeamong the poor 7. In secondaryand higher education, the social rates of return
   are not very high due partly to higher government subsidy in terms of increased
    operatingbudget and teachers' salaries.This is also reflectedin the indexof subsidization
   particularlyfor higher education,whichindicatesthe degreeto whichcosts are borne out
   by the public sector compared to the private sector. The high private returns to
   secondaryand higher educationimply that investmentin these levels lead to high wage-
   payingjobs, an importantparadigmin a rapidly growing economysuch as China. But,
   high wages may also reflect institutional pressures (minimum wages) rather than
   augmented productivity resulting from human capital formation by these levels of
            8
   education. By the same token, the paper credentialappears to be more important than
   the quality of educationat these levels.

    1.30 Table 1.10reports the basicMincerianearningsfurnctions,      showingthat the overall
   marginal private rate of return to one year of extra schooling,irrespectiveof education
   level, is around 12.5 percent. The marginal returns to education for three levels of
   education are displayedin the last column. These rates are consistentwith the earlier
   findingsas they are close to those based on the "complete"method in Table 1.9. One
   exception to findings found in other developingcountriesis the positive coefficient on
   the squared term for schooling (S2). Thus, contrary to the common pattern of
   diminishing returns to spending in additional years of schooling, there is slightly
   increasing return in China. The results perhaps reflect the changingeconomyin which
   large agriculturalsector is utilizing labor inputs of mostly less educatedworkers, and a
   rapidly growing modem sector that is demandingmore highly educated and qualified
   labor.

   1.31 Overall, three important conclusionsemerge fromrforegoing analysis. First, the
   high social rates of return to primary and secondaryeducation suggestthat the demand
   for labor with these educationallevels still exists. The governmentmay allocate more
   resources to these levels of education in order produce more graduates, but such
   allocations should be selective at the secondarylevel; since costs vary significantlyby
   fields of study, as does the demand. Second,the social rate of return to higher education


   7   See for example, Lockheed, Jamison and Lau, 1980,"Farmer Education and Farm
Efficiency: A Survey in Education and Income" World Bank Staff Working Paper No 402.
Authors found that on average, farmers with four years or more of primary education were at
least 7 percent more productivethan those with no primary schooling.
   8 Labor productivity improvedslowly until 1987. See I.J. Singh,"Industrialproductivity in
SocialistEconomies", 1991.
                                             18



        Table 1.10: Returns to Education (Mincerian Earning Function), 1992


                                                                            Rate of
                                                                            Return
                                   Basic              Dummzy               (Percent)


constant                           10.12               12.03
Schooling(s)                       12.20
Schooling (S2)                      0.005
  squared
Experience (E)                      0.031               0.034
Experience squared (E2 )           -0.0002             -0.0001
No education (Base)                 0.000
 Primary                                                0.621                14.5
 Secondary                                              1.025                13.0
 University                                             1.831                10.2
                                    0.274               0.268

Note: Caveats of ROR Analysis.
Source: Staff calculations from MOL survey 1992.


investments is below the returns to investment in alternative forms of capital, casting
doubt whether further public expenditures in this subsector are justifiable. But reducing
expenditures may jeopardize the already very narrow access to higher education,
particularly among the poor. The attendance rates in higher education are very low,
capturing only one fifth of secondary school graduates. The students who exit from
education system create an oversupply of labor with senior secondary schooling, thereby
affecting their wages and eventually dampening returns on investment. It would be
prudent to target governmental resources to those fields of study where the demand in
labor market is high. Third, the access to higher education should be enhanced by
introducing alternate methods of financing higher education, such as scholarships and
student loans that eventually would help reduce the differences in rates of return.

1.32 Finally, one important issue concerning the social rate of return is that it is based
on monetary earnings (or productivity) and excludes any "externalities" associated with
education. These are benefits of education that go beyond what the individual investor
                                                19



    realizes in terms of increasing earnings, such as effects on others by having more literate
    people around, or better nourished and health children if a mother is educated.
    However, this is not really a critique of the rate of return analysis. It is a possible
    addition to education benefits that are unmeasured, as for any other investment in a
    country.9

     1.33 State Owned Enterprise Expenditures. From 1986 to 1993, educational
    expenditures by state owned enterprises (SOE) rose steadily from 2.33 to 8.54 billion
    Yuan; representing an annual average growth rate of 5.6 percent (Annex 1, Table 5).
    The SOE expenditure on education as a share of GNP renmained     virtually stable at around
    0.023 percent, and relative contribution by the cenitral and local SOEs varied
    considerably in response to economic swings and production capacities. Overall, the rise
    in SOE spending on education is a result of government's avowed policy of promoting
    school enterprise finances throughout China. Better-off SOEs can continue financing
    schools in their localities. If SOEs incur losses; and are umable to finance their schools,
    local governments usually try to help them. Indeed, SOEs are not only generating
    additional resources for education their involvement has created a stronger sense of
    commitment to the overall educational enterprise.




    9 For
        recent attempts to quantify education externalities, see P. Romer, "Increasing Returns
and Long Run Growth" Journal of Political Econony 94, 1986, and R. Lucas, "On the
Mechanics of Economic Development" Journal of Monetary Economics 21, 1988.
                                              20



                                        II. HEALTH


   2.01 From 1983 to 1993, at least three salient features have characterized public health
   financing in China. First, real spending on health grew modestly, yet as a percentage of
   GNP it actually declined from 0.77 to 0.48. Moreover, health spending is low when
   compared with that in neighboring Asian countries whose income levels are similar.
   Second, capital expenditure in real terms fell; the bulk has gone to medical equipment
   purchases in hospitals above the county level. And, third, the ratio of expenditures on
   curative health care to preventive health services is very high.

   2.02 As shown in Table 2.1, real spending on public health in China was Y9.15 billion
   in 1993, representing per capita public health spending of Y8.3 (US$ 1.45 at the 1993
   exchange rate). Public health expenditures grew at 5.1 percent annually in real terms,
   and were slightly been affected by the decline in the share of total public expenditures
   of GNP. Overall public expenditure on health remained at around 0.6 to 0.7 percent of
   GNP, representing 2 to 3 percent of total govermmentexpenditure. As with educational
   expenditure, the spending levels and budgetary shares for the health sector of China are
   quite low when compared with those in neighboring Asian countries whose income levels
   and economic structure are similar. During the period of 1980-90, public health
   expenditure in 16 Asian countries averaged about 1.5 percent of GNP and 5 percent of
   total government expenditure (Table 2.2). The low level government spending for health
   is chiefly attributable to the existence of funding from the private sector, which accounts
   for about 1.4 percent of GNP, and to comparatively low capital costs in the public
   sector, particularly in rural health facilities Table 1 (Annex 2). However, under the
   assumption that data on user fees and household surveys are representative of the entire
   population, private health expenditure in China should actually be higher; in 1993, they
   constituted 42 percent of total health expenditure. This ratio of private to total health
   expenditure is not unusually high in comparison with those observed in the other low and
   middle-income countries in Asia, China appears to be more like Sri Lanka and the
   Philippines°.

   2.03 The composition of public health expenditure has changed steadily from 1983 to
   1993; growing share of recurrent expenditure of total health expenditure is attributable
   to the continuous decline in capital expenditures and the government's policy of
   maintaining higher non-salary expenditure. From 1985 to 1987, civil works absorbed the
   bulk of capital expenditure. Since 1988, however, capital budgets have been dominated
   by medical equipment purchasing, and physical infrastructure maintenance which,
   however, is relatively inexpensive. Although the ratio of salary to operating cost in health


     10 From various health sector studies of the World Bank, private as a proportion of total
national health expenditure varies as follows: Bangladesh (60%, 1987), India (59%, 1987),
Indonesia (60 %, 1990), South Korea (67%, 1990), Sri Lanka (41 %, 1990), Viet Nam (59 %,
1990) and Philippines (46%, 1990).
                                        Figure 2.1




              RelationshipBetween Health Expenditureand
                                                   (1993)
               GNP per Capita in 16 Asian Countriies
  Per capita health expenditure (US$)



  80 _
                       Thailand




  50 -




  30-         SiL*


       China/   PFhinpnes
       Laost S Indor,esa Myanmar
  10




       0       1,000      2,000         3,000   4,000   5,000   6,000   7,000
                                             per
                                          GNP capita




Source: Table 8 (Annex 2)
                                         21




                    Table 2.1: China: PublicE:xpenditure Health
                                                       on
                                    (billionYuan)


                                                                   As a percentageof
                 CurrentPrices                  1987 Prices                Public
Year        Recurrent Capital Total       RecurrentCapital Total   GNP Expenditure

1983         3.88      0.61     4.49       3.00     0.47   3.48    0.77    3.5
1984         4.44      0.74     5.18       3.61     0.60   4.22    0.74    3.3
1985         5.03      1.06     6.09       4.54     0.95   5.50    0.71    3.0
1986         5.96      1.18     7.14       5.63     1.12   6.75    0.74    3.1
1987         5.95      1.14     7.09       5.95     1.14   7.09    0.63    2.9
1988         6.66      1.05     7.72       5.95     0.94   6.89    0.55    2.9
1989         7.44      1.01     8.45       6.05     0.82   6.87    0.53    2.8
1990         8.12      0.73     8.85       6.15     0.55   6.70    0.50    2.6
1991         8.78      0.68     9.46       6.24     0.48   6.72    0.47    2.5
1992        10.38      0.95    11.33       7.05     0.65   7.70    0.47    2.6
1993        11.94      1.05    12.99       8.41     0.74   9.15    0.48    2.7

Average      7.1      0.9      8.1         5.7      0.8    6.5     0.60     2.9

Annual     26.1      21.1     19.9         4.8      6.5    5.1
Average
Growth (%)

 Note: Includescentral and local governmentexpenditures.
 OLSQ growth rates
 Source: MOPHand Staff estimates.
                                            22




Table 2.2: BasicIndicatorsof OverallHealthDevelopment,1993




                        Public health expenditure
                                      As a           Mortality rate              Percentage
                        As a          percentage      Jper 1000)      Life       changein infant
                        percentage of public                          expectancy mortalityrate
                        of GNP        expenditure Infant Maternal     at birth    (1965-93)

Bangladesh                0.8         4.5          92      3.5           55          -21
Bhutan                     1.7        2.6         127      NA            48          NA
China                     0.5         2.7          34      0.1           70          -61
India                     2.0         6.7          80      2.5           61          -43
Indonesia                 1.0         3.8          66      3.3           61          -37
Korea                     0.8         2.2          13      0.1           71          -60
Lao PDR                   2.8         4.9         125      6.6           50          NA
Malaysia                  2.9         6.8          15      0.3           71          -51
Myanmar                   2.1         6.8          72      0.7           60          -49
Nepal                     1.2         4.3         101      5.8           53          -29
Philippines               1.2         4.9          51      0.9           64          -36
Papua New Guinea          3.6        10.1          61      6.0           54          -54
Sri Lanka                 1.7         4.5          20      0.3           72          -54
Thailand                  1.6         6.1          26      1.0           69          -53
Viet Nam                  0.8         3.3          45      1.2           65          -50

Mean                   1.6            5.0          62      2.3           62          -46
PopulationWeightedMean                             56      1.4           65          -50

  Source: World Bank data file.
                                              23



   sector of China is not exceptionally high by middle-income Asian standards, a steady
   influx of physicians into existing facilities at a rate not matched by increases in the
   recurrent budget sometimes caused insufficient allocation of non-salary operating budgets
   in the mid-1980s, particularly for equipment and medical supplies". Consequently, not
   only did the salaries of health workers remain relatively low, but the quality of services
   in many provinces also deteriorated. The government was forced to attach an overriding
   priority to recurrent expenditure for the operations and maintenance of medical
   equipment and existing infrastructures. Although a favorable fiscal situation allowed
   capital expenditures to be restored somewhat in 1992-93, they are significantly less than
   those in 1986-87.

   2.04 The ratio of curative to preventive health care expenditure rose steadily in the past
   decade, primarily due to a shift in resource allocation away from rural cooperative health
   services to hospitals at county level and above (Table 3, Annex 2). Of total expenditures
   on curative care, spending in hospitals increased from 64 percent in 1980 to more than
   70 percent in 1993 stemming largely from purchases of medical equipment, higher fees
   for hospital services and raise in doctors' salaries. More importantly, however, has been
   the rising costs of overall health services particularly since the mid 1980s. Compared
   with neighboring Asian countries, China spends well over half of government spending
   on curative health services, well above the Asian average and close only to Sri Lanka and
   the Philippines. Other close comparators like Malaysia or Thailand that achieved better
   health indicators, spend a smaller proportion of their budget on hospitals and related
   curative health services. These differences may be ascribed to differences in the age
   pattern of the population and to rates of mortality'2 .

   2.05 Sources of fmancing health expenditures. Financing for health expenditures in
   China comes from three basic sources: the government budget, patient fees, and
   insurance sponsored by both the government and state enterprises. The most important
   features of Figure 2.2 are the continuous decline in the government budget for health,
   and the dramatic rise in patient contributions. Another interesting finding is that the
   spending level by the rural collective fund has declined much sharply starting in 1983.
   Of total health expenditures, the share of the government budget dropped from 25 percent
   in 1980 to only 14 percent in 1993, while the share of household contributions (patients'



    "In 1989, salary to non-salary operating ratio was 18/56, and comparable figures in other
Asian countries are as follows: Bangladesh (51/23), Nepal (53/37), Indonesia (39/36), Viet Nam
(16/65).

    12 Data from Health sector studies by the World Bank indicate that, in 1990, curative care
as a percentage of total public health spending varied was as follows: China (71%), Sri Lanka
(70%), the Philippines (65%), Malaysia (58%) and Thailand (57%). The average for all Asian
countries listed in Table 2.2 was 55 percent.
                               Figure 2.2




                                   China
                   Source of Financing Health Expenditure
                             (Pecentage Share)
percent


 90




 70 -




 40


                                            No:,
 20-




                   1980         1985           1990           1993
                                       Year

        9         Govemment
            Central                GIESb           RuralCollectiveFund

      []Local   Govemmenta         Labor           PatientFees

       a LocalGovemment aggregate
                      =         expenditure province county+ municipality
                                          of        +
       bGEIS:
            Govemment
                    Employees
                            Insurance
                                    System




        Source: Table 2 (Annex 2).
                                           24



fees) rose from 23 percent to 37 percent; the expenditure share in insurance sponsored
by state enterprises (labor insurance) increased from 30 percent to 37 percent. The
introduction of a user-fee policy and the dismantling of the cooperative health system
have been responsible for this shift in the sources of financing health expenditures from
the government to SOE insurance and patient fees.

2.06 With the devolution of health services financing, local governments now finance
more than over 80 percent of total government expenditures in health. But as revenue
collection and expenditure allocation effectively came under the control of local
governments, they shifted health resources toward greater spending on income-generating
investment. Indeed, this practice led to a precipitous drop in their share of total health
financing, from 24 percent in 1980 to only 12.9 percent in 1993. Moreover, since
collected revenue is shared up from the local governments to the center, the central
government is essentially recipient, and has thus had little leverage to increase spending
on the health sector. Consequently, the central government's share of total health
expenditures declined from its peak of 1.8 percent in 1986 to only 0.8 percent in 1993.

2.07 The dramatic rise in expenditure contributions by GEIS and labor insurance
schemes is attributable to the rising number of claims from individuals age 60 and older,
a legacy of declining fertility rates and then attendant aging of the population; but it is
also due to the higher prices of drugs and medical supplies, and to purchase expensive
medical equipment as technological advancements have been introduced. Moreover,
"moral hazard" factors have escalated the costs of health services even further--that is
the existence of the GEIS and labor insurance schemes have created incentives for health
providers to offer more care than patients actually need. Overprescriptions by health
providers, impelled by more profit earning opportunities, have led to the
overconsumption of an ever expanding array of newly developed and expensive
technologies for diagnostic tests and surgical procedures. Also, substantial revenues
from drug sales provide incentives for hospitals to over-prescribe.

2.08 The most remarkable trend, however, has been the growing contribution of patient
fees to total health expenditures; these include direct fees for hospitals, clinics and
physicians. Since 1983, patient fees have more than doubled in real terms, and have
compensated for the decline in expenditures by rural collectives. The elimination of
communal agriculture in the 1980s dismantled rural collectives, and those who had been
covered under this scheme were forced to pay for health services. Rural cooperative
medical care services were financed partly with a public welfare fund supported by rural
collective enterprises partly from the contributions of insured farmers (2 to 10 yuan per
person monthly). The elimination of collectives adversely affected the medical care
scheme, and farmers who were better-off under the management-contract system later
lost interest in contributing to the scheme. Today, the scheme covers only 10 percent of
the total rural population, a sharp drop from the 60 percent coverage in 1980. Compared
                                               25



    with neighboring Asian countries, the changing pattern of health financing sources
    resembles the pattern only in Korea, in which a health insurance system is also emerging.
    Although other countries also show a high level of private spending, as in China, they
                                              3
    have very few health insurance schemes' . The expenditure shift to patient fees and
    government-sponsored risk-sharing pools appears to be consistent with the policies of
    other countries in which mortality rates are similar and which are undergoing health
    sector reform.


    Equity Aspects of Health Expenditure

   2.09 Table 2.3 presents a summary of the average number of inpatient and outpatient
   visits by province, ranked by per capita income from poor to rich. The usage figures
   indicate how public health spending has been targeted. ThLey  show that the redistributive,
   or target, impact of public health spending was highest for users in the poorest provinces.
   In 1980, the average number of inpatient and outpatient visits were highest among the
   two poorest quintiles of provinces, but these indicators are at odds with an apparent
   reduction in usage among poorer provinces in 1993. The elimination of the rural
   cooperative health scheme and the introduction of user fees prompted costs of access to
   health services to rise among the poor.


            Table 2.3: Annual Average Number of Inpatient and Outpatient Visits
                       in Provinces by Quintiles Per Capita Income

            Quintiles of Per
            Capita Income              1980                   1993
            in Provinces     Inpatient    Outpatient Inpatient Outpatient


             I                 2.9           3.7           1.8        2.8
             II                2.6           3.5           1.7        2.7
             m                 2.7           3.4           1.9        2.4
             IV               2.2            2.8           2.11       2.5
             V                1.8            2.5           1.6        2.2

            Source: Staff Estimates from MOPH data.




   "In Korea, insurance accounted for 25 % of total health spending in 1989; in the Philippines
and Sri Lanka, insurance contributed only about 4% and 2 %, respectively, to total health
expenditures in the same year.
                                             26



2.10 Indeed, according to the 1993 National Health Survey (NHS), utilization varies
positively with household income per capita. Province-wide data indicate greater under-
use of service in poorer provinces is associated with economic reasons, and vice versa.
Of total households, 58.8 percent in rural areas and 39.8 percent in urban areas do not
use public health facilities due to economic difficulties. Another factor is the deterioration
in quality of public health services due to the reduced public spending since the mid-
 1980s. The emergence of private sector and liberalization of pharmaceutical sales
permitted users, particularly the rich, to move away from public sector to seek costly and
perhaps better quality private health care providers. This is partly substantiated by the
 1993 NHS. Of total households 11 percent in urban areas and 3 percent in rural areas did
not seek public health provider due to poor quality, and about 7 percent used private
health clinics but mostly in big cities and wealthy urban areas. A third factor may be the
declining age-specific morbidity rates, requiring fewer visits to health facilities. But, age
and gender-specific mortality and morbidity rates have shifted from infectious to more
endemic diseases, and not declined dramatically to warrant such marked reduction in the
utilization of health services. This overall decline in public inpatient and outpatient
utilization in poorer provinces indicates that the redistributive impact of public health
expenditures is increasingly becoming less favorable. This finding clearly indicates more
intervention in preventive services would be desirable on both equity and efficiency
grounds.

2.11 Household Health Expenditure. Annual per capita household expenditures on
health care are shown in Table 2.4, ranked according to quintiles, from rich to poor in
urban areas. The poorest households spend more on drugs and medicine than do richer
households. The top quintiles spend more on services that require fees for physicians
and hospitals. The amount spent on health care rises as households income increases and
increases more than proportionately with consumption. This pattern is revealed by the
elasticity of expenditures, which exceeds unity for each item with respect to total
consumption expenditures. One important point is that the poor are not paying lower
prices for services rendered by physicians and hospitals, given that per capita
expenditures on services do not vary widely by income groups.

2.12 The expenditure elasticities of various types of health and medical items provide
valuable clues about whether government subsidies are targeted more at the poor. If the
subsidy is not differentiated by the category of service or by the type of recipient, and
if modern health care is treated as a luxury good, then the subsidy will not help the poor.
But if the services are differentiated by the category of care, then it is easier to identify
specific services for public subsidies. The findings here suggest that fees for services are
a good option, given that expenditure elasticities are low. Although primary health care
is a pro-poor service, it now appears that the poor are spending less on this health care
component; thus, these services are increasingly becoming more of a luxury good and
service for them. A similar breakdown of private expenditure on various medical items
is not available for rural areas. Yet the broad picture by province reveals that
households in rural areas of low income provinces spend less than those in high income
provinces (SSB, 1994).
                                             27



           Table 2.4: Per Capita Household Expenditures on Medical and Health Care,
                                       by Income Group
                                         Quintiles (Yuan)


                        Quintiles of household per capita income Expenditure   T-ratio
                          I        II      III     IV         V   Elasticity

  Instrument             0.20     0.28      0.31 0.41        0.58    0.85      1.04
  Primary health care    0.40     0.66      0.93  1.33        1.72   1.45      1.93
  Drugs                 23.43    26.69     31.83 33.38      :38.35   8.11      4.56
  Herbal medicine        2.04     2.69      3.47 4.24        6.67    6.63      2.87
  Servicecharges         3.95     5.17      4.84 7.06        8.00    1.15      0.77
  Othera                 0.47     0.51      0.72 1.01        0.85    0.94      1.23
     Total              30.47    36.00     42.10 47.43      56.16

   a. Home visits.
   Source: SSB: Urban IncomeExpenditureSurvey, 1992.


Outcomes and Efficiency

2.13 To unravel how public health expendituresaffectedthe healthoutcomes, we first
order provinces by per capita income from poor to rich, and then map per capita health
expendituresand infantmortality rates with 1993 data. The figure 2.3 exhibits how the
maximumoutcomes--areductionin infant mortalityrates--hasbeen achievedwith least-
cost interventions.The figure clearly shows that public spendingon health is regressive--
that greater health expendituresare made in richer provinces. A more instructivefinding
is that income and vital health indicatorsappear to have played a considerablerole in
budgeting decisions.Poor provinces, including Henan, Guizhou, Yunnan, and Jiangxi,
in which infant mortality rates are higher than the national average, received meager
budgetary resourcesand spent less. Contraryto pro-health risk targeting, poor and high
infant mortality provinces--Hunan,Sichuan, and Guangxi--receivedconsiderably low
health budgetary outlays.Chief outliers are Tianjin, Beijing, and Shanghai, which have
high per capita income, a low infant mortalityrate, and hjighper capita health spending.
(TechnicalAnnex A)

2.14 Althoughgovernmenthealth expenditureshave been pro-poor in 1980s, yet such
progressivityhas been reduced in the late 1990s and theix efficacy appears to be quite
modest in poorer provinces (Figure 2.4). The Figure 2.3 also shows that lower income
                                                               Figure             2.3



        CHI-NA:Infant                      Mortality                    &- Percapita                         Health               Expenditure

                                                               Yuan                                      per           1,000      live         births
    Poor
                    Guizhou                                                              ______________
                    Guanaxi
                        Anui
                      Gansu
                    Sichuan                                                                -________
                     Jiangxi                                                               -            =               3
                    Yunnan
                    Shaanrxi
s                    ~~Hunban                                                      _______________f

                    Ninaxia
    [               Oinc~~hai:_\\\i\\\\\'
                    %Vnxi                                                          ____-_E                   _I
        Inner      Mo1Yolia                                                                                       -


                      Hebei                                                        _
                    Xinjilang                                                          ______
                      Fuj'ian"______
             He     ionjiang
                  Shgn?don_
                     Jia;ngst                                                              -
                   Zhejiing                                                       _             _
                   Liaoning                                                   _
                 Guangdono,                                              _
                     Tianjin                                                       -
                        Beijing        -              _          __-_
                   ShanghalI
    Rich                              40         30              20                10               0       20          40     60         80     100         120
                                                                        M         PHEXP                  M            IMOR

    Provinces              are ranked            by percapita                           income



                                                          Figure             2.4



                         CHINA: Percapita                                    Health                     Expenditure

        40



        30 _                                 __                              _          __                                        -   _                  /




        20.




         olo
             1      3             5    7     9            11        13           15        17           19            21     23       25         27          29
         Poor                                                                                                                                           Rich
                                                           ^     1993                           1982

Provinces                 are ranked             by percapita                          income
                                            28



provinces benefitted less than wealthier provinces from China's progress at reducing
mortality. In some cases, the interprovincial variations have widened during the 1982-93
period, because more rapid declines in infant mortality rates have occurred in provinces
whose infant mortality rates were initially lower. A marked disparity in infant mortality
persists ranging from a very low 11 per 1,000 live births in Beijing to a high of 98 per
 1,000 live births in Tibet in 1993. Likewise, the maternal mortality rate varies widely
by province -- only 6.5 per 10,000 pregnancies in Shandong to a high 73 per 10,000
pregnancies in Tibet. These variations are associated prinarily with differences in per
capita household expenditures by province, as well as in govermment expenditures, the
lack of health facilities, or both by province. Based on household income expenditure
surveys, a simple statistical analysis with regressions shows that infant mortality rates by
province are negatively associated with per capita household expenditures. The estimated
elasticity is -0.31, indicating that a 10 percent increase in expenditures tends to reduce
infant mortality by 3 percent, while a 10 percent increase in doctors and nurses per capita
(a proxy for health facility) is associated with a 4 percent lower infant mortality rate.

2.15 How have health expenditures translated into greater access to basic health
facilities and services? Data on the coverage of sanitation and safe water supply, and the
proportion of infants immunized with DPT3 in all provinces, indicate the answer. Figures
2.5 through 2.7 reveal that observed differences in the provision of these public health
services by province. Sanitation coverage varies proportionately with income. Although
poorer provinces have benefitted from greater sanitation coverage, significant variations
exist among poorer provinces. Public safe water coverage also varies widely by province.
Safe water is available to more than 80 percent of the population in high-income
provinces, but reaches less than 50 percent of the population in most of the poor
provinces. Among all public health service programs, immunizations of diphtheria,
pertussis, and tetanus of third dose (DPT3) served poorer provinces well, and its
relationship with income was stable except for a few cases.

2.16 Relative Effectsof Determinants HealthStatus. For morethantwo decades,
                                   of
China has made remarkable stride in improving the health status of its population. Given
the extent of this progress, it is important to examine whether the improvement in health
outcomes can be linked to specific components of public expenditure or, as is also
possible, whether the achievement is due mainly to gains in overall income overtime.
Thus, this section posits to what extent can particular government policies such as public
expenditure on health, access to various health services iniprove the health status of the
population?

2.17 Regressions are used to assess the impacts of changing government policies and
income levels on infant mortality rates and life expectancy at birth. The data used are a
time-series of 1983-93 from cross-section of 29 provinces in China. To isolate the
relative effects of each determinant of health improvement:, three different econometric
                                        Figure 2.5




                         China: Coverage of Safe Water by Province
                                  (% households)

100



 80 -....                                ..



 60 -...                                                                              ...


 40...........




   o
       1         3   5     7   9   11      13   15   17      19   21   23   25   27         29
           Poor                                                                       Rich
                                   69 1993                19B2




                                        Figure 2.6




                         China: Coverage of Sanitation            by Province
                                  (% households)
100



 80(



 60



 40-



 20




       1      3      5    7    9   11     13    15   17     19    21   23   25   27         29

   Poor                                  19182Rich
                                                                      Figure 2.7




                             CHINA: COVERAGE OF DPT3 BY PROVINCE
      % of infants

100
 9 0 . , . .......
       ....                  .. . ....               ............   ........
                                                                           ,    _.__R....
                                                                               ,,.......



 6 01 . . .........                                                                                             .. .1.
                                                                                                                     .
                                                                                                                    ..




   B0          3          5              7g      9/             1        13          1          7        1        1                      3            2               7       2
 40                                                                                                  .... __
                                                                                                          ..._...                              .. .          ........ .-. ........
                                                                                                                                                                   .....
                                                                                                                                                                  __._          ..
 30                  _....                     , .........
                                              ..,.,     ..                                           .................   ....... ...........
                                                                                                                               .                          .............. ....._._..........
                                                                                                                                                                       --            ....




  O        I                                                    1        1       |   1 |        1 |          1 t          1 t            1 ,
      1        3          5              7       9            11         13          15         17       19              21           23              25             27             29

   POOR                                                __                                                                                                     RICH
                                                             Nt DPT92
                                                                    a                       °    DPTB2

Note: Provinces are ranked by percapita                                                                         income
                                                 29



                                Table 2.5: Regression Results
                           Dependent Variable: Infant Mortality Rate

     Right hand                          OLS                  Fixed             Random
   side vanables                       Estimates              Effects           Effects


   log Income per capita               -0.688*                0.049             -0.509*
                                       (9.828)                (1.28)             (-6.97)
   log per capita Health                0.249                -0.114*            -0.531*
     Expenditure                        (4.44)               (-2.20)            (-8.16)
    Safe water                         -0.234*                 .052               NC
                                       (-4.178)                (1.3)
    Sanitation                           -.154               -0.286*             -0.200
                                       (-1.811)              (-3.404)           (-0.83)
   DPT3                                 0.134                -0.259*            -0.242*
                                        (1.13)               (-2.192)            (-3.15)
         N                                348                  348                348
   Adj. R2                               0.55                  0.96               0.20


   *Indicates significant at 5% confidence level.
   Numbers in parentheses are the t-statistics.

                                                             4
   procedures are used, and Table 2.5 presents the results1 .The empirical results indicate
   the importance of income, public health expenditures, sanitation and DPT3. Income is
   consistently the most important variable in reducing mortality. The OLS estimates show
   that provinces with coverage of safewater and sanitation facilities have had low infant
   mortality rates. The positive coefficient on public health expenditures reflects that more
   money is being spent in those provinces where infant mortality rates are high. The impact
   of such expenditures is manifested in the fixed and random effect columns. Indeed, the


    14  To disentangle the relative impacts of income growth and public policy variables
(expenditures, sanitation, safewater, DPT3) on infant mortality rates and gains in life expectancy
at birth, three methods are used, simple OLS without provincial dummies; a fixed effect model
with provincial dummies, and random effect with generalized least squares (GLS). (See
Technical Annex C .for details)
                                             30



results show that overtime, greater allocation of government resources did have a
favorable impact in reducing infant deaths, suggesting that higher expenditure, generally
synonymous with expanded supply of drugs, consumables, and equipment, can be
effective in improving health status.

2.18 Immunization and sanitation clearly have discernible independent iinpact on
improving health status in all empirical specifications. The DPT3 variable measures the
percentage of one-year old children receiving a full series (3-shots) of diphtheria,
pertussis and tetanus immunizations. Results also show that the efficacy of the logistical
support of health system because the completion of three-shot sequence requires a fully
organized personnel who interact routinely with the population. Likewise, sanitation
needs community support and better coordination with health personnel. Interventions in
sanitation and immunization, spearheaded by commune health system in China have
attained remarkable successes to combat infectious diseases and reduce infant deaths.
Expanded coverage of these services over the past decade as shown in Figures 2.6 and
2.7 tend to substantiate these findings.

2.19 But when an interactive term between income and government health expenditure
is included in a separate regression, a somewhat interesling result emerges (Technical
Annex D). In this case the coefficient on the expenditure variable by itself is the first
derivative if income is zero that tends to counterbalance the direct effect of expenditure
on infant mortality given a negative coefficient estimate on the interaction. It is difficult
to disentangle autonomous government spending from implicit subsidy associated with
the presence of more doctors and therefore induced by more demand and income. Other
than income, such public health services as santiation and DPT3 still show strong effects
on improved health status. At least this is consistent across all regression estimates, and
further justifies improved access to these basic services among rural and poor people.

2.20 Finally, the impact of government policy on health status is assessed with
regression analysis using the cross-sectional country level data. In each country, a panel
of cross-section (provinces or regions) and different time-series data were utilized. The
results in several studies indicate uniformly favorable impact of immunization and water
supply on health status in the Philippines, Malaysia and Viet Nam, while sanitation and
imnmunization more effective in China. Poor country such as Viet Nam faced stricter
               are
economic constraints than others in the sample during the period under review, and so
improved health status is a result of rising income growth than public health spending.
The dissimilar effects of public health spending in Viet Nam also reflects greater
importance of other health promoting interventions for reducing infant deaths in a period
of deteriorating health infrastructure, high incidence of infectious and water-borne
diseases and poor quality of services. (See Technical AmnexD)
                                        31



                             III.   SOCIAL SECURITY


Introduction

3.01 For the past decade,the socialsecuritysystemof Chinahas facednew challenges.
Amongthe majorschemesin the system,the unfundedpay-as-you-go        (PAYG)retirement
pension in whichpensions are paid from the current   contributionsof activeworkers and
current revenuesare increasinglybecomingunsustainable.The reasons are compelling.
On the one hand, demographicfactorsincludingan aging of the populationhave begun
to add pressure on the system, creatinga larger number of pensioners than there are
contributors.Currently, pension schemeis supporting3 pensionersper one contributor
on average.Exacerbatingthis trend is the fact that PAYG systemsdo not accumulateany
savings to pay for the pensions they promise. On the other hand, faster market
orientation may threaten the integrity of and thus the confidence of workers in the
systems. For example, enterpriserestructuringand real wage growth may jeopardize
expanding coveragefor new active workers. Too, rising unemploymentrates due to
enterprisereformsmay force government allowearly retirementandto relax eligibility
                                         to
criteria. Given these demographic and economic factors, workers in many state
enterprises have begun to fear that the system will no longer be able to provide the
pensions that they are promised, thus they may resort increasingly to evading their
pension contributions.

3.02 In 1980s, China became one of the first socialist countries to initiate a
comprehensive   reform of its plannedeconomicsystem;it is now consideringthe reform
of its socialsecurityparticularlypensionandunemployment    benefit schemeto ensure that
they accord with the principles of the country's new socio-economicpolicies. The
current PAYG system may suffer increasingly from enterprise reforms and other
problems associated with declining production-- a shortfall of resources that are
transferred from an economicallyactive populationto a growing number of retired
persons. The plannedeconomicsystemoffered very generouspension benefits, and the
system was unfunded. But, initiallyas the transitionbegan, governmentrevenuesfell,
includingthose earmarkedfor the pension system, reducing average pensionbenefits in
real terms. At the same time, opportunitiesfor retired persons to supplement their
incomewith secondary    jobs becameless available. Alongwith economicproblemscame
cultural changes in the tradition of shared housing and income pooling particularly in
rural areas. Younger people are now looking to save more money to meet their own
needs rather than using their savings in pooling arrangementswith older generations.
Thus, conflicts between the entitlementof retired persons and the living standards of
younger active personsare emerging. Perhaps the most damagingresult of this conflict
 is the response of young active workers--thereal losers in the system--asthey may try
 to evade paying their pension contributions. Policymakers are thus increasingly
                                         32



recognizingthat the Chinese socialsystemmust be reformedif better financial solvency
and a more equitableand efficient system are to be achieved.


                          OVERVIEW OF THE SYSTEM

3.03 In China social security benefits are grouped into two main categories: social
insuranceand social assistance. Socialinsurancehas five major schemes:(a) retirement
pension; (b) unemployment insurance; (c) work-injury insurance; (d) medical-care
insurance; and (e) maternity insurance for female workers. While social assistance
covers three major schemes: (a) social welfare; (b) disaster relief; and (c) special
assistance for disabled army personnel, families of martyrs, and resettlement for
                                         for
demobilizedsoldiers.Theresponsibilities the two main programsare dispersedamong
three ministries: the Ministry of Civil Affairs (MOCA), the Ministry of Personmel
(MOP), and the Ministryof Labor (MOL).The MOLoverseessocial insuranceprovision
for employeesworking in the state-ownedenterprises(SOE), the MOCA is responsible
for social assistance, while social insurance for employees working in government
organizationsand public institutionsfall under the jurisdiction of the MOP (Table 3.1).


A. Social Insurance

3.04 Retirementpension. All socialinsuranceschemesare of the definedbenefit type
that is benefits are calculated on the basis of employee's payments and contributory
services.The contributionlevelsboth by the employersand employeesare set to provide
sufficient income support to employees for contingencies such as retirement,
unemployment,injuryand disability,sicknessandmaternit,y--and administrativecosts
                                                              the
 of the schemes. Although retirement pensions provides mandatory basic universal
coverage for governmentemployees,coveragein SOEs sometimesvaries according to
the type of employeeand pooling arrangements.

           Level of benefits: Basically,the retirement pensions are a percentage of the
      last standardwage or salary excludingbonuses. Unlike many countries, retirees
      in China continue to live in houses provided by the government during their
      employment, and pay rent equivalentto 1 to 2 percent of pensionableincome.
      Currently, pensions are based on a national scale, ranging from 75 to 90 percent
      of the standardwage, and depend on the numberof years of serviceand dates of
      employment.The minimumvesting period for a full pension is 25 years, but in
      certain occupationsearly retirementis allowed. In such cases, minimumyears of
      service for pension eligibility varies from 10 to 15 years. In 1993, the average
      replacement rate was 75 percent. But due to the lack of price and wage
      adjustments,high inflationhas actuallyeroded the real value of pension benefits
      in recent years.
                                              33



           Table 3.1 China: Main Features of the Social Security System

     Type of Benefit       Responsible         Beneficiaries               Benefit       Source of Finance
                             Agency


A. Social Insurance        MOP, MOL,     All employees of the       Money and service   Budget: Central and
                           SOEs, and     government                                     local govermments,
    Retirement pension     some COEs     organizations, state-                          Contribution:
    Unemployment benefit                 owned public                                   Employers, and the
    Work injury benefit                  institutions, SOEs, and                        employees
    Medical-care benefit                 some COEs
    Maternity benefit

B. Social Assistance:      MOCA          Homeless children,         Money and service   Budget:Central and
    Social Welfare                       childless poor elderly,                        local governments,
                                         the disabled, and                              Donations: Non-
                                         homeless mentally                              governmental
                                         disturbed in urban areas                       sources


    Disaster Relief        MOCA          Disaster-stricken areas    Money, food,        Budget Central and
    Social Relief          MOCA          Households below           clothing, medical   local governments,
                                         poverty line               service             Donations: Non-
                                                                                        governmental
                                                                                        sources

    Special Assistance     MOCA          Disabled army              Money and job       Budget: Central and
                                         personnel, families of     allocation          local governments
                                         martyrs, and
                                         demobilized soldiers




            Level of contributions: Operating on a pay-as-you-go (PAYG) basis, the
       Chinese pension system is strictly an unfunded defined-benefit scheme. Employees
       contribute 3 percent of their standard wage. Contributions are mandatory
       including contract workers, non-managerial and non-professional employees
       employed since 1986. The employee contributes 12 percent of total wage bill,
       making a total contribution rate of 15 percent. However, due to differences in
       pooling arrangements the rate of employee contribution varies from 2 to 3 percent
       of either total or basic wage depending on the category of employees, financial
       position of SOEs in different provinces. For example, all employees of SOE in
       Beijing contribute 2 percent of total wage, whereas in Guanghan and Yantai is 3
       percent of basic wage. Moreover, for contract workers, there is no fixity in
       employer's contribution to their pension fund because it is subsumed in the SOE's
       overall contribution to the pool.

          Demographicfactors: Demographic factors have played a crucial role in
       Chinese pension system and will inevitably be more important to its sustainability.
                                           34



        More than two decades ago, China attained the demographic transition of low
        birth and death rates; consequently, the proportion of people age 60 and older has
        increased progressively to 9 percent in 1993, while the system dependency ratio
        which indicates the total burden on the active working population, will exceed 10
        percent in the year 2000 and projected to to rise by 20 percent in the year 2030.
        Besides the growing maturity of the pension schemes, which enables more
        workers to become eligible for a pension as tirme progresses, the much more
        slower growth of active workers relative to retired population will aggravate the
        system dependency ratio even further.

 3.05 Unemployment insurance. Since 1986, the government has also provided
unemploymentcompensationprimarily to workers in SOEs. Essentially, the scheme
covers workers who are laid-offbecauseSOEshave been restructuredor gone bankrupt,
workers who are terminated for disciplinaryreasons, amdcontract workers upon the
completionof their contract. Sincefew SOEsin Chinahave undergonerestructuringthe
scheme appears to be a compensationthat goes to workers on short term contracts.
Before the mid of 1980s, unemploymentwas not even discussed in the country for
political concerns. Workers were kept in SOEs regardless of whether they created
redundancy.But after the start of enterprise reform in the early 1980s and especially
sincethe implementation the Bankruptcy
                          of                 Law in the mid-1980s,the proportionof loss-
making state enterprises has remainedmore than 30 percent for years. Indeed, actual
urban unemploymentrate has risen from 2 percent in the mid-1980sto 2.6 percent by
the end of 1993. Infact, it is estimatedthat more than 30 percent of workers in SOEsare
redundant.

 3.06 To meet the increaseddemandfor financingthe redeployingthe unemployedand
redundantworkers, the staterequestedlocal governmentssetup their own unemployment
insurancefund through contributions     pooled from enterprises. Currently, enterprises are
required to pay 1 percent of their total pre-tax wage bills to the local unemployment
fund, which financesunemploymentrelief and training in the localities. Accordingto a
state regulation, the relief benefit for the unemployedshouldbe higher than local social
relief standard, and lower than the local minimumwage criterion. Therefore, the benefit
varies from region to region according to the local social relief and minimum wage
criterion. Normally, the replacementrate varies from 50 to 75 percent of average basic
wage and depends on the duration of employment. The maximum duration of
unemploymentinsuranceis two years, if employedfor 5 years or more, and one year if
employed for less than 5 years. As with the retirement pension, the beneficiaries of
unemploymentinsuranceare permittedto stay in houses offered by the government,but
they do not receive separateallowancefor house rent. Unlikemany developedcountries,
there is no mandatoryactivejob searchand training programsfor beneficiaries.In 1993,
about 87 percent of SOE workerswere coveredby the scheme. However, the fund has
                                           35



reportedly been misused, only about 10 percent of the fund has been used for
unemployment relief since the pooling started in 1986.

3.07   Medical-care insurance. Medical care insurance covers both retired and active
employees in the government and SOEs. The GEIS covers health insurance for the
governmentemployeesand the Labor Insurancesupervisesthe coverage for employees
of the SOEs. In 1986, contractworkers becameeligible for health insurance. There is
no minimum qualifyingperiod for eligibility.Coverageincludesthe costs of treatment
by doctor, maternitycare, hospitalbed, surgeryand ordinarymedicine. Employeespay
registrationfees both for outpatientand inpatientcare in public health facilitiesand for
house calls. Mandatorycontributionby employeeswas introduced in 1986, and it is
equivalent to 5 percent of employee's wage (Chapter II). For dependents, medical
insurancecoversthe full cost of treatmentby doctor in designated public healthfacilities,
and 50 percent of the cost of surgery and ordinary medicine. The national norm for
contract workersand their dependentsis same as for permanentworkers.

3.08 Work-injury     insurance. Government    organizations,                    and
                                                             public institutions SOEs
finance work-injury insurance for their employees as part of their social insurance
obligations. There is no minimum qualifying period. Temporary disability benefits
include 100 percent of basic wages during the temporarydisabilityuntil recovery or the
onset of a permanentdisability. Permanentdisabilityinsurancecovers 80 to 90 percent
of standard wage. The minimum benefit is 50 yuan a month. Like medical-care
insurance,work-injuryinsurancesystemsis also under reform in an experimentalstage.
Currently, over 700 cities and countieshave joined the work-injurypooling program.
In additionto localdifferencesin the rate of contribution,there is also sectoraldifference
in the rate of contribution because the rate of work-injury incidence varies among
different sectors. Currently, the average rate of contribution among all participating
enterprises ranges between0.1-3 percent of their total wage bills.

3.09 Maternity insurance. Permanent and contract employees in the government,
SOE and some COE, are eligiblefor this benefit. Maternityinsurancecoversbenefitsfor
medical-careduring child birth, and incomefor female workersduring maternityleave.
The standardpackageof benefitsinclude100 percentof earningpayableby employerfor
upto 90 days due to child birth, and 100 percent of eamings payable for 15 to 45 days
in case of abortion. Normally, employer bears the entire cost, but in some SOEs
contributionto the funds have startedbut it differs among localities.


B. Social Assistance

3.10 As one component of the social safety net, social assistance benefits are
determined according to certain demographicentitlement categoriessuch as homeless
                                           36



children, childless poor elderly, the disabled, homeless mentally disturbed in urban areas,
disabled army personnel, the families of martyrs, demobilized soldiers, and people in
disaster-stricken areas. The Ministry of Civil Affairs is the responsible agency at the
central level for the provision of social assistance to these people. Assistance takes the
forms of financial help, material support, and job allocation.

3.11 Social welfare. During the 1987-93 period, local governments spent a total of
Yuan 8.6 billion for homeless children, childless poor elderly and homeless mentally-
disturbed. The money was used to construct welfare homes, payment for welfare
workers, as well as the living for the assisted. Besides local governments, expenditures
on social welfare are supplemented by community resources. In addition to financial and
material assistance, the local governments also provide job opportunities to disabled
people, enabling them to engage in economically productive activities to support
themselves. At present, China has over 50,000 welfare enterprises in urban areas where
the disabled make up the bulk of the employees. During the period of 1989-93, these
enterprises received a total of Yuan 490 million credit from state banks at discount
interest rates for technical upgrading. Meanwhile, the China Lottery Fund-Raising
Committee for the Disabled also donated a total of Yuan 29.8 million.

3.12 Disaster relief. China's disaster relief scheme provides financial and material
support to areas stricken by floods, earthquake, and cyclones. This category of social
assistance is the only one that is financed by the central government. During the 1987-93
period, the central government appropriated a total of Y8.6 billion for disaster relief,
while local government spent Yuan 1.83 billion as counterpart fund. Although actual
expenditure on disaster relief varies from year to year according to the severity of
disasters, it is an earmarked item in the state budget. The relief fund is mainly used to
provide food, clothing, housing and medical-care to the disaster-stricken people. Of all
the disaster relief fund from the central government, 30 percent are to be used by local
governments to assist local poor people specifically.

3.13 Special assistance. As with social welfare, special assistance financed by the
local governments, which render fnancial support to the families of martyrs, disabled
army personnel, and provide job services for the demobilized soldiers resettled in their
areas. This expenditure captures a large share of total spending on social assistance in
local government budgets, averaging 43 percent during the period of 1987-94.


Public Expenditure: Trends, Structure and Composition

3.14 The social security system is financed by three major sources: the revenue budgets
of the central and local governments; the budget of state owned enterprises (SOE); and
employee contributions. With regard to social insurance, the central and local
                                            37



governments finance the benefits of their respective employees in government
organizations and public institutions for retirement pension, medical-care benefit, work
injury and maternity. However, under special circumstances, the fiancing of the social
insurance is also shared by both of these organs of the government. On the other hand,
the bulk of social assistance in China is financed by local governments, the central
government only provides disaster relief to localities stricken by severe natural disasters.
However, although expenditures on disaster relief is an earmarked item in central
government budget every year, wherever this source is not enough, the local
governments have to meet their own needs through their own resources. Apart from local
governmental resources, expenditures on social welfare is supplementedby resource from
the community-level collective economies run by the neighborhood communities in urban
areas, and the village communities in rural areas. Some of these communities take care
of the needy in their own areas, especially of the childless poor old people, the disabled,
the homeless mentally-disturbed, and the family of the army-men and martyrs. The
expenditures on such assistance are met from the income of the productive or service
enterprises run by these communities. However, the coverage and quality of such
assistance vary according to the income of these enterprises.

3.15 Within social insurance, retirement pension spending as a share of GNP remained
stable around 1.2 percent between 1987 and 1993 (Table 3.2). The spending levels in
China are low compared to those in Eastern European countries and Korea. In 1992,
public expenditures on pension in Eastern Europe averaged about 8 percent of GNP
(Table 3.3). As with former socialist countries, the share of pension contributions from
wages is relatively high, which would tend to an underestimate of China's current
pension spending as a percentage of GNP. Since the pension system is funded from
workers' earnings, an increase in wages as a share of GNP will mean higher
contributions to the pension funds in the future. In addition, if newer supplementary
pension scheme, introduced in some cities, is assumed to mature at the 1992 level,
pension spending as a percentage of GNP could hover around 1.4 to 1.5 percent. Thus,
actual pension expenditure could be higher than the share reported in Table 3.3.
Furthermore, as the population continues to age, as more younger people move to private
sector, the number of active people in public sector will decline, thus reducing
contributions and eventually imposing a greater fiscal burden on public expenditures for
pension.

3.16 Figure 3.1 plots the relationship between per capita GNP and public expenditure
on pension as percentage of GNP in a sample of former socialist countries, China and
Korea in 1992. Overall, public spending in pension is positively associated with income
per capita. Part of this pattern is due to the fact that pension system becomes expensive
with aging of population particularly in Poland, Bulgaria and Hungary. The figure shows
that China is an outlier (negative) predicted for its income level, reflecting that the
country is underspending on pensions given its income level. In this sample, spending
                                             38



                                       of
     Table 3.2: Structureand Composition PublicExpenditureon Social Security



                                  1987 1988 1989 1990 1991 1992 1993


As a percentage of GNP

A. Social Insurance                 1.2     1.1     1.2    1.3    1.3    1.3     1.3
    Retirement Pension             1.2      1.1     1.2 1.3       1.3 1.3        1.3
    Unemployment Insurance         0.02     0.02    0.02 0.02     0.02 0.02      0.03
B. Social Assistance               0.25     0.23    0.24 0.24     0.27 0.20      0.17
As a percentage of
Public Expenditure

A. Social Insurance                6.6      7.1     7.3   7.0     7.8    8.4     9.4
    Retirement Pension             5.5      6.0     6.2   6.4     6.8    7.3     8.2
    Unemployment Benefit           1.1      1.1     1.1   1.0     1.0    1.1     1.2
B. Social Assistance               1.3      1.2     1.3   1.2     1.4    1.1     1.1


Source: Annex 3, Table 1.


pattern in China resembles closely only with that of Slovak Republic. The statistical
relationship explains only about 20 percent of its variation. But such weak influence in
this cross-sectional analysis does not contradict the fact that in single country, higher per
capita income tends to raise public spending on pensions.

3.17 In addition to pension scheme for retirees financed by earmarked taxes, China
provides medical insurance for retirees and active workers (chapter II), and
unemployment insurance. Real public expenditure on unemployment insurance has
increased from Yuan 2.6 billion in 1987 to 4.1 billion in 1993. This represents a stable
share in GNP of 0.02 percent, and in total public spending of 1.1 percent. However,
compared to all OECD and Eastern European countries, government expenditure on
unemployment benefits in China is very low because under the command economy the
country was not accustomed to unemployment, resulting into low urban unemployment
by international standard (2.6%), even after correcting for underestimation.
                                           39



     Table 3.3: Public Expenditureon Pensionsin SelectedCountries(1975-90)
                            (as a percentageof GNP)


 Country                                    1975         1980        1985         1990


 Bulgaria                                   5.0          6.2         6.8          8.8
 Czech and Slovak Republic                  6.4          7.2         7.8          7.8
 Hungary                                    5.6          7.8         8.9          9.9
 Poland                                     4.3          5.8         6.8          7.7
 Romania                                    3.5          3.8         4.4          7.2
 U.S.S.R.                                    -           -           5.8          6.8
 China                                       -           -           1.0          1.2
 High-income OECD countries                 8.4          9.8         9.7          9.8
 Low-income OECD countries                  4.4          5.9         8.0          8.2
 OECD average                               7.1          8.2         8.9           -


 Source: National Statistics Commission; OECD Social Data Bank; International
         Monetary Fund (1991); and International Labor Office.


3.18 Figure 3.2 shows levels of public spending on pension in selected countries.
Expressed as a share in GNP and total public spending, China and Malaysia are at the
low end of the spectrum while other countries occupy the high end. Generally, the
largest expenditure in social assistance is special assistance, which accounted for 0.37 to
0.47 percent of total government expenditure between 1987 and 1993, followed by social
welfare, with the pattern broken only with the upturn in disaster relief 1991 due to
unusual natural calamities such as earthquake and cyclone; expressed in relation to total
government spending, disaster relief outpaced other schemes in that year, reaching its
peak of 0.58 percent.
                                                                     Figure 3.1



                                Relationshipbetweenexpenditureon pensionas percentageof GNP
                                                 and GNPper capita

               14-
                                                         _ Ukrair                   Specification: Pension    a + B log GNPPC

               12 -

                          Uzbekistan
               10-                                           Hungary
     0z                           Georgia
                                          6
                                     zulgaraech                      na              Estonia
            08                               PolndCzech          *
          co                                   a
                                            URomanj                  *Belarussi
                 75Albania
          r O    6- Albani                                a
                                                   Kyrgyzstan      Latvia
     C 0.
     C.
                                               AzerbaijaS* Rsi
                                                            Russia
                 4u
                 4                                 Kazakistan-


                 2-

                                     Slovak                    Malaysia
                 0                *China

                      0                 1000          2000                3000            4000        5000      6000       7000
                                                                                 GNP per capita




                                                                     Figure 3.2



                                             Pensionas a percentof GDPand Public Expenditure

                                GDP                                              Public Expenditure


      Bulgaria
Czech Republic

      Hungary 2.:|

          Polandc

     Romania

       Albania              .        ....                                                                     Z
      Slovenia

          Russia
            China

    Singapore

            Korea

     Malaysia


                     10          5             0         5                10         15          20      25      30       35

                           % of GDP                                            % of PublicExpenditure
                                            40



               IV. MAIN ISSUES AND STRATEGIC DIRECTIONS


4.01 Despite commendable efforts by government to achieve greater balance in how
educational, health and social security services are delivered, two broad problems have
plagued all three sectors in recent years. First, the distribution of services, especially
education and health, continues to discriminate against the poor, largely because they
have not benefitted from economic growth and are being choked off from service access.
Second, some current mechanisms and arrangements for devolving government financing
for services have led to over-decentralization and loss of central government control that
have created ineffective targeting. Although dramatic economic growth and several fiscal
reforms have generated more government revenues in the past two years, the National
People's Congress has attached high priority to repaying the public debt and financing
the budget deficit, it has only modestly increased the 1995 State budget for education,
public health and social security. Obviously, these increases are by no means sufficient
for augmenting necessary public investments in the social sector. As such, the most
important interventions would be to increase the aggregate level of spending by local
governments that have surplus budgets and face little intersectoral competition, and
improve expenditure targeting by central and local governments that face budgetary
pressures. This is crucial, because each province competes with each other to obtain a
greater share of investment outlays, often, the location of projects rather than their
importance becomes a more critical criterion for central government allocation decisions.
Since local governments assumed major responsibilities for social sector expenditures,
future tax reforms should allow enhanced tax autonomy for them.                     Also,
intergovernmental grants scheme should be geared to ensure greater equity. Greater
complementarity should also be fostered between public and private financing, and
assistance for high priority needs should be sought from new or existing external donors
whose commitments have been modest in the past.


Education

4.02 Economic conditions of China have improved dramatically since 1991, and the
economy is forecast to grow at a faster rate than the school-age population, suggesting
that budgetary constraints on education will probably be eased. Economic rates of return
reinforce the justification for more spending in primary and secondary education mainly
to improve its capacity to enroll and retain poor students in economically disadvantaged
areas, and in higher education mainly for expanding specific fields of study in order to
foster an externally efficient education system. Based on the preceding analysis the
following strategies deserve urgent attention to bring about growth in the desired
directions. First is the targeting of subsidies for primary and secondary education, so that
more resources go to better education for poorer children, second is the improvement of
                                                41



    efficiency in higher education, and third improving student finance schemes in higher
    education.

    4.03 Improving Targeting of Basic Education. One emerging problem that may
    become critical in the education system is its weak capacity to retain students in primary
    and secondary levels. Significant selection occurs within cycles of education through
    dropping out in several poor provinces. This results in adverse selectivity in the sector.
    Children from poor families leave the school system ear]lierto start working with their
    parents so targeting subsidy to the poorest of the poor children even at primary schools
    is complicated, because it requires attracting students to a type of education that they may
    find very expensive in direct and opportunity costs and Inot appropriate. The targeting
    of the subsidies for basic education can be improved by increasing public spending
    through higher per student payments to schools serving poorer children and by reducing
    the payments to schools in better-off areas. With increased resources, the schools serving
   poor students would be able to increase availability of specialized personnel, equipment,
    and other services which would compensate for the disadvantages poor students bring
   from home and improve the overall learning environment. Analysis of regression already
   showed that increased public spending actually stimulates survival prospect of students.
   Complementary measures would include pricing reform to reduce the private cost for the
   poorest children. Such reform includes elimination of tuition and fees, distribution of free
   textbooks, and allocation of money to compensate for opportunity cost as was done in
   Sri Lanka. Area-based price reforms should be done on the basis of targeting schools
   only in poverty counties. Another step involves food for education in primary and
   secondary schools only for poor children as done in Bangladesh. So far, the measures
   adopted in Sri Lanka and Bangladesh have proved successfill at enhancing enrollment and
   retention rates among poor students.

   4.04 Secondary Education. Enrollment and survival rates in secondary education are
   lower in China than in other Asian countries at similar income levels. Our analysis
   suggest that this is mainly due to the lack of demand stemming from cultural factors
   rather than the lack of spaces. But the dynamics of China's new labor market should
   increase demand for secondary school graduates, necessitating some forms of
   governmental interventions. Although the present policy of 9-year compulsory education
   that extends the junior secondary level does enhance coverage, yet additional intervention
   via fmancial incentives for school attendance covering direct and foregone earnings
   particularly in less developed areas would be needed with the caveat that this incentive
   could be reversed when further justification no longer applies 15 Currently, the


    15 Malaysia and Sri Lanka incorporated reversibility in their compensation design. The
specific circumstances justifying their initial introduction did not continue when attendance and
cohort survival rates improved markedly.
                                               42



   underlying assumption for such intervention is not inconsistent with the empirical findings
   of transition rates from primary to junior secondary education. But empirical support for
   higher government subsidy for senior secondary education to promote attendance and
   stimulate survival prospects is apparently weak. However, this conclusion does not
   nullify the case for more narrowly targeted subsidy and financial incentives such as
   scholarships for poor but meritorious students in senior secondary education.

   4.05 Higher Education. Although social rate of return from higher education is low,
   yet future demand in China's expanding labor market argues for finding ways to enable
   universities to expand current enrollments. Once the demand for secondary school
   graduates rises, which is a likely prospect in itself, this pool will be enlarged from which
   only an expanded tertiary education can absorb the intake. In particular, engineering,
   business administration, agriculture, and medicine are becoming priority fields of study.
   But diversifying the content, type and level of courses offered at public institutions will
   require addressing three problems: internal inefficiency, disproportionate contributions
   from public and private sectors and inefficient student finance schemes.

   4.06 Improving Internal Efficiency. The structure of costs revealed that unit costs
   are very high in public tertiary education stemming largely from low pupil-teacher ratios
   and high research and development costs. Two important measures could alleviate this
   burden and make this subsector internally more efficient. First, internal structure of
   universities should be reconfigured by eliminating duplication of programs, streamlining
   administration, and increasing teachers' workload in tune with international standard.
   Second, the several currently underutilized campuses and facilities should be merged,
   creating economies of scale. Rather than building new classrooms for future enrollments
   existing facilities could be used in double shifts. Evidence from a sample Chinese higher
   education institutions show that unit costs decline steadily with rising enrollments. By
   way of illustration, a simulation based on regression analysis and data on size distribution
   show that a potential savings of more than 30 percent of the current operating cost of
   those institutions with enrollments of 2,000 students'6 . However, consolidating facilities
   to reap economies of scale is hampered by the current configuration of various line
   agencies, central and local, each with its own institutions. Rather than establishing
   separate facilities and compete for each other, all line agencies should work together to
   improve administrative arrangements that would raise modest pupil-faculty ratios.

   4.07 Selective Increase in User fees. Private tuition and fee contribute only 5 percent
   to total recurrent university expenditures, and in recent years cost-of-living subsidies have
   captured 20 percent of total public higher education spending. Both areas could be


    16 For corroborative evidence see also W. Min, (1990),"Chinese Higher Education: Mode
of Expansion and Economies of Scale" Jiaoyo Yanjiu , October, Beijing.
                                             43



 improved by reducing the burden of subsidization on the public sector. Given the high
private rate of returns to higher education, government subsidy may limit private
 spending that students and their parents want to make. Indeed, there are two indications
that the private sector is able and willing to pay: the existence of excess demand for
education i.e more applicants than available places, and donations. Rising enrollments
reflect that increasing user fees selectively would not reduce the demand for higher
education but instead allow government to mobilize private contributions for expanding
access and thus improving equity. However, setting tuition and fees according to costs
per student, rather than in absolute amount would be desirable so that charges could be
raised in accordance with inflation. The government should further reduce subsidies for
the cost of living that students would incur whether or not they are enrolled. These non-
instructional subsidies are still high relative to many Asian comparators and should not
be a substitute for private spending that could be made anyway by many households.
Such step can significantly alleviate crowding-out problem, freeing-up government
resources for expanding enrollments in other levels of education.

4.08 Student Finance Schemes. The foregoing indicat-esthat high subsidies and low
user fees in public higher education tantamount to a blanket scholarship and student loan
schemes that support a limited part of the entire system, which not only creates adverse
impact on equity but also thwarts the efficacy of public expenditure. As discussed below
such outcomes can be redressed by reconfiguring the existing mechanisms for student
fmance. It is clear that priority fields of study will be offered mostly in public
universities. But in future private (Minban) universities will continue to grow not only
in response to increasing demand but also as a resull: of limited places in public
universities. In such situation, there will be more pressure to expand the access to public
universities. But given the high unit cost in public relative to private universities, such
expansion may cause further inefficiency and greater social selectivity because the better-
off students will get more opportunities who would otherwise have gone to private
institutions.

4.09 Scholarship scheme targeted to poor and qualiEied students can avoid such
perverse outcomes. Although such scheme is already in place in China, yet, public
universities are still guaranteed a captive clientele and thus insulated from the competitive
pressures of having to justify the quality and cost-effectiveness of their educational
services to users and communities. So scholarships should be untied students should be
allowed to select institution of their choice, public or private. Thus spurring competition
and promoting efficiency. The student loan schemes are also available in 37 public
universities. However, to avoid re-emergence of blanket subsidy and higher default rates
in future, loans should be given only to students who do not qualify for scholarships but
who cannot fund studies from their own sources. Repayment programs should be made
                                                44



    into income-contingent payment systems where loans are repaid according to the post-
    graduate earnings of the borrower'7 .

    4.10     Enhancing Distance Education. Oualitv and Resource Generation. Further
    development of distance education would expand coverage and reduce unit costs. But
    many technical colleges and post secondary institutions produce graduates who are
    preoccupied with the need to obtain a formal degree for formal sector jobs, with
     inadequate attention to necessary job skills imparted through education and vocational
    training system. This leads to a declining quality of higher education. As such, adopting
    high standard of admissions, training and instructions, relaxing strict regulations for
    private universities, and providing accreditation on a differentiated scale of academic
    excellence are critical. Finally, although ample scope exits for cost savings and revenues
     from the sales of services to SOEs, donations, too much dependence on these sources
    may be counterproductive. Experience shows that the advantage in these cases may
    diminish beyond a revenue recovery rate of 15 percent, and educational institutions may
    turn into profit-making organizations. Evidence shows that in several OECD countries
    these sources do not contribute more than 10 percent of total university revenues" .    8
    University-run enterprises played an important role in income generation and may
    continue to do so. But institutions should undertake investments on the basis of
    comparative advantage and the demand for their products, sale and marketing of new
    technologies.


    Health

   4.11 The preceding analysis shows that the changing socioeconomic environment is
   transforming mortality risk factors in China. In particular, growing industrialization and
   greater concentration of the population have increased pollution emissions and discharges,
   and have strained infrastructure. Consequently, more age-specific deaths are coming from
   cancer, cardiovascular diseases, respiratory and other chronic diseases. Conversely,
   fewer deaths are coming from infectious and communicable diseases because health
   infrastructure for these diseases is well placed. The diagnosis and treatment of chronic
   diseases however require sophisticated technology, trained staff and special facilities,
   which eventually raise overall costs. Many patients should be able to finance rising costs
   with existing insurance or own resources, but the poor and indigent patients will



    17 In countries where credit culture and discipline within the student population is relatively
weak, early preparation for income-contingent deferred payment scheme is better. Such schemes
have led to low default rates in Tunisia, Australia, Sweden, and Quebec Province of Canada.

    1   For evidence see OECD, "Financing Higher Education" Paris, OECD 1990.
                                            45



increasingly be shut out from these medical services. An important issue is how current
practice of extending free or below-cost hospital care to poor people can be continued
in the face of rising hospital costs and how government financing can be executed more
effectively to poor areas without affecting quality of services. Economic reforms have
emphasized revenue generating curative services rather than basic preventive health care.
As a result several problems in health financing have emerged--an unbalanced allocation
of government resources, an unbalanced allocation of medical labor input, rising
insurance costs, a narrow redistribution of public health expenditures, and an unfavorable
user fee structure.

 4.12 Intrasectoral allocation of resources. The recent surge in infectious and
 respiratory diseases particularly in poorer provinces point to the fact that there has been
 a resource allocation problem, manifested in disproportionately greater spending on
 curative health services relative to preventive care; thereby benefitting mostly government
employees, SOE workers and urban population at the expense of a vast majority of rural
population including the poor. As Table 3 (Annex 2) indicates spending on curative care
is high both in real terms and as a share of total public health expenditure. Although this
patterm reflects clearly the epidemiological transition and a greater demand for hospital
care, funding for disease prevention services, township hospitals (the first referral
points), anti-epidemic stations, and health promotion activities are being squeezed out
disproportionately, leading to a much higher incidence of risk factors. Further reduction
in funding would increase the risk of premature deaths. ;Several health risks are linked
to the behavior of individuals, and have a combined synergistic effect. In most cases,
eliminating or reducing exposure to risk factors--a preventive intervention-- is more
effective and less expensive than secondary treatment. ]3y way of illustration, cancer
treatment in the form of surgery, chemotherapy and radiation is less effective and more
expensive than preventing exposure to such risk factors as smoking, alcohol intake,
environmental pollution, and the absence of screening and immunization services.

4.13 A major health policy priority for the government should be to increase the
aggregate level of spending on preventive health services and health promotional
activities. Indeed, regression analysis based on cross-provincial level data suggests that
such preventive measures as safe water, immunization and sanitation have had a profound
impact on the health status of the population. And a case for increased levels of spending
for these services is particularly strong in provinces where current levels of public health
spending are much below the national mean. Owing to the rapid economic growth in
China, the extent to which greater allocation can realistically be provided is not
constrained by vigorous intersectoral competition for resources that had previously shifted
resources away from the health sector toward income generating projects.

4.14 Another policy priority for the government is to generate savings by improving
the efficiency of treatment protocols for major chronic diseases. For example, less severe
                                            46



medical cases can be detected and treated by nurses or paramedical staff in inexpensive
township medical centers with simple screening or diagnostic procedures and inexpensive
drugs. The MOPH has already taken actions to shift treatment protocols, yet many
township medical centers especially in poorer provinces still lack the equipment, supplies,
physical facilities, and outreach programs necessary to ensure effective therapy and
diagnostic treatment to build awareness in the community about how the onset of
morbidity and endemic diseases can be delayed or prevented. Public spending on health
promotional activities should be targeted at training health personnel to inform and
motivate the population about the benefits of changing their lifestyle ,providing incentives
to industries to adopt less hazardous technologies, and developing family hygiene
regimens.

4.15 Still another policy priority for the government is to raise tobacco tax where
feasible, and redirect hidden hospital subsidy to health prevention and promotional
activities. All public hospitals receive subsidy in the form of rent-free buildings and
medical equipment. In several cases, these subsidies benefit hospitals that are already
well developed and cater those who are already insured. For all practical purposes,
subsidies to public hospitals should be eliminated or reduced gradually and any accrued
savings from it should be directed at rural township health centers. Any deficit in hospital
expenditures can be financed from co-payment from insured patients, or a trust fund of
money from hospital profits and government revenue can be created to meet any future
hospital budget deficits.

4.16 Allocation of Labor Input. Alternative mixes of labor inputs are influenced by
the allocation of public spending which in turn influences health outcomes. In China,
indices of labor input for western medical treatments show an inegalitarian allocation
across provinces. Poor provinces are deprived of more qualified medical staff. Table 4.1
shows the distribution of more qualified health personnel (doctors, assistant doctors) and
less qualified health personnel (specialized nurses, nurses, and midwives). Computed on
the basis of relative personnel ratios (RPR) the richest quintile of provinces-- Beijing,
Shanghai, Guangdong, Zhejiang, Tianjin and Liaoning clearly have an oversupply of all
medical staff but particularly of more qualified staff, as the value of RPR far exceeds
unity. Conversely, the poorest quintile of provinces--Guizhou, Xinjiang, Anhui, Henan,
Guangxi and Sichuan (Annex 2 Table 7)--have an undersupply of all medical staff
including more qualified personnel. Other less disadvantaged provinces in the second
quintile--Jiangxi, Yunan, Shannxi, Hunan, and Tibet--also have an undersupply of more
qualified medical personnel especially doctors.

4.17 Besides their impacts on health outcomes, the different mixes of health personnel
has important underpinnings on cost efficiency. In provinces with high morbidity and
mortality rates from infectious and communicable diseases, redeploying the personnel
mix toward more preventive health services staff like nurses and midwives can generate
                                           47



                   Table 4.1 Labor Input Mixes by Province (1993)


   Province
 by quintile                         Relative Personnel Ratio (RPR)a
 of per capita                   Special         Asst.
    income        Doctor          Nurse          Doctor         Nurse           Midwife


  I                0.81            0.77           0.82             0.67           0.74
  II               0.94            0.95           1.07             0.90           0.61
  III              1.34            1.23           1.34             1.26           1.05
  IV               0.86            1.15           1.24             1.16           1.40
    V              1.90            2.38           1.76             2.48           1.33


RPR = (Provincial Staff/Provincial Population)/(National Staff/National Population)
RPR exceeding one indicates oversupply and less than. one indicates undersupply of
medical personnel.

Source: Staff estimates and Table 7 (Annex 2).



further cost savings. To attract and retain more qualified staff in poorer provinces, the
government should also provide such incentives as hciusing, bonuses, education for
dependents, and better posting after service in rural areas. To minimize future imbalance
in labor input mixes, another powerful policy instrument is to upgrade training and
retraining for less qualified health personnel and then substitute them for more qualified
staff where feasible. To implement these policies, the government should ease its
currently rigid controls on remuneration to make it more amenable to local needs.
However, whether the local or the central government should be given authority to
choose the structure of such benefit package remains an open question.

4.18 Health Insurance. To control the high costs of medical services the GEIS and
Labor insurance schemes have introduced mandatory payroll deductions, devised several
cost-sharing formulas and promoted lump sum fixed capitation payment to providers.
Mandatory contributions, introduced in 1986 but implemented in 1991 currently vary
from 2 to 5 percent of employee wages. Cost sharing formula was adopted in 1991.
Accordingly, users pay from 20 to 40 percent of costs after the fixed amount is met by
the insurance scheme. Finally, capitation, introduced in 1991, sets limits on payments
to hospitals and doctors by the GEIS and Labor insurance schemes through negotiations.
                                           48



Employees must select their doctors and use hospitals designated by the insurance
contracts, and are also reimbursed only for essential drugs listed by the MOPH.
Moreover, in some cases, hospitals retain unused money from lump sum capitation
payments by the government, but are liable to reimburse only 50 percent of any over-
spending from that.

4.19 While these measures are commendable two important problems with the health
insurance system have emerged in recent years: rising costs and uneven coverage. From
 1982 to 1993, real government spending on two insurance schemes--GEIS and Labor--
has increased dramatically, even faster than revenues Table 2 (Annex 2), and some
beneficiaries of the scheme are better-off than others. Three interlocking shortcomings
are inhibiting the efficiency and equity of the system. First, although the cost sharing
formula is intended to be uniform, co-payments in the Labor insurance scheme vary from
one SOE to another according to their financial solvency. For instance, in SOEs with
inadequate financial assets, the formula applies only to medical care, other SOEs share
the costs of both medical care and drugs. In addition, many SOEs allow employees to
claim unused money from their free allowance at the end of the year. In such cases
employees bear all or a substantial part of unforeseen medical costs that exceed the
cashable allowances, making it harder to justify the principle of mixing good or bad risks
that is crucial to social insurance. Second, the 50 percent government financing of over-
spending in lump-sum capitation creates an unnecessary burden on its budget. And, third,
while there is a policy for pooling funds, yet separation of funds for the GEIS and Labor
Insurance scheme may create distortions and impair such pooling in the future. The
government should pursue four options to redress these shortcomings: centralize the
management of health insurance, standardize co-payment rate, pool medical costs, and
introduce health card system in rural areas. Implementing the first and second options
would facilitate implementing the third, all of which would make the system operationally
more efficient. The fourth option would promote equity.

4.20 One problem with cost-sharing formula is that it gives providers incentives to
select patients who require high-cost treatment, reject those with high risk, and pile on
costs of marginal and questionable services. Conversely, standardization of coverage
cannot be implemented when financial resources of SOEs vary considerably, and
marginally profitable SOEs may find it difficult to maintain minimum coverage.
Moreover, in future, serious problems may crop up if SOEs continue to face bankruptcy
or merger due to ongoing SOE reforms. Therefore, centralizing insurance scheme
management would enable authorities of both GEIS and Labor insurance to monitor
patients' cases more effectively with a view toward reducing unnecessary services, and
standardizing co-payment rates would simplify administrative mechanisms.

4.21 Gradual steps for pooling medical costs should also be sought through expanding
coverage for contract workers and their dependents, and by unifying GEIS and Labor
                                         49



insurance schemes.A combinedpool would create a mechanismfor smoothingthe cost
                                                                  it
of health servicesover time by equalizingit over the emaployees, would also permit
                     to
cross-subsidization those employees who are less able to contribute to cost sharing,
and would give the local governmentsgreaterbargainingpower in negotiatingcontracts
on capitation payments. Some cities, such as Shanghai, Yanai, and Quendao are
considering unifying GEIS and Labor insurance schemes with compulsory payroll
deductions equivalentto 5 percent of wage. Another policy option would be to make
hospitals liable for all overspendingon the capitationpayments, forcing them to invest
and create trust funds with profits and unused capitation money, thus gradually
eliminating the 50 percent govermnent contribution to such over-spending. Free-up
resourceswould give additionalleverageto the governmentfor subsidizingthe poor and
indigentwho are not coveredby insurance.

4.22 While China has been successful at covering a vast majority of its urban
populationwith its two major healthinsuranceschemes,its rural risk-coverage programs
have weakenedsignificantly.The rural populationnow relies on governmentsubsidyand
their own resources.To remedythis problemone alternativewould be to introducerural
health card system.Thailandhas this schemein operationsince 1983. Under this scheme
households can purchase an annual health card, usually priced at less than the annual
average householdexpenditure.It entitles the purchaserto a fixed number of treatments
and an unlimited number of visits for preventiveservices such as maternal and child
health care and immunizations.This schemehas sustainedfor more than a decade and
accordingto evaluationsurveysthis is very popularamongrural householdsin Thailand.
Village-level drug revolving funds are another scheme for local health insurance
coverage.

4.23 Finally, as the economycontinuesto grow, the private sector will expand as will
the privatehealthproviders. Peoplemay need supplementary   insuranceandprivate sector
can play an importantrole to cover additionalhealth services.In areas where the private
sector can step in without much difficulty, the govenmment encourage private
                                                                can
insurance thus promoting greater diversity and competition in the supply of health
services. Specific legislation should be enacted to guard against abuse, promote
community risk-rating and forbid exclusionof pre-existingmedical conditions. Since
mandatory social insuranceis already in place implementingsafeguards over emerging
private insuranceshould not be too difficult.

4.24 Targeting of resources: One implicationof the stiuctureof sourcesof fmancing
is that the insuredconsumethe bulk of healthservicesin China, thereby shiftingservices
away from the vast majority of uninsured and indigent patients. Compounding this
inequity is the wide disparity in the distribution of provincial government health
expendituresparticularly in the past few years. Further econometricanalysis with 1993
data indicatesthat province-levelpublic spending on health responds more strongly to
                                              50



   changes in GDP in high-income provinces than in low-income provinces. Moreover, a
   one percent growth in provincial income is associated with much greater increase in
   government health expenditure in richer provinces than in poorer ones. These results are
   disturbing, since they imply that even with equitable growth in income across provinces--
   an unlikely prospect--the distribution of provincial government expenditure on health will
   become more unequal. If economic growth in the province alone is less likely to reduce
   the imbalance in provincial health expenditures, a strong case can be made for the central
   government to target an increasing share of central government resources to the poorest
   provinces. '9

   4.25 While a larger proportionate increase in government spending on health in the
   poorer provinces than in the richer provinces will clearly promote equity, its relative
   effect on efficiency is not immediately obvious. Indeed, the latter depends much on the
   marginal impact of increased public health expenditure on health outcomes. Table 4.2
   reports the impact of a 1,000 yuan increase on infant mortality rates by quintile of
   provinces. An increase of 1,000 yuan in per capita provincial government health
   spending in high-income provinces is associated with drops in infant mortality of only
   0.12 percent. Conversely, similar increase in per capita health expenditure leads to a
   decline of infant mortality by 3.04 percent in poorer provinces. Thus, a 1,000 yuan

            Table 4.2: Impact of a 1,000 yuan increasein provincial government
                           health expenditureon infant mortality, 1993


              Quintileof         Percentage         Infant       Maternal
              Provinces         of population      mortality     mortality


                I                   30.8             -3.04          -1.01
                II                  15.6             -2.45          -0.67
                InI                 17.2             -1.05          -0.44
                IV                  24.2             -0.43          -0.23
                V                   13.9             -0.12          -0.03


              Source: Staff estimate from MOPH data.


    l9 The implied GDP elasticity of provincial government health expenditures varies from a
low of -0.098 for Guizhou, the poorest province in China, to a high of 2.56 for Shanghai, the
richest province.
                                            51




increase in health spending yields much faster decline ia infant and maternal mortality
in poorest provinces. Because poorest provinces account for more than 30 percent of the
total population of China, a strategy of redistribution of government health expenditure
from richer to poorer provinces will generate a larger overall decline in infant mortality,
and as such will improve the overall efficiency of public health financing.

4.26 User Fees. The combination of recent worsening income distribution, reduced
government health spending and price decontrols (resulting in higher prices for health
care and other consumables) reduced utilization of health services, consequently
worsening their health status particularly in poor provinces. The government is concerned
about cushioning the poorest of the poor from further higher health costs stemming from
changing age pattern of diseases and greater cost recovery in public health sector. Indeed,
better targeting of public expenditure to improve access of the poorest to health facilities
would depend on two policy instruments: reducing user costs associated with treatment
and drugs, and improving the quality of care for those who actually gain access.

4.27 A more effective mechanism should be devised to protect the poor from user fee
increases; the current mechanism in place in China to ecempt the indigent from fees is
not working. As household expenditure data showed there is a weak relationship between
household income and the average price paid for treatmrient,that is, households in the
bottom income quintile spend a disproportionately high share (13 percent) of their total
health expenditure on one annual visit. They also spend more on drugs, which are
generally not subsidized. Of total private health spending, the poorest households pay 77
percent on drugs compared with 68 percent by the richest quintile. Thus, the burden of
paying for service charge and drugs is higher among the poor than the better off. Since
this disproportionate burden prevents the poor from usinig public health facilities, then
improving access for them may require selective price reductions for essential services;
the poorest hard-to-reach group should be exempt from paying fees, which may be
compensated for by increases in public spending and other insurance fees. This needs an
appropriate pricing policy that differentiates prices by the income of users

4.28 Finally, improving the quality of public health services would enhance utilization.
One indicator of the quality gap is the variation across inicomegroups in the amount of
household expenditure on public health, but another indicator, available from the 1993
NHS, is that 15 percent of households do not use public health facilities due to the lack
of beds (4.1 %) and the absence of proper services and treatment (11.3 %). Interprovincial
variation is even wider. Another indicator is that the poor are less likely than the better-
off to receive treatment by qualified and trained doctors when they are sick. Spending
on medical equipment, supplies, drugs, and physical infrastructure in township health
centers and the redeployment of trained personnel in needy areas would considerably
improve the quality of services available to the poor.
                                           52



Social Security

4.29 In the social security subsector, the social assistance schemes absorb an
insignificant proportion of the government budget; they are also well targeted, and do not
currently suffer from any serious design problem. The bulk of public expenditure is
absorbed by two social insurance programs: retirement pension and health insurance.
Although unemployment insurance currently consumes only a negligible proportion of
government budget, the continued restructuring, merger, and bankruptcy of SOEs may
create more unemployment and in the future adding surplus labor to the pool. If these
problems go unchecked, unemployment compensation may become a budgetary burden.
Therefore, some actions are now needed to make this scheme more efficient.


Program of Pension Reform

4.30     Given its current economic reforms political realities, and culture, China should
undertake a step-wise approach to reforming its pension system. First, the government
should restructure the benefit formula and then unify pension schemes of government
employees, SOEs and COEs. These medium-term measures could be implemented
within 3 to 5 years. Eventually, these measures would reduce the spending levels of the
pension fund, in turn reducing the distorted incentives and gradually improving the
financial condition of the system. However, given demographic pressure, the medium-
term program will not be able to ensure the sustainability of the system in the long run.
Thus, a bold reform, incorporating a multi-tier (pillar) system, will be necessary to avoid
any painful transition later. Enlightening here is the Communique of the Third Session
of the Fourteenth Party Congress on November 11, 1994 where multi-pillar funded
system has been endorsed as a final objective of old age pension scheme in China.

4.31 Program of medium-term reforms. The medium-term reform priorities are to
(a) improve the benefit formula; and (b) unify all individual pension schemes. To
accomplish these, the following measures are important.

   (a) Restructuring benefit formula:

   Increase the normal retirement age. China's normal retirement ages are lower than
   those in East European countries and all OECD countries--60 for men and 55 for
   women on average. In view of country's current and future economic conditions and
   projected demographic profile, the present low retirement ages appear unjustifiable.
   The retirement age could be set at 65 for men and 60 for women. But, this age
   threshold should be raised gradually, increasing it by one year every two or three
   year. Raising retirement ages will immediately yield benefits of a higher active
   worklife ratio in turn containing the costs of the system. To illustrate, according to
                                         53



a recent simulation by the Ministry of Labor, a rise irnthe retirement age to 65 in the
year 2020 would yield three contributors per pensioner, higher than the two
contributors currently.

Introduce an alternative linear accrual factor. The declining linear accrual factor
used currently creates relatively high pension rates for all income brackets and for
periods of service of fewer than 40 years; it also creates higher pension expenditures
than does a constant accrual factor. The decreasing accrual rates also discourage
workers from remaining in their current jobs beyond the standard retirement age. To
minimize these distortions, the government should establish a reduced accrual rate of
credited service, which could be linked to net salary after pension contributions are
deducted.

 Introduce an actualized lifetime income for Dension.            The current method of
 calculating pensions on the basis of the last basic wage, excluding bonus, creates
 several problems. First, the method fails to establish clear links between contributions
 and benefits because two workers with the same length of services and lifetime
 income may get completely different pension solely on the basis of the difference in
 their final standard wage. Second, it leads to system abuse as it gives an incentive to
employers to set arbitrarily higher wage at the time of retirement, translating into
higher pension. In this respect, employees having good relationships with employers
tend to get much higher final wages than others. The thke in final wage for preferred
employees thus creates unnecessary distortions. Third, because of inflation any
diminution in real value of final basic wage naturally reduces real value of
replacement rates. Alternatively, the use of lifetirne actualized earnings for pensions
should be introduced gradually, preferably including one additional year each calendar
year. Indeed, this method would avoid system abuse and eliminate the uneven
treatment of workers from different age groups and length of services. Normally,
actualization should be based on changes -in the consumer price index or average
wages.

Impose ceilings on contributions and benefits. At present, there is no ceiling on
either contributions or benefits. Ceilings provide early signals to employees about the
scope of the public pension system. An imposition of a benefit ceiling would release
additional resources that could be used to replace the income of the poor. Another
potential salutary effect of ceilings is that it encourages the development of
complementary private pension schemes.

Introduce pension indexation. The absence of routine inflation-adjusted pensions
erodes the real pensions of older beneficiaries during the high inflationary period.
Thus, introduction of automatic inflation adjusted pensionable income would maintain
its real value, and creates incentives to contribute. It is worth pointing out here that
                                           54



   price indexation should not be used when real wages fall, because lower real wages
   would translate into less revenues from contributions. When real wages rise, real
   pensions could be adjusted. The proposed indexation could be adopted once economy
   stabilizes. In addition, the government should apply actuarial decrements for early
   retirees and actuarial increments for late retirees to eliminate distortions that favor
   either group. Should the policy of subsidized housing for retirees and unemployed
   change, and if rents are raised in line with market levels, the government should also
   take into account of housing costs in pension and unemployment benefit calculations
   to offset potentials of homelessness. Also, as discussed below, the proposed second
   and third pillars in the scheme should be able to provide additional cash for
   cushioning such eventualities.

4.32       (b) Unify the pension system and ensure portability. China's old age
pension scheme is highly decentralized. Wide regional differences in standard of living
may well justify a decentralized approach. For that reason the experimental new pension
schemes continued to evolve uniquely at the local level. Also, it is easier to implement
a scheme designed by municipal and provincial government than one imposed by the
central government. But, decentralized system may become increasingly unmanageable
once number of pensioners rises much faster than hitherto due to aging and divestiture
of SOEs. One option is to adopt provincial old age pension schemes with a set of
national guidelines that would allow portability of pension. Such a move would certainly
generate greater economies than the present separate pools for SOE, COE and
government employees. Perhaps a unified pension system--with its single organizational
structure would be most desirable in order to promote portability, foster the efficiency,
improve fund management, and prevent evasion. Different rates of contributions and
benefits, and status division between government employees and SOE workers hinders
mobility which inturn hampers portability. Unification into a common contribution and
benefit formulas would ease this problem. Likewise, efficiency gains can be attained by
economies of scale from merging all branches and by pooling the human resources from
the MOL, MOP and MOCA into one specialized unit for social insurance. Through
unification it would also be easier to adopt new legislation that covers all components of
social security including fund management. However, a larger redistribution of assets
across SOEs is being resisted by new SOEs with few retirees and subsequently precluding
a move toward a national pension scheme.

4.33 Prograin of long-term reform. When combined with on-going macroeconomic
and SOE reforms the proposed medium-term reforms should be able to create
preconditions for operating a multi-pillar pension scheme. These reforms would help
effect improved accounting and auditing standards, create a pool of trained personnel,
improve tax administration, and foster tradeable financial market instruments. Some of
these preconditions currently exist in China. A multi-pillar system should consist at least
three components:
                                                 55



     4.34         (a) The first pillar. When all pension schemes are unified, the system could
     then be modified, and to become ultimately the first pillar of a multi-pillar system. The
     first pillar would be modelled in tune with the current PAYG basis, but there will be
     some mandatory contributions from employees and employers. Employees should
     contribute from their salaries, and employer i.e government should preferably finance
     this pillar from general revenues than payroll taxes. Benefits could be flat rate or
     earning related with a low ceiling. The flat- rate component of the first pillar would seek
     to augment a redistributive objective, replacing higher proportion of income for poor
    people and vice versa. The earning-based component would seek to achieve the objective
    of intergenerational fairness. In view of socio-economic condition of China, the flat rate
    pension should pay a minimum pension equal to 30 percent of average earnings. All
    pension payments should be based on net earnings and could be proportional to the actual
    length of service. Thus, workers who have the full length of service would receive the
    full minimum pension. Under the second component, pensions should be based on
    average actualized lifetime earnings. Both components of the first pillar should be
    indexed (Table 4.3). It is worth pointing out here that the MOL has proposed a scheme
    similar to what is being advanced here; a defined benefit scheme with two components,
    flat payment based on average city wage and second payment based on a percentage of
                           20
    average lifetime wage /.

   4.35      An initial contribution rate of 5 percent of payroll would be sufficient to begin
   with. The rates should rise to a maximum of 10 percent, depending on the demographic
   profile of active workers. If the dependency ratio increases, contribution rates should
   be increased, if not, then the normal retirement age would have to be raised which would
   in turn increase expenditure savings. Experience in OECD countries suggests that a
   contribution rate of 10 to 15 percent for the first pillar is sufficient for paying a 30
   percent net pension 21/. To demonstrate, consider the following simple example:
   If Wt= net real wage growing at g per year, c= contribiution rate, n=length of active
   life, m = length of retired life and p = net pension or replacement rate i.e the ratio of
   pension to wage. If n=40 and m=15 and p=30%, then the required contribution rate
   is:
              E W* (l+g)        t*c*n=E    W* (1+g) t*p*m

                   E       W*(1+g) t*p*m
                       E    W*(1+g) I*n
              c   = p * m /n



   20/ At present, the city of Shanghaihas individualaccountssystem.Two other citiesChongquingand
Wuhan are also implementingsuch scheme.

   21/ See for example Dimitri Vittas,"TheSimple Algebra of P'ensionPlans", World Bank, CECFP
Mimeo, 1993.
                                            56



c = (0.30) * (15)1 40 = 11% ; similarly, ceteris paribus, for net replacement rate of
70%, the required contribution rate would be 26% per annum.



            Table 4.3: China: SuggestedLong-Term Program of Pension Reform


  Pillars                   ContributionRates          BenefitRates              Objectives
                            Employee Employer

 First Pillar                5%         10%        (1) Flat rate:         Redistributive
 (Mandatory)                                       Minimumpension,
 Government                                        independent salary
                                                                  of
                                                   level but
                                                   proportionalto actual
                                                   length of career,
                                                   i.e., flat rate amount
                                                   times years of
                                                   service.

                                                   (2) Earnings-             Intergenerational
                                                   related: Basedon          fairness
                                                   lifetimeand linear
                                                   accrualrate.

                                                   (1) and (2) together
                                                   shouldgenerate
                                                   pensionequal to 30-
                                                   35% of net average
                                                   earnings.

 SecondPillar            5% (*)                    Pensionequal to           Strong linkages
 (Mandatory)                                       40% of actualized         betweencontribution
 Private                                           averagelifetimenet        and benefits
                                                   earnings.
                                                                             Creates capital market
 Third Pillar            Varies                    Supplementary
 (Voluntary)                                       pension
 Private

(*) may vary accordingto the growth rates of real wages and real interest.
                                               57



    4.36       (b) The second pillar. The second pillar would be a mandatory savings
    scheme based on individual capitalization accounts. This pillar, currently functioning in
    Chile, Mexico, Malaysia and Singapore, aims to provide strong and transparent links
    between contributions and benefits, and to foster the creation of capital markets.
    Normally, contribution rates of 10 to 15 percent yield a pension equal to 40 to 45 percent
    of average net earnings. However, contributions are assessed according to the projected
    growth in wages and interest rates. The success of this pillar crucially depends on proper
    management. To reduce operating costs and achieve reasonable yields on investments
    the pillar should be managed by a central authority, either public or private, that
    maintains records, and should rely on private fund management companies to invest
    funds in productive ventures 2/.

    4.37 Individual account systems have been set up in some municipalities on a pilot
    basis to encourage contributions, but accrued benefits registered on individual accounts
    are actually notional. Since market is still underdeveloped, investment risk is high, and
    looses may be caused by mismanagement through institutional defects, developing this
    fully-funded, defined-contribution pillar would take sorne time. Experience in other
    countries in Western Europe show that failure to establish the funded pension scheme
   based on defined contributions may lead to the creation of an occupational and partially-
   fumded company-based pension system which would be less efficient and thus thwart the
   tranSition to a market-oriented economy. Because, company-based system would create
   long term liabilities for the firms, blocking efforts to restructure enterprises, and
   jeopardizing the fiscal balance if the government is forced to bail out failed pension
   schemes. Partially-funded schemes are generally anti-poor since they tend to exclude
   low-paid workers, small businesses, and agricultural worlcers, thereby benefitting higher
   income workers. As such the poor would be forced to rely on the public system as their
   only source for old age income security. Thus, occupational pension funds with high
   return are beneficial only to the rich, creating an inequitable system. Another problem,
   emerging particularly in industrialist countries from these funds is their inability to
   provide inflation-adjusted pensions, because, the employers normally do not have a way
   to hedge the risk through a suitable investment strategy. If the pension is not indexed
   to any wage or price level, the beneficiary will loose real value of pension due to
   inflation.'3



    I Chile introduced this pillar in 1981. It is privately managed, and has high operating costs
but high real rates of return.

    'Consider an inflation of 5 percent per year, then the value of 10,000 Yuan will have fallen
to 3773 Yuan if an employee retires after 20 years of service. The value of first year's benefit
will ultimately have a real value of 1352 Yuan after 40 years and continue to fall each year. If
the employer indexes the pension to cost of living both before after retirement, then the
employee will have a benefit worth of present purchasing power annually for rest of his/her life.
                                               58



   4.38 Too often, the employer-based final salary in the defined benefit plans of many
   OECD countries are biased in favor of late-comers who usually come from senior
   positions and adversely affect early leavers when their pensions are not portable. Given
   these factors, the Chinese government should develop the second pillar after privatization
   is soundly anchored. It is interesting and relevant to note that Chile introduced this pillar
   even before it undertook comprehensive banking reforms. In this respect, China is better
   positioned because it has taken actions to reform commercial banks.

   4.39 China can set an initial contribution rate of 5 percent by employees. Once real
   wages increase, expenditures in the public system will be reduced, allowing contributions
   to be raised to a maximum of 15 percent. To avoid a state monopoly and ensure
   operational efficiency, the govermnent should make the second pillar mandatory, but
   have it privately managed only for the investment of funds. As in Chile, Malaysia, and
   Singapore, there should be one account per pensioner and one investment fund per fund
   management company, thus fostering simplicity and transparency. If the contribution
   rates are insufficient, the government could transfer to this pillar some of SOE or own
   assets, selected for privatization, in the form of shares of stocks in those firms. This
   strategy would help pay some of the government tax and debt arrears. Divestiture of
   state enterprises can generate revenues to support investments, both domestic and
   foreign, in large, or long-term, risk-bearing assets that in turn can support this pillar.
   If appropriate, non-inflationary funding of this pillar through foreign grants is another
   possibility.

   4.40 (c) The third pillar. The first and second pillars together could provide a
   pension equal to at least 70 to 75 percent of net earnings. The third pillar should be a
   voluntary one such as standard bank deposit, life insurance, or marketable securities,
   whose main purpose would be to supplement the income security of the persons after
   they retire. The smooth operation of the third pillar would depend not only on the
   success of the first two pillars and but also on whether the banking system is reformed
   to ensure that voluntary pension plans and the interests of pensioners are protected.
   Foreign direct investments can play an important role in augmenting higher returns and
   accumulation of resources both for the second and the third pillars through establishing
   global funds for sales. However, macroeconomic stabilization, political stability,
   regulatory flexibility in foreign investment, and currency convertibility are key
   prerequisites for operating these portfolio investments in China. Moreover, funded
   pension plans that opt to use international management services to diversify and improve
   their performance will require legislative changes in several industrialized nations 24/




    24/ Legal constraintsand mental reservationsare the main barrier to increasedWestern European
investmentin developingcountriesand other non-domesticassets. See L.E Davanzoand L.B. Kantz,"
Towards a GlobalPension Market"Journal of PortfolioManagement,1992.
                                                59



    4.41 Transitional Arrangement. On the basis of administrative, financial, actuarial
    and legal aspects of the social pension scheme, transitional arrangements should be made
    because the proposed multi-pillar structure should take into account of the varying ability
    of different cohorts of workers to adjust savings behavior and their employment situation.
    As such, workers within five years of the minimum retirement age at the time of the
    introduction of the new system should have the old system. For workers between five
    to fifteen years of retirement should have two options: either retire under the old system,
    or transfer to the new system if it is more congenial. Specific transitional arrangements
    could also be established. The new multi-tier (pillar) system should be applied to all new
    workers.

    Unemployment Insurance

   4.42        While unemployment insurance in most OECD countries provides job search
    incentives, the system in China is purported first to encourage labor mobility and second
    to facilitate adjustments of labor supply in SOEs, mainly by reducing surplus labor.
   Although introduced by the central government, the scheme is supervised by local
   governments. Since its inception in 1986, the scheme has fostered labor mobility but
   done little to adjust SOE labor supply. Because a significant proportion of the surplus
   labor represents the old and unskilled, and because the scheme does not provide any
   incentives for training and retraining. So if they are lail off the prospect of being re-
   employed is slim, making the scheme virtually irrelevant for them. Moreover, the
   experience of labor markets in developed countries show that the long-term unemployed
   are less likely than short term unemployed to find reemployment. The current duration
   of unemployment benefit is already long (2 years), extending it would do nothing solve
   the issue of redundant labor. Rather the government needcs adopt a proactive strategy
                                                                 to
   focusing on job creation and retraining programs.

   4.43 To make unemployment benefits more meaningful and relevant first, training and
   retraining programs should be provided in the scheme, tailored specifically to upgrade
   skills of surplus labor force. Second, where conditions permit, SOEs may hand over their
   welfare services to local governments. And, then they can provide subsidy to employers
   if they hire long-term unemployed, to compensate them fo,rhaving to upgrade their skills
   or to retain workers they may otherwise have displaced. But, strict eligibility criteria
   should be followed such as the length of unemployment period of the workers.
   Otherwise, such wage subsidy may be abused5. Third, in an expanding economy like
   China, a lump sum severance payment to long-time workers, in lieu of monthly



    25/ For instance, employers may accept subsidy but hire workers without utilizing that subsidy
(substitutioneffect), and an unemployedworker may not be hired because they are not eligible for the
subsidy (displacementeffect).
                                              60



   unemployment could be introduced. This will provide workers with a stock of capital that
   can be invested in other economic activities or private businesses, and they would have
   an incentive to seek jobs more quickly, since they would no longer have any economic
   benefit to remain unemployed for the duration of the insurance. Experience in OECD
   countries show that unemployment compensation has a negative effect on post-
   displacement wage and the duration of unemployment--the longer and greater the
   benefits, the longer the spells of unemployment 26. Moreover, the cost of administering
   one lump sum payment would be lower than the costs of monthly installment.




    26Thus unemployment    benefitsprovideneeded compensationto jobless workers, but they do not
necessarilylead to an equivalentor higher payingjob. See W. Narendranathan al,"Unemployment
                                                                          et
BenefitsRevisited" The EconomicJournal,June 1985
ANNEX TABLES
                                       ANNEX 1



            Table 1: Public Expenditure by Level of Education


                 -Current Prices (bil Yuan)-                  ----   (Percentage)----
     Year      Primary   Secondary     Higher        Prirary         Secondary    Higher

     1984       4.23       4.2          3.27           36.2            35.9       27.9
     1985       5.26       5.21         4.02           36.3            36.0       27.7
     1986       6.11       6.05         4.69           36.3            35.9       27.8
     1987       8.03       7.75         6.06           36.8            35.5       27.7
     1988       8.13       7.06         6.15           38.1            33.1       28.8
     1989      10.30       9.53         8.46           36.4            33.7       29.9
     1990      11.77      10.58         8.84           37.7            33.9       28.3
     1991      12.64      11.35         9.63           37.6            33.8       28.6
     1992      15.42      13.69        11.06           38.4            34.1       27.5
     1993      21.33      20.03        18.22           35.8            33.6       30.6

Source: SEdC
                                       ANNEX 1


Table 2: Expenditureon Education By Level of Government

        Year      Central    Local      Total

        1987       12.9      87.1       100
        1988       14.1      85.9       100
        1989       13.0      87.0       100
        1990       13.7      86.3       100
        1991       12.9      87.1       100
        1992       13.2      86.8       100
        1993       12.8      87.2        100


a. Local Government= Provincial+County+Municipality

Source: SEdC
                                            ANNEX 1


Table 3: Sources of Financing Total Education
                (Percentage)



 Sources                    1991     1992       1993


Government Budget           68.4     67.2       65.7

EducationLevy               10.1     10.5       9.4

Tuition & Fees              4.8      5.1        8.2

School Enterprise            5.4     5.4        6.1

Community Fund Raising       8.6     8.8        6.6

Other                       2.7      3.0        4.0


Source: SEdC
                     ANNEX 1

Table 4: Public Expenditure on Education by Category



              (BillionYuan)      Percentage
                      Non-              Non-
         Personnel Personnel Personnel Personnel


1990      24845      10807         69.7      30.3
1991      27100      11770         69.7      30.3
1992      33889      12515         73.0      27.0
1993      39951      11911         77.0      23.0


Source: SEdC
                                          AN'NEX 1


Table 5: SOE Expenditure on Education (billion Yuan)



                                               Share (%)      ___

 Year         Central   Local     Total     Central Local


 1986         1.10       1.23     2.33       47.2      52.8
 1987         1.13       1.52     2.65       42.6      57.4
 1988         1.49       1.86     3.35       44.4      55.6
 1989         1.97       2.13     4.10       48.0      52.0
 1990         2.23       2.40     4.63       48.2      51.8
 1991         2.81       2.58     5.39       52.2      47.8
 1992         2.24       4.04     6.28       35.6      64.4
 1993         4.42       4.12     8.54       51.8      48.2


Source: MOF (Culture & Education Department)
(Continued)

                                         ANNEX 1

                                          Table 6


                                PrimaryEducationExpendituresby Province
                                                         0
                                             (MillionYuan)
  Province                  1989        1990        1991        1992       1993
Beijing                     196.63     242.69       262.34    306.7       409.91
Tianjin                    149.02      173.4        184.15    209.88      258.04
Hebei                      383.54      461.64       489.59    611.92      814.27
Shanxi                     226.54      264.27       284.41    363.78      481.87
Inner Mongolia             270.15      304.46       300.22    361.92      492.99
Liaoning                   472.52      539.49       587.07    689.66      857.18
Jilin                      293.98      321.65       355.46    423.31      521.43
Heilongian                 360.46      392.1        435.12    503.48      648.13
Shanghai                   285.48      313.35       349.00    433.21      564.83
Jiangsu                    556.52      652.36       686.74    848.83     1213.48
Zhejiang                   405.04      457.48       490.54    591.79      814.58
Anhui                      332.17      379.16       428.18    519.27      734.44
Fujian                     412.07      461.97       515.13    612.21      751.28
Jiangxi                    274.59      312.98       329.97    424.16      558.11
Shandong                   486.47      554.31       612.36    808.68     1084.23
Henan                      382.21      439.17       476.26    633.1       902.67
Hubei                      286.38      337.37       355.65    438.18      585.17
Hunan                      465.81      531.6        593.97    735.15      830.97
Guangdong                  764.31      873.56       974.64   1198.42     1782.1
Guangxi                    463.94      537.64       601.66    710.09      945.57
Hainan                     106.78      132.28       138.27    166.59      247.24
Sichuan                    743.29      853.9        885.88   1092.2      1381.77
z                          245.46      286.43       307.79    394.4       489.38
Yunnan                     476.95      544.36       604.25    729.33      953.27
Tibet                       56.66       65.77        60.7      70.05       82.24
Shannxi                    236.28      259.47       287.83    344.29      499.92
Gansu                      199.63      225.25       232.94    282.75      368.53
Qinghai                     65.99       68.03        71.42     87.3        99.98
Ningxia                      66.95      74.81        76.38     90.06      103.59
Xinjing                    273.74      319.74       349.22    419.25      491.09

a. Local Government Expenditure = Aggregate of (province + county + township).
(Continued)

                                        ANNEX 1

                                         Table 6


                                      PrinmaryEducation: Per-pupil
                                       Public Expenditure by
                                          Province (Yuan)
         Province             1990         1991       1992           1993
       Beijing               243.91       258.97     302.17      401.09
       Tianjin               212.81       212.89     241.24      290.91
       Hebei                  65.44       67.59       85.80      104.96
       Shanxi                 88.89       94.36      119.43      153.32
       Inner                 130.00      128.30      155.33      209.16
       Liaoning              132.46      149.95      177.43      232.55
       Jilin                 116.50      133.03      166.40      202.65
       Heilongjian            98.59      112.41      131.77      173.62
       Shanghai              281.54      310.22      375.40      479.08
       Jiangsu               106.56      115.48      144.19      203.98
       Zhejiang              122.85      135.28      163.79      226.78
       Anhui                  59.88       69.36       84.74      122.20
       Fujian                137.08      150.36      171.49      206.51
       Jiangxi                69.52       76.68      100.23      134.84
       Shandong               67.75       75.13       96.92      127.03
       Henan                  45.69       50.45       66.80        94.88
       Hubei                  54.15       57.79       70.79        93.79
       Hunan                  76.62       86.38      106.70      119.17
       Guangdong             116.91      125.13      153.05      214.17
       Guangxi                95.92      104.18      120.60      154.83
       Hainan                137.12      136.90      168.27      231.72
       Sichuan                92.15      100.50      122.28      150.80
       z                      65.85       70.95       90.40      111.27
       Yunnan                121.84      136.55      165.01      214.80
       Tibet                 432.70      361.31      391.34      428.33
       Shannxi                73.36       79.10       90.79      125.70
       Gansu                  93.23       95.82      115.03       147.47
       Qinghai               138.84      152.61      193.14      225.69
       Ningxia               112.16      115.55      139.41       164.95
       Xinjing               172.18      178.63      210.78       240.85

       a. Local Government Expenditure = Aggregate of (province + county
          + township).
(Continued)

                                        ANNEX 1

                                            Table 6



                                  Secondary Educaton Expenditures by
                                            (MillionYuw)a
       Province          1989        1990         1991        1992        1993
       Beijing          235.01     254.27        273.87     310.95       506.24
       Tianjin           141.25     162.49        173.27    195.28       253.43
       Hebei            454.86      520          549.08     655.97       905.85
       Shanxi           285.18     314.78        327.56     398.36       505.48
       Inner            235.18     253.96        249.25     294.79       434.22
       Liaoning         471.41     509.06        556.03      645.9       857.65
       Jilin            251.15     279.05         305.4      356.4       481.08
       Heilongjian      313.85     343.71        359.97     416.34       580.51
       Shanghai         285.06     301.61        327.85     407.35       600.37
       Jiangsu          510.98     594.16        629.49     769.95      1113.1
       Zhejiang         364.82     398.82        429.05     517.77        731.1
       Anhui            306.57     352.58        388.06     462.65       552.06
       Fujian           273.96     304.43        340.01     404.88       550.06
       Jiangxi          239.44     260.32        278.66     342.74       458.39
       Shandong         608.83     685.57        746.49     935.55      1332.59
       Henan            519.65     572.11        617.97     779.82      1066.45
       Hubei            377.13     425.32        442.99     531.27       706.44
       Hunan            441.75     478.4         576.99     696.08       777.36
       Guangdo           580.8     623.3         665.86     868.29      1510.1
       Guangxi          246.33     285.23         310.7     371.59       501.45
       Hainan             62.43      69.58         76.03      86.9       123.62
       Sichuan          607.29     718.89        739.25     878.21      1126.47
       Guizhou          124.72      146.2        154.48     189.27       246.24
       Yunnan           272.58     326.85        353.39     421.75       643.62
       Tibet              34.77     42.65          37.51     39.8          43.79
       Shannxi          281.46     299.25        325.14     379.63       487.26
       Gansu            163.45     179.88        188.72     230.47       281.68
       Qinghai            56.61      61.53         65.24     75.59         87.58
       Ningxia            58.9       62.79         63.26     77.19         85.89
       Xinjing          194.48     227.21        242.3      281.22        336.1

      a. Local Govermnent Expenditure = Aggregate of (province + county + township).
(Continued)

                                        ANNEX 1

                                         Table 6


                                       Secondary Education:Per-pupil
                                       Public Expenditure Province
                                                         by
                                                  (Yuan)a
       Province                 1990         1991         1992          1993
       Beijing                621.69        633.96       653.26         967.95
       Tianjin                470.99        472.13       516.61         651.49
       Hebei                  250.48        247.22       279.97         365.70
       Shanxi                 217.09        228.58       283.13         365.23
       Inner                  238.24        234.92       282.10         438.61
       Liaoning               281.25        294.66       340.31         448.80
       Jilin                  233.51        246.29       283.76         371.78
       Heilongjian            190.84        201.21       238.45         349.07
       Shanghai               619.32        634.14       718.43        1031.56
       Jiangsu                210.77        218.12       265.32         378.73
       Zhejiang               235.15        237.44       283.09         395.19
       Anhui                  161.66        173.78       205.08         238.99
       Fujian                 290.49        300.63       339.10         430.07
       Jiangxi                144.06        149.90       191.58         251.45
       Shandong               186.70        200.19       241.81         337.19
       Henan                  162.30        172.81       216.50         293.87
       Hubei                  201.10        201.63       239.20         310.66
       Hunan                  189.62        224.51       273.51         310.45
       Guangdo                266.37        279.54       358.21         544.97
       Guangxi                202.58        208.52       241.61         305.58
       Hainan                 276.11        307.81       329.17         451.17
       Sichuan                182.37        192.97       262.70         344.38
       Guizhou                151.50        154.17       191.38         245.01
       Yurman                 263.80        274.37       332.09         507.59
       Tibet                 2244.74       1705.00      1990.00         1903.91
       Shannxi                225.68        240.67       287.16         375.97
       Gansu                  186.60        196.58       253.82         323.03
       Qinghai                283.55        299.27       374.21         442.32
       Ningxia                221.09        217.39       259.03         321.69
       Xinjing                263.28        300.62       364.27         468.11

      a. Local Government Expenditure = Aggregate of (province + county +
         township).
(Continued)

                                          ANNEX 1

                                           Table 6


                                 Higher Education Expenditure by Province
                                          (Million Yuan)a
        Province      1989        1990         1991       1992        1993
       Beijing      480.35      559.46        615.59     686.68      867.85
      Tianjin        160.62      178.99       198.27     230.91      267.88
      Hebei          187.38     211.99        234.84     257.42      326.65
       Shanxi        132.77      147.16       163.57     177.47      232.07
      Inner           90.26       86.13        98.25     105.21      163.59
      Liaoning      329.97      398.46        453.03     529.56       645.6
      Jilin         233.01      248.53        273.98     314.06      382.29
      Heilongji       225.4     254.82        276.24     295.51      392.75
      Shanghai      398.81      439.59         502.8     558.94      699.21
      Jiangsu       377.45      415.94        466.97     523.04      688.92
      Zhejiang        181.5       194.6       219.34     249.65      312.21
      Anhui         144.91      171.73        147.76     188.51       237.2
      Fujian        129.29      142.73        154.77      180.5      219.64
      Jiangxi       102.34      115.89         130.2     146.25      164.25
      Shandong        361.8     380.73         423.8     430.48      584.83
      Henan         169.49      195.54         226.1     284.04      372.75
      Hubei         340.33      389.67         401.4     499.12      650.53
      Hunan         219.07      262.25        282.27     310.62      412.07
      Guangdo       273.31      291.86        325.13     404.94      510.08
      Guangxi         90.61     103.12        111.53     133.82      171.19
      Hainan          24.09       27.19        31.11      37.26       56.58
      Sichuan       365.57      411.73        452.18     515.55      604.56
      Guizhou         77.71       81.89        81.23      95.42      119.01
      Yunnan        130.43      119.54        133.75     1184.35     316.33
      Tibet           19.53       17.45        19.66      21.99       28.76
      Shannxi       260.96      251.82        354.43     394.44      471.44
      Gansu           88.61     103.12        112.55     124.87      161.78
      Qinghai        24.39        24.56        26.18      30.36       41.33
      Ningxia        21.55        26.39        24.01      30.22       41.94
      Xinjing        95.64      102.18        107.03      118.4      160.65


      a. Local Govermnent Expenditure = Aggregate of (province + county +
         township).
                                   ANNEX 1

                                     Table 6




                                  Higher Education: Per-pupil
                                 Public Expenditure by Province
                                          (Yuan)a
 Province             1990          1991          1992             1993
 Beijing             3996.14       4493.36       4703.29          5458.18
 Tianjin             3442.12       3965.40       4356.79          4464.67
 Hebei               2789.34       3173.51       2993.26          3110.95
 Shanxi              2830.00       3207.25       3113.51          3743.06
 Inner               2691.56       3169.35       3094.41          4421.35
 Liaoning            3239.51       3624.24       3981.65          4138.46
 Jilin               3404.52       3805.28       3975.44          4445.23
 Heilongjian         3185.25       3496.71       3476.59          4091.15
 Shanghai            3632.98       4297.44       4436.03          5337.48
 Jiangsu             2829.52       3220.48       3463.84          3785.27
 Zhejiang            3243.33       3655.67       3782.58          4219.05
 Anhui               2769.84       2383.23       2547.43          2928.40
 Fujian              2548.75       2866.11       3059.32          3431.88
 Jiangxi             2069.46       2325.00       2358.87          2313.38
 Shandong            3591.79       3924.07       3499.84          3847.57
 Henan               2444.25       2757.32       3087.39          3584.13
 Hubei               2997.46       3087.69       3565.14          4170.06
 Hunan               2980.11       3207.61       3376.30          3712.34
 Guangdong           3040.21       3496.02       3749.44          4359.66
 Guangxi             2713.68       3014.32       2909.13          3356.67
 Hainan              3398.75       3888.75       4657.50          5658.00
 Sichuan             2920.07       3206.95       3242.45          3396.40
 Guizhou             3032.96       3124.23       3534.07          4103.79
 Yunnan              2780.00       3110.47       3840.63          6326.60
 Tibet               8725.00       9830.00      10995.00          9586.67
 Shannxi             2623.13       3770.53       3829.51          4029.40
 Gansu               3124.85       3410.61       3468.61          4044.50
 Qinghai             4093.33       4363.33       5060.00          5904.29
 Ningxia             3298.75       3001.25       4317.14          4194.00
 Xinjing             3296.13       3344.69       3587.88          4119.23


a. Local Government Expenditure = Aggregate of (province + county +
   township).
                                                                ANNEX 2


Table 1: CHINA:TotalHealthExpenditure
(millionYuan)



                     CurrentPrices                      1987Prices                  PercentageShare       As a percentageof GNP
                                National                          National                                               National
Year        Public    Private    Total         Public   Private     Total           Public   Private   Public   Private   Total

1983       10686        7153     17839         8271       5536      13807           59.9     40.1       1.8       1.2     3.1
1984       12248        8591     20839          9970      6993      16963           58.8     41.2       1.8       1.2     3.0
1985       14309        8867     23176         12921      8007      20928           61.7     38.3       1.7       1.0     2.7
1986       18116        9612     27728         17138      9093      26231           65.3     34.7       1.9       1.0     2.9
1987       20844       12789     33633        20844      12789      33633           62.0      38.0      1.8       1.1     3.0
1988       26556       16716     43272        23711      14925      38636           61.4      38.6      1.9       1.2     3.1
1989       30886       20422     51308        25111      16603      41714           60.2      39.8      1.9       1.3     3.2
1990       35814       23798     59612        27132      18029      45161            60.1     39.9      2.0       1.3     3.4
1991       41047       30435     71482        29173     21631       50805            57.4     42.6      2.0       1.5     3.5
1992       48690       35850     84540        33077      24355      57432            57.6     42.4      2.0       1.5     3.5
1993       52076       37801     89877        36673      26620      63294            57.9     42.1      1.9       1.4     3.3


Average    28297       19276     47573        22184      14962      37146            60.2     39.8      1.9       1.3     3.1

  Note: Public Sector includes central & local governments, and SOE expenditures.
  Source:MOPH.
                                        ANNEX 2

                  Table 2: Sources of Financing Total Health Expenditure
                              (Current Prices, Million Yuan)


                          Domesticsources
                                             Rural
       Governmentbudget          Insurance  Collective Patient Foreign
Year    Central   LocaF        GEIS'   Labor Fund       fees    aid    Total


1980        152      3033        668     3725     2110      2944      NA   12632
1981        172      3244        786     3964     2030      3608      NA   13804
1982       214       3761        908     4677     2100      3648      NA   15308
1983       233       4258       1090     5105     1040      6113      NA   17839
1984       286       4898       1269     5795     1230      7361      NA   20839
1985       400       5686      1544      6678     1430      7437      NA   23176
1986       509       6631      1888      9088     1640      7972      NA   27728
1987       534       6551      2221     11538     1940     10819      NA   33633
1988       572       7143      2912     15929     2550     14166      NA   43272
1989       574       7872      3811     18629     2427     17995      NA   51308
1990       563       8286      4434     22531     2609     21189      NA   59612
1991       552       8907      5041     26547     3450     26985      NA   71482
1992       585      10748      5811     31546     4658     31192      NA   84540
1993       623      12403                                             NA


 NA: Not Applicable under the Budget Reporting System

 a. Local government = aggregate expenditure of province + county + township
 b. GEIS: Government Employees Insurance System.

 Source: MOPH.
                                       ANNEX 2

           Table 3: Uses of GovermnentHealth Expenditure (percentage)


                       Million Yuan                           Share
Year      Preventive        Curative     Total   Preventive    Curative   Total


1980         556            1785         2341     23.8          76.2      100
1981         580            1939         2519     23.0          77.0      100
1982         639            2285         2924     21.9          78.1      100
1983         705            2497         3202     22.0          78.0      100
1984         850            2771         3621     23.5          76.5      100
1985         960            3038         3998     24.0          76.0      100
1986        1114            3620         4734     23.5          76.5      100
1987        1142            3586         4728     24.2          75.8      100
1988        1314            3890         5204     25.2          74.8      100
1989        1473            4349         5822     25.3          74.7      100
1990        1620            4759         6379     25.4          74.6      100
1991        1760            5057         6817     25.8          74.2      100
1992        2346            5743         8089     29.0          71.0      100
1993        2857            6356         9213     31.0          69.0      100


Source: MOPH.
                                            ANNEX 2

                                 Table 4: Regession Results:
                           Dependent Variable: Infant Moitality Rate



Right hand                     OLS                 Fixed               Random
side variables                Estimates           Effects              Effects


log Income per capita           -0.688*            0.049               -0.509*
                                (9.828)            (1.28)               (-6.97)
log per capita Health            0.249            -0.114*              -0.531*
  Expenditure                    (4.44)            (-2.20)              (-8.16)
Safe water                      -0.234*            .052                 NC
                                (-4.178)           (1.3)
Sanitation                        -.154           -0.286*              -0.200
                                (-1.811)          (-3.404)             (-0.83)
DPT3                             0.134             -0.259*             -0.242*
                                 (1.13)           (-2.192)              (-3.15)
N                                 348               348                 348
Adj. R2                          0.55              .096                 0.20


*Indicates significant at 5% confidence level.
 Numbers in parentheses are the t-statistics.

Note: For details see Technical Annex C.
                                                         2
                                                    ANTNEX

                     Table 6: Total Local Government Health Expenditure (Nillion Yuan)a



        Province        1982   1983   1984   1985   1986   1987   1988   1989   1990   1991     1992   1993


Beijing                  106    115    139    196    250   277     275   275    243    266      352     366
Tianjin                   88    102    107    145    156    154    147    175   166    206       186    198
Hebei                    154    171    198   222     256   253     289   302    331    338      398    433
Shanxi                   110    132    157    170    186    173    171    198   236    259      290    316
Inner Mongolia           103    120   231     166    188   173    200    208    214    228      260    293
Liaoning                 214   234    254    334     375   359    373    423    472    460      480    614
Jilin                    123    143    156    192   239    219    236    260    261    271      317    341
Heilongjian              146    163    181   215    251    263    278    294    319    323      389    427
Shanghai                 123    140    158   214    260    261    309    353    350    371      453    531
Jiangsu                  187   233    256    276    346    343.   368    402    420    494      554    632
Zhejiang                 142    160    190   226    293    288    313    349    361    408      462    517
Anhui                    128    138    145    173    194   183     196   217    238    266      303    297
Fujian                   107    118    130    147    172   166     194   217    228    267      325    371
Jiangxi                  109    111    127    143    167   153     173   198    199    211      267    296
Shandong                263    298    320    373    417    438    507    579    620    648      784    864
Henan                    196   211    232    266    307    307    347    372    388    423      490    518
Hubei                    187   205    216    257    304    301    315    338    354    381      530    613
Hunan                    153    165    184   221    260    260    280    291    303    329      415    458
Guangdong               209    232    252    336    375    389    413    509    452    517      849    1234
Guangxi                  102    112   139     159    192   197    210    216    236    259      308    357
Hainan                     0     0      0      0      0      0     43     60     78       70    111     118
Sichuan                 249    278    313    363    451    419    455    499    568    590      678    767
Guizhou                  79      86    99     120    138   131     141    156   169     176     206    237
Yunnan                   117    144    173    181   218    227    248    274    310    341      391    506
Tibet                     35    42     50     68      57    64      67    67     74       77     81     109
Shannxi                  105    121    129    145    162    164    169    195   209    233      252     275
Gansu                     76     87    107    116    147    129    148    153    178    176      190    207
Qinghai                   34    47     58      55     61     57     62     62     64      65      89    109
Ningxia                   36     36    43     51      51    52      54     53    58       60     79      81
Xinjing                   82    114    128    155    158    151    162    177    198   202      261     318
        Total           3763   4258   4872   5685   6631   6i51   7143   7872   8297   8915    10750   12403




           a. Local Government Expenditure = Aggregateof (province + county + township)
           Source: MOPH.
                                         ANNEX 2
                    Table 7: Relative Personnel Ratio by Province (RPR)'

                                       Special     Assistzznt
         Province            Doctor     Nurse        Docior       Nurse    Midwife

 Guizhou                       0.57       0.58         1.1.6       0.65        0.77
 Guangxi                       0.67       0.99        0.74         0.75       0.62
 Anhui                         0.60       0.74        0.78         0.57        1.03
 Henan                         0.80       0.70        0.68         0.64       0.75
 Gansu                         1.02       0.84        0.85         0.85       0.43
 Sichuan                       1.23       0.79        0.77         0.59       0.90
 Jiangxi                       0.96       0.98        0.73         0.87       0.80
 Yunnan                        0.75      0.87         1.09         0.78       0.62
 Shaanxi                       1.38       1.00        1.00         0.92       0.71
 Hunan                         0.96      0.81         0.73         0.83       0.73
 Tibet                         1.08      0.82         1.49         0.66       0.23
 Ningxia                       1.12       1.26        1.41         1.37       0.62
 Qinghai                       1.48      1.64         1.25         1.09       0.95
 Shanxi                        1.65       1.07        1.29         1.30       1.12
 Inner Mongolia                1.45      1.06         1.91         1.34       1.08
 Hubei                         1.34      1.34         0.96         1.25       1.56
 Hebei                         0.95      0.66         0.96         0.81       0.60
 Jilin                         1.21      1.65         1.67         1.78       1.02
 Xinjiang                      0.72      1.85         1.93         1.74       1.03
 Fujian                        0.94      0.67         1.00         1.02       1.92
 Heilongjiang                  1.21      1.62         1.55         1.56       1.06
 Shandong                      0.72      0.93         0.68         0.89       1.20
 Jiangsu                       0.80      1.03         1.15         0.92       1.38
 Zhejiang                      0.83      0.77         1.16         0.87       1.76
 Liaoning                      1.22      2.00         1.46         2.16       1.00
 Guangdong                     0.89      0.73         0.97         1.22       1.02
 Tianjin                       2.71      2.69         1.09         2.31       1.11
 Beijing                       2.67      3.73         2.24         3.78       1.02
 Shanghai                      2.01      2.74         3.03         2.91       2.53

a. RPR= (Provincial Staff/Provincial Population)/(Nationa] Staff/National
          Population)
   RPR exceeding one indicates oversupply and less than one indicates undersupply
   of medical personnel.

   Source: Staff estimates from MOPH data.
                                      ANNEX 2

                Table 8: Per Capita GNP and Health Expenditres (US$)
                               In Selected Asian Countries, 1993




                                Per Capita                     Per Capita
                                  GNP                            Health
 Country                                                        Spending


Bangladesh                         220                            7.8
Bhutan                             180                            7.0
China                              470                           10.1
India                              310                           32.0
Indonesia                          670                           13.0
South Korea                       6790                          379.0
Laos                               250                            8.0
Malaysia                          2790                           70.0
Myanmar                            200                            7.5
Neptal                             175                            7.3
Pakistan                           420                           24.5
Philippines                        770                           15.4
Papua New Guinea                   950                           37.0
Sri Lank                           540                           31.0
Thailand                          1910                           74.0
Viet Nam                           185                            3.0

 Average


Source: World Bank Data File.
                                         ANNEX 3


Table 1: Public Expenditures on Social Security (Current Prices, Billion Yuan)



                                 1987     1988     1989     1990     1991        1992   1993


A. Social Insurance a/           15.9     19.1     21.9     25.7      30.1       36.5   46.2

      Retirement Pension         13.3     16.1     18.7     22.2     26.1        31.8   40.5
      Unemployment Benefit        2.6      3.0      3.2      3.5      4.0         4.7    5.7

B. Social Assistance             28.0     31.9     38.0     42.0      55.0       48.0   53.0

      Social Welfare b/           8.6      9.0     10.0     12.0     16.0        14.0   17.0
      Disaster Relief            10.4     11.9     14.0     14.0     22.0        16.0   16.0
        Central                   9.0     11.0     11.0     11.0     12.0        13.0   15.0
        Local                     1.4      0.9      3.0      2.0      6.0         3.0    2.0
      Special Assistance b/       9.0     11.0     14.0     16.0     17.0        18.0   20.0

C. Total (A +B)                  48.9     51.0     59.9     66.7      81.1       74.5   100.2


a! Excludes Medical Benefits. See Chapter II on Medical Benefits.
b/ Local level only

Sources: Statistical Year Books, MOL, MOCA and Staff estimates.
                                        ANNEX 3




Table 2: Public Expenditure on Social Security (1987 Prices, Billion Yuan)



                                 1987    1988     1989    1990      1991     1992   1993


A. Social Insurance a/           15.9    17.1    17.8     19.5      21.4     24.8   32.1

       Retirement Pension        13.3    14.4    15.2     16.8      18.6     21.6   28.1
       UnemploymentBenefit        2.6     2.7     2.6      2.7       2.8      3.2    4.0


B. Social Assistance             28.0    28.4    30.9     31.8      49.1     32.6   36.8
      Social Welfare b/           8.6     8.0     8.1      9.1      11.4      9.5   11.8
      Disaster Relief            10.4    10.6    11.4     10.6      15.6     10.9   11.1
        Central                   9.0     9.8     8.9      9.1      11.4      8.8    9.7
        Local                     1.4     0.8     2.4      1.5       4.3      2.0    1.4
      Special Assistance b/       9.0     9.8    11.4     12.1      12.1     12.2   13.9

C. Total (A+B)                  43.9     45.5    48.7    51.3       70.5     57.4   68.9



a/ Excludes Medical Benefits. See Chapter II on Medical Benefits.
b/ Local level only

Sources: Statistical Year Books, MOL, MOCA and Staff eslimates
                                                 Annex 3

                      Numberof Retirees(Year-End)and RetireeDependencyRatio*


                            1987       1988       1989        1990       1991    1992    1993
Total Retirees
(in 100 million)            19.68      21.20      22.01      23.01       24.33   25.98   27.80
DependencyRatio              6.7        6.4        6.2        6.1         6.0     5.7     5.4

                  staff ratio in all sectors and ownershipenterprises.
* Retiree-to-active
Source: StatisticalYear Bookof China 1994
                                          EmployeeRatio in the State Sector, 1993
                          Retiree-to-Active


              Regions            Retiree-to-Active           Regions              Retiree-to-Active
                                 Employee Ratio                                   Employee Ratio

      National                         5.3               Henna                           6.7
      Beijing                          4.5               Hubei                           6.1
      Tianjin                          4.2               Hunan                          4.8
      Hebei                            6.4               Guangdong                      5.9
      Shanxi                           6.5               Guangxi                        6.0
      Inner Mongolia                   7.4               Hainan                         4.3
      Liaoning                         4.4               Sichuan                        4.5
      Jilin                            6.5               Guizhou                        5.4

      Heilongjiang                     6.2               Yunnan                         4.9
      Shanghai                         2.8               Tibet                          6.0
      Jiangsu                          5.2               Shaanxi                        5.6
      Zhejiang                        5.3                Gansu                          6.5
      Anhui                           6.0                Qinghai                        4.6
     Fujian                            6.0               Ningxia                        6.0
     Jiangxi                          6.1                Xingjiang                      4.4
      Shangdong                       6.0            _                 _

   Source: StatisticalYear Bookof China 1994.


                                Balanceof the EnterprisePensionPooling Fund


                 -________
                    -________
          _______________                                 Y100illion
                                                             m
                        1987        1988          1989      1990           1991      1992       1993
 Contribution           74.26       132.23     172.15      202.73      294.93       376.69     349.00
 Expenditure            59.62       110.06     142.40      172.29      247.38       326.16     309.00

 Accumulative
 Balance                19.94       43.25      73.22       103.51      164.35       222.85     293.00

Source: (1) Statistical Year Book of China 1993; and (2) MOF.
                    CONTRIBUTIONS TO THE RURAL PENSION FUND

                            Y100 million                              As % of Total

               Village &       Individual        Total        Village &       Individual
                 TVEs           Farmers                        TVEs            Farmers

      1991        0.15            0.35           0.50            30               70
      1992        2.22            4.52           6.74            33               67

      1993        2.27            5.28           7.55            30               70

    Source: Ministry of Civil Affairs


                                      Urban Unemployment

                Unemployed             Urban Unemployment Rate            Redundant Labor
                 (1 million)                    (%)                       in SOEs (person)
    1987             2.77                        2.0
    1988             2.96                        2.0
    1989             3.78                        2.6
   1990              3.83                        2.5
    1991             3.52                        2.3                           29,257
    1992             3.64                        2.3                          103,556
   1993              4.20                        2.6

Source: Statistical Year Book of China 1994, Year Book of Labor Statistics of China 1992, 1993


                         Balance of Enterprise Unemployment Funds*

                                      Unemployment Funds (YIO million)
                            Receipts            Expenditure            Current year
                                             ___________                 Balance
             1987              5.07                0.86                    4.21
             1988              5.78                1.58                    4.20
             1989              6.52                1.71                    4.81
             1990              7.43                1.87                    5.56
             1991              8.37                2.50                    5.87
             1992              10.53               4.73                    5.80
             1993

           *The figures exclude contribution from COEs before 1991.

           Source: Financial Year Book of China 1993
                                 TECHNICALANNEX A

                                    A NOTE ON DATA


               Provincial Data In China, the governmentbudget, includingsocial sector,
containsfiscal transactionsby four levels of government:the central governmentand the
three levels of local govermnent:provincial, county and township. Province-leveldata on all
levels, such as school enrollment, income, governmentexpenditures,governmenteducational
and health expendituresare the aggregate of province + county + township. Whereas,
province-leveldata on all rates, such as cohort survival, transition, infant mortality, DPT3,
coverageof sanitationand water supply etc are the average of province + county +
township.Province-leveldata is the lowest possibledisaggregationavailablethus far on a
reasonably lengthytime-series (more than 10 years).


              Public Expenditures Unless otherwise noted, public expendituresin this report
include governmentbudgetaryexpenditures,excludingspendingby State Owned Enterprises
(SOE). They are treated separatelybecause comparisoncountries in the region have data on
governmentexpenditures,thus ensuring consistentcross-countrycomparisons. Historically,
the SOE expenditureson educationand health in China accounted,respectively, for 5% to a
maximumof 10%, and 20% to a maximumof 25% of total 'public" (Government+SOE)
expenditures.Thus, the gov'ernment still the dominantprovider of education and health
                                   is
services.

              Rich versus Poor Provinces It is worth pointing out here that the three largest
municipalities- Beijing, Tianjin and Shanghai- are very rich compared to all provinces in
terms of per capita income, and at the same time, they also enjoy provincialstatus.
                                         TECHNICALANNEX B

   CALCULATIONOF THE RATE OF RETURNTO INVESTMENTSIN EDUCATION


1.              Two main methods for estimating rate of return to investments in education were
considered, namely (a) Elaborate method and (b) earning function method. The following
sections illustrate these methods.

2.             (a) Elaborate Method:              Following Psacharopolous (1973, 1985), the basic
equation for this method is given by,

                      N
                   F, (YH - YA G(+r)
                                              c                                                ~~~~~~~~~~~~~~
                            = (Ys)t (1 +r)t
                          t-1


Where, Y = labor earnings, H = higher education, S = secondary education and r = rate of
return which is obtained by solving the above equation. The estimated rates of return for 1988
showed in the text were derived from an iterative computer program that begins with an arbitary
value of r and keeps modifying it by small increments until the left-hand side equalizes the right-
hand side. For private rate of return, the only cost of education under evaluation is the
opportunity cost of staying on in school beyond the age 18 instead of working in the labor
market. Conventionally, this opportunity cost is measured by the earnings of labor with
secondary school qualifications. For social rate of return, public unit costs of education are
included in the right-hand side of equation (1).

3.              In this respect, the earnings should be before-tax, whereas in the private rate of
return, earnings should be net of tax. But, as Psacharopolous (1985) points out, the post versus
pre-tax treatment of earnings does not make any significant difference in the calculation of the
rate of return. It is the added cost (unit costs) of schooling that chiefly explains the difference
between these two rates of return. As a result, a social rate of return remains lower than a
private rate of return.

4.             The earning function method: The earning function starts with a regression
equation of the following form:

                Iny        co+OS +A
                            a mnY aY
                                ay as                                                          (2)
                            1Y a
Where Y is income and S is the level of schooling. This semilog equation is an inflexible
functional form. The implication is one of increasing returns to education, unless second-order
regressors are included. The parameters of a semilog equation are:


                  InY= ac: 1 S +/ 2 S 2 +z
                         +
                aInY        daY
                 as       Y as
                       = fI+20   2S


                  ay =   l Y+2t32 S


If 3 is constant, as Y increases aYlaS also increases. This imposes an inflexibility on the
coeficients, which can be resolved by incorporating a higher-order terms in the regression
equation. (3)

In tune with equation (3) the Mincerian model specifies a relation of the form


                InY= o÷ +fl2EX + 3E
                      +/S                    2                                                (4)

where Y is income, S is schooling, E is experience and S x E is the schooling-experience
interaction variable. As noted before, this function as the disadvantage of being inflexible but,
if squared terms are included, the inflexibility is largely overcome.

The estimated regression coefficient ,Bis interpreted as the average private rate of return to one
extra year of schooling. Mincer also proved that

                     amnY
                       as
i.e., the rate of return is nothing but the relative change in earnings (alnY) as a result of a
change in schooling (aS). It is more appealing if one adds an educational level dimension to this
average rate of return. The first way is to add an e.s2 term in equation (4) where e is the
estimated coefficient on years of schooling squared. A differentiation with respect to S yields

                r = ,B +2eS                                                                    (5)


By substituting different values of S in the RHS of equation (5), a regression-derived rate of
return corresponding to different levels of schooling can be computed. Thus, for elementary
education (S =6), secondary education (S= 12) and higher education (S= 16) can be assigned.
Another way is to specify different levels in the earnings function by means of a series of
dummy variables (as reported in Table 1.10), viz. ELEM, SEC and HIGH having a value of 1
if the person belongs to the particular educational level and 0 else:


                InY= a +0 1 ELEM+0 2 SEC+0 3 HIGH+I¾LX+               2
                                                               5 Ex                        (6)

From this equation, the return to different levels of education are derived from the estimated
coefficients


                T (Elementaryvs.illiterate)
                                                SE

                T (Secondary vs.   elementary) =02   SE

                r(Higher vs. secondary)       = S3 -
                                                SH- sS

Where,        SE = elementary schooling
              Ss -secondary schooling
              SH = higher schooling
                                   TECHNICALANNEX C

                         A NOTE ON REGRESSIONANALYSIS


              The results from regression analyses with panel data are presented in Table 2.5
(Chapter 2) and Table 5 (Annex2). This sectionbriefly explains the methodologyand
equations.From a policy perspective, it is important to determinethe relative importance of
the pathwaysthrough which public policies and income influencehealth performance. The
basic time-series (1982-93),cross-sectional(29 provinces:aggregateof
province+county+townshipfor all levels such as income, expenditures,and average of
province+county+townshipfor all rates such as infant mortality, DPT3 etc) relationship
posited herein determinesthe net health outcomes.

             Equationsindicatingthe ultimate effectsof public policies and income on
health outcomes,are as follows:

(1)            lnGDPa=ao+a 2Zit+,xr=c oxnHEXP+U,+ea,

(2)                 I,LEB   1   =[o+0iZi,+3 21nGDP +E.=,) 03nHEXP+U +ei
                                                 5


              Where, GDP= real per capita provincialincome, Z= a vector of health
promotingservices, which includesthe coverageof safe water, sanitationand immunization
services by province. HEXP= provincialgovernmenthealth expenditureper capita,
whereas, r = recurrent expenditureper capita, and c = capitalexpenditureper capita.
IMR= infant mortalityrate by province, LEB= life expectancyat birth by province.

       First, the parameters al,cc2,c03,,1,02,03 are estimatedwith the simple OLS.
But, with the panel data (time-seriesand cross-section),the opportunityexists to remove the
possible bias introducedby unmeasurableor unobservable, time-invariantfactors, which
might be correlated with the includedright hand side variables. If the unobserved
endowmentsare time-invariant,for example, culture, religion, politicalenvironment,
agloclimateetc, then a fLxedeffects procedure is appropriate. The fixed-effectsmodel is
obtainedby transformingthe variablesinto deviationsfrom the individualmeans or by
introducingregional dummies. Since Ui terms in (1) and (2) are fixed, and they drop out of
the model. Consequently,unbiased and consistentestimatesare obtainedusing OLS. This
procedurethus remediesthe effects of all time-invariantomittedvariables '/.



     '/ For example, in equation (2), the transformationof the variables into deviationsfrom
the individualmeans is as follows:
         The unobserved regional effects may be random and causes biases in the estimated
 standard errors and subsequently nullyfy standard tests. The random effects procedure
 accounts for both time-invariant and time-varying error components. In the fixed effects
 model Uis are treated as parameters while in the random effects model the Ui s are treated as
 components of error term like eit. It is worth pointing out here that random effects
 procedure ignores any correlation between the errors and time-varying observed variables
 and can be estimated with generalized least squares procedure (GLS). To determine whether
 the fixed or random effects procedure is appropriate for the present data, Hausman-Wu
 specification test is used which shows fixed-effect model is better. But, impelled by
 intellectual curosity, we presented all three models to see how they differed in terms of
 magnitude and statistical significance.


        Description of Samnple The number of observations (N) in Table 2.5 constitutes 29
 provinces (cross-section) x 12 years (time-series 1982-93)= 348. In Table 5 (Annex 2)
 sample represents sub-national geographic entities in other countries. Details are presented
 below.

Malaysia:    14 states x 4 years = 56
Philippines: 12 regions x 9 years  108
Viet Namn: 7 regions x 5 years = 35




(IMRit-IMRi)= (GDPit-GDPi) + (HEXPit-HEXPi) + (Zit-Zi) + (eit-ei), since Uis are
fixed, they drop out of the model.

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:1
posted:8/30/2011
language:English
pages:115